monorail
/
Archipelago
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
Archipelago/worlds/yachtdice/YachtWeights.py

2357 lines
134 KiB
Python
Raw Normal View History

YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
yacht_weights = {
("Category Ones", 0, 0): {0: 100000},
("Category Ones", 0, 1): {0: 100000},
("Category Ones", 0, 2): {0: 100000},
("Category Ones", 0, 3): {0: 100000},
("Category Ones", 0, 4): {0: 100000},
("Category Ones", 0, 5): {0: 100000},
("Category Ones", 0, 6): {0: 100000},
("Category Ones", 0, 7): {0: 100000},
("Category Ones", 0, 8): {0: 100000},
("Category Ones", 1, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Ones", 1, 1): {0: 100000},
("Category Ones", 1, 2): {0: 100000},
("Category Ones", 1, 3): {0: 100000},
("Category Ones", 1, 4): {0: 100000},
("Category Ones", 1, 5): {0: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Ones", 1, 6): {0: 33491, 1: 66509},
("Category Ones", 1, 7): {0: 27838, 1: 72162},
("Category Ones", 1, 8): {0: 23094, 1: 76906},
("Category Ones", 2, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Ones", 2, 1): {0: 100000},
("Category Ones", 2, 2): {0: 100000},
("Category Ones", 2, 3): {0: 33544, 1: 66456},
("Category Ones", 2, 4): {0: 23342, 1: 76658},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Ones", 2, 5): {0: 16036, 1: 83964},
("Category Ones", 2, 6): {0: 11355, 1: 88645},
("Category Ones", 2, 7): {0: 7812, 1: 92188},
("Category Ones", 2, 8): {0: 5395, 1: 94605},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Ones", 3, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Ones", 3, 1): {0: 100000},
("Category Ones", 3, 2): {0: 33327, 1: 66673},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Ones", 3, 3): {0: 19432, 1: 80568},
("Category Ones", 3, 4): {0: 11191, 1: 88809},
("Category Ones", 3, 5): {0: 3963, 2: 64583, 1: 31454},
("Category Ones", 3, 6): {0: 3286, 2: 96714},
("Category Ones", 3, 7): {0: 57, 2: 99943},
("Category Ones", 3, 8): {2: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Ones", 4, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Ones", 4, 1): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Ones", 4, 2): {0: 23349, 1: 76651},
("Category Ones", 4, 3): {0: 11366, 1: 88634},
("Category Ones", 4, 4): {0: 3246, 2: 71438, 1: 25316},
("Category Ones", 4, 5): {0: 1466, 2: 98534},
("Category Ones", 4, 6): {0: 7, 2: 99993},
("Category Ones", 4, 7): {0: 2, 2: 31222, 3: 68776},
("Category Ones", 4, 8): {3: 99999, 2: 1},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Ones", 5, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Ones", 5, 1): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Ones", 5, 2): {0: 16212, 1: 83788},
("Category Ones", 5, 3): {0: 4879, 2: 69906, 1: 25215},
("Category Ones", 5, 4): {0: 1513, 2: 98487},
("Category Ones", 5, 5): {0: 484, 2: 31541, 3: 67975},
("Category Ones", 5, 6): {3: 99785, 2: 215},
("Category Ones", 5, 7): {3: 100000},
("Category Ones", 5, 8): {4: 66815, 3: 33185},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Ones", 6, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Ones", 6, 1): {0: 33501, 1: 66499},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Ones", 6, 2): {0: 11326, 1: 88674},
("Category Ones", 6, 3): {0: 2289, 2: 79783, 1: 17928},
("Category Ones", 6, 4): {0: 10, 3: 68933, 2: 30973, 1: 84},
("Category Ones", 6, 5): {0: 4, 3: 99996},
("Category Ones", 6, 6): {2: 1, 4: 67785, 3: 32214},
("Category Ones", 6, 7): {4: 100000},
("Category Ones", 6, 8): {4: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Ones", 7, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Ones", 7, 1): {0: 27838, 1: 72162},
("Category Ones", 7, 2): {0: 8807, 2: 68364, 1: 22829},
("Category Ones", 7, 3): {0: 75, 3: 62348, 2: 35246, 1: 2331},
("Category Ones", 7, 4): {0: 6, 3: 99994},
("Category Ones", 7, 5): {3: 29500, 4: 70500},
("Category Ones", 7, 6): {4: 100000},
("Category Ones", 7, 7): {4: 30322, 5: 69678},
("Category Ones", 7, 8): {5: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Ones", 8, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Ones", 8, 1): {0: 23156, 1: 76844},
("Category Ones", 8, 2): {0: 5678, 2: 75480, 1: 18842},
("Category Ones", 8, 3): {0: 28, 3: 99972},
("Category Ones", 8, 4): {3: 32486, 4: 67514},
("Category Ones", 8, 5): {4: 100000},
("Category Ones", 8, 6): {5: 74125, 4: 25875},
("Category Ones", 8, 7): {6: 60476, 5: 29297, 4: 10227},
("Category Ones", 8, 8): {6: 99999, 5: 1},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Twos", 0, 0): {0: 100000},
("Category Twos", 0, 1): {0: 100000},
("Category Twos", 0, 2): {0: 100000},
("Category Twos", 0, 3): {0: 100000},
("Category Twos", 0, 4): {0: 100000},
("Category Twos", 0, 5): {0: 100000},
("Category Twos", 0, 6): {0: 100000},
("Category Twos", 0, 7): {0: 100000},
("Category Twos", 0, 8): {0: 100000},
("Category Twos", 1, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Twos", 1, 1): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Twos", 1, 2): {0: 69690, 2: 30310},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Twos", 1, 3): {0: 57818, 2: 42182},
("Category Twos", 1, 4): {0: 48418, 2: 51582},
("Category Twos", 1, 5): {0: 40301, 2: 59699},
("Category Twos", 1, 6): {0: 33558, 2: 66442},
("Category Twos", 1, 7): {0: 28182, 2: 71818},
("Category Twos", 1, 8): {0: 23406, 2: 76594},
("Category Twos", 2, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Twos", 2, 1): {0: 69724, 2: 30276},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Twos", 2, 2): {0: 48238, 2: 51762},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Twos", 2, 3): {0: 33290, 2: 66710},
("Category Twos", 2, 4): {0: 23136, 2: 76864},
("Category Twos", 2, 5): {0: 16146, 2: 48200, 4: 35654},
("Category Twos", 2, 6): {0: 11083, 2: 44497, 4: 44420},
("Category Twos", 2, 7): {0: 7662, 2: 40343, 4: 51995},
("Category Twos", 2, 8): {0: 5354, 2: 35526, 4: 59120},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Twos", 3, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Twos", 3, 1): {0: 58021, 2: 41979},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Twos", 3, 2): {0: 33548, 2: 66452},
("Category Twos", 3, 3): {0: 19375, 2: 42372, 4: 38253},
("Category Twos", 3, 4): {0: 10998, 2: 36435, 4: 52567},
("Category Twos", 3, 5): {0: 7954, 4: 92046},
("Category Twos", 3, 6): {0: 347, 4: 99653},
("Category Twos", 3, 7): {0: 2, 4: 62851, 6: 37147},
("Category Twos", 3, 8): {6: 99476, 4: 524},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Twos", 4, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Twos", 4, 1): {0: 48235, 2: 51765},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Twos", 4, 2): {0: 23289, 2: 40678, 4: 36033},
("Category Twos", 4, 3): {0: 11177, 2: 32677, 4: 56146},
("Category Twos", 4, 4): {0: 5522, 4: 60436, 6: 34042},
("Category Twos", 4, 5): {0: 4358, 6: 95642},
("Category Twos", 4, 6): {0: 20, 6: 99980},
("Category Twos", 4, 7): {6: 100000},
("Category Twos", 4, 8): {6: 65250, 8: 34750},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Twos", 5, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Twos", 5, 1): {0: 40028, 2: 59972},
("Category Twos", 5, 2): {0: 16009, 2: 35901, 4: 48090},
("Category Twos", 5, 3): {0: 6820, 4: 57489, 6: 35691},
("Category Twos", 5, 4): {0: 5285, 6: 94715},
("Category Twos", 5, 5): {0: 18, 6: 66613, 8: 33369},
("Category Twos", 5, 6): {8: 99073, 6: 927},
("Category Twos", 5, 7): {8: 100000},
("Category Twos", 5, 8): {8: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Twos", 6, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Twos", 6, 1): {0: 33502, 2: 66498},
("Category Twos", 6, 2): {0: 13681, 4: 59162, 2: 27157},
("Category Twos", 6, 3): {0: 5486, 6: 94514},
("Category Twos", 6, 4): {0: 190, 6: 62108, 8: 37702},
("Category Twos", 6, 5): {8: 99882, 6: 118},
("Category Twos", 6, 6): {8: 65144, 10: 34856},
("Category Twos", 6, 7): {10: 99524, 8: 476},
("Category Twos", 6, 8): {10: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Twos", 7, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Twos", 7, 1): {0: 27683, 2: 39060, 4: 33257},
("Category Twos", 7, 2): {0: 8683, 4: 54932, 6: 36385},
("Category Twos", 7, 3): {0: 373, 6: 66572, 8: 33055},
("Category Twos", 7, 4): {8: 99816, 6: 184},
("Category Twos", 7, 5): {8: 58124, 10: 41876},
("Category Twos", 7, 6): {10: 99948, 8: 52},
("Category Twos", 7, 7): {10: 62549, 12: 37451},
("Category Twos", 7, 8): {12: 99818, 10: 182},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Twos", 8, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Twos", 8, 1): {0: 23378, 2: 37157, 4: 39465},
("Category Twos", 8, 2): {0: 5602, 6: 94398},
("Category Twos", 8, 3): {0: 8, 6: 10911, 8: 89081},
("Category Twos", 8, 4): {8: 59809, 10: 40191},
("Category Twos", 8, 5): {10: 68808, 12: 31114, 8: 78},
("Category Twos", 8, 6): {12: 98712, 10: 1287, 8: 1},
("Category Twos", 8, 7): {12: 100000},
("Category Twos", 8, 8): {12: 59018, 14: 40982},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Threes", 0, 0): {0: 100000},
("Category Threes", 0, 1): {0: 100000},
("Category Threes", 0, 2): {0: 100000},
("Category Threes", 0, 3): {0: 100000},
("Category Threes", 0, 4): {0: 100000},
("Category Threes", 0, 5): {0: 100000},
("Category Threes", 0, 6): {0: 100000},
("Category Threes", 0, 7): {0: 100000},
("Category Threes", 0, 8): {0: 100000},
("Category Threes", 1, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Threes", 1, 1): {0: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Threes", 1, 2): {0: 69569, 3: 30431},
("Category Threes", 1, 3): {0: 57872, 3: 42128},
("Category Threes", 1, 4): {0: 48081, 3: 51919},
("Category Threes", 1, 5): {0: 40271, 3: 59729},
("Category Threes", 1, 6): {0: 33201, 3: 66799},
("Category Threes", 1, 7): {0: 27903, 3: 72097},
("Category Threes", 1, 8): {0: 23240, 3: 76760},
("Category Threes", 2, 0): {0: 100000},
("Category Threes", 2, 1): {0: 69419, 3: 30581},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Threes", 2, 2): {0: 48202, 3: 51798},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Threes", 2, 3): {0: 33376, 3: 66624},
("Category Threes", 2, 4): {0: 23276, 3: 49810, 6: 26914},
("Category Threes", 2, 5): {0: 16092, 3: 47718, 6: 36190},
("Category Threes", 2, 6): {0: 11232, 3: 44515, 6: 44253},
("Category Threes", 2, 7): {0: 7589, 3: 40459, 6: 51952},
("Category Threes", 2, 8): {0: 5447, 3: 35804, 6: 58749},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Threes", 3, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Threes", 3, 1): {0: 57964, 3: 42036},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Threes", 3, 2): {0: 33637, 3: 44263, 6: 22100},
("Category Threes", 3, 3): {0: 19520, 3: 42382, 6: 38098},
("Category Threes", 3, 4): {0: 11265, 3: 35772, 6: 52963},
("Category Threes", 3, 5): {0: 6419, 3: 28916, 6: 43261, 9: 21404},
("Category Threes", 3, 6): {0: 3810, 3: 22496, 6: 44388, 9: 29306},
("Category Threes", 3, 7): {0: 1317, 6: 30047, 9: 68636},
("Category Threes", 3, 8): {0: 750, 9: 99250},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Threes", 4, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Threes", 4, 1): {0: 48121, 3: 51879},
("Category Threes", 4, 2): {0: 23296, 3: 40989, 6: 35715},
("Category Threes", 4, 3): {0: 11233, 3: 32653, 6: 35710, 9: 20404},
("Category Threes", 4, 4): {0: 5463, 3: 23270, 6: 37468, 9: 33799},
("Category Threes", 4, 5): {0: 5225, 6: 29678, 9: 65097},
("Category Threes", 4, 6): {0: 3535, 9: 96465},
("Category Threes", 4, 7): {0: 6, 9: 72939, 12: 27055},
("Category Threes", 4, 8): {9: 25326, 12: 74674},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Threes", 5, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Threes", 5, 1): {0: 40183, 3: 59817},
("Category Threes", 5, 2): {0: 16197, 3: 35494, 6: 48309},
("Category Threes", 5, 3): {0: 6583, 3: 23394, 6: 34432, 9: 35591},
("Category Threes", 5, 4): {0: 5007, 6: 25159, 9: 49038, 12: 20796},
("Category Threes", 5, 5): {0: 2900, 9: 38935, 12: 58165},
("Category Threes", 5, 6): {0: 2090, 12: 97910},
("Category Threes", 5, 7): {12: 99994, 9: 6},
("Category Threes", 5, 8): {12: 73524, 15: 26476},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Threes", 6, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Threes", 6, 1): {0: 33473, 3: 40175, 6: 26352},
("Category Threes", 6, 2): {0: 11147, 3: 29592, 6: 32630, 9: 26631},
("Category Threes", 6, 3): {0: 2460, 6: 21148, 9: 55356, 12: 21036},
("Category Threes", 6, 4): {0: 997, 9: 29741, 12: 69262},
("Category Threes", 6, 5): {0: 831, 12: 76328, 15: 22841},
("Category Threes", 6, 6): {12: 29960, 15: 70040},
("Category Threes", 6, 7): {15: 100000},
("Category Threes", 6, 8): {15: 79456, 18: 20544},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Threes", 7, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Threes", 7, 1): {0: 27933, 3: 39105, 6: 32962},
("Category Threes", 7, 2): {0: 7794, 3: 23896, 6: 31832, 9: 36478},
("Category Threes", 7, 3): {0: 1321, 9: 40251, 12: 58428},
("Category Threes", 7, 4): {0: 370, 12: 74039, 15: 25591},
("Category Threes", 7, 5): {0: 6, 15: 98660, 12: 1334},
("Category Threes", 7, 6): {15: 73973, 18: 26027},
("Category Threes", 7, 7): {18: 100000},
("Category Threes", 7, 8): {18: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Threes", 8, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Threes", 8, 1): {0: 23337, 3: 37232, 6: 39431},
("Category Threes", 8, 2): {0: 4652, 6: 29310, 9: 45517, 12: 20521},
("Category Threes", 8, 3): {0: 1300, 12: 77919, 15: 20781},
("Category Threes", 8, 4): {0: 21, 15: 98678, 12: 1301},
("Category Threes", 8, 5): {15: 68893, 18: 31107},
("Category Threes", 8, 6): {18: 100000},
("Category Threes", 8, 7): {18: 69986, 21: 30014},
("Category Threes", 8, 8): {21: 98839, 18: 1161},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Fours", 0, 0): {0: 100000},
("Category Fours", 0, 1): {0: 100000},
("Category Fours", 0, 2): {0: 100000},
("Category Fours", 0, 3): {0: 100000},
("Category Fours", 0, 4): {0: 100000},
("Category Fours", 0, 5): {0: 100000},
("Category Fours", 0, 6): {0: 100000},
("Category Fours", 0, 7): {0: 100000},
("Category Fours", 0, 8): {0: 100000},
("Category Fours", 1, 0): {0: 100000},
("Category Fours", 1, 1): {0: 83260, 4: 16740},
("Category Fours", 1, 2): {0: 69514, 4: 30486},
("Category Fours", 1, 3): {0: 58017, 4: 41983},
("Category Fours", 1, 4): {0: 48389, 4: 51611},
("Category Fours", 1, 5): {0: 40201, 4: 59799},
("Category Fours", 1, 6): {0: 33496, 4: 66504},
("Category Fours", 1, 7): {0: 28052, 4: 71948},
("Category Fours", 1, 8): {0: 23431, 4: 76569},
("Category Fours", 2, 0): {0: 100000},
("Category Fours", 2, 1): {0: 69379, 4: 30621},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Fours", 2, 2): {0: 48538, 4: 51462},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Fours", 2, 3): {0: 33756, 4: 48555, 8: 17689},
("Category Fours", 2, 4): {0: 23070, 4: 49916, 8: 27014},
("Category Fours", 2, 5): {0: 16222, 4: 48009, 8: 35769},
("Category Fours", 2, 6): {0: 11125, 4: 44400, 8: 44475},
("Category Fours", 2, 7): {0: 7919, 4: 40216, 8: 51865},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Fours", 2, 8): {0: 5348, 4: 35757, 8: 58895},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Fours", 3, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Fours", 3, 1): {0: 57914, 4: 42086},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Fours", 3, 2): {0: 33621, 4: 44110, 8: 22269},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Fours", 3, 3): {0: 19153, 4: 42425, 8: 38422},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Fours", 3, 4): {0: 11125, 4: 36011, 8: 52864},
("Category Fours", 3, 5): {0: 6367, 4: 29116, 8: 43192, 12: 21325},
("Category Fours", 3, 6): {0: 3643, 4: 22457, 8: 44477, 12: 29423},
("Category Fours", 3, 7): {0: 2178, 4: 16802, 8: 43275, 12: 37745},
("Category Fours", 3, 8): {0: 488, 8: 20703, 12: 78809},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Fours", 4, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Fours", 4, 1): {0: 48465, 4: 51535},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Fours", 4, 2): {0: 23296, 4: 40911, 8: 35793},
("Category Fours", 4, 3): {0: 11200, 4: 33191, 8: 35337, 12: 20272},
("Category Fours", 4, 4): {0: 5447, 4: 23066, 8: 37441, 12: 34046},
("Category Fours", 4, 5): {0: 2533, 4: 15668, 8: 34781, 12: 47018},
("Category Fours", 4, 6): {0: 2058, 8: 19749, 12: 58777, 16: 19416},
("Category Fours", 4, 7): {0: 1476, 12: 45913, 16: 52611},
("Category Fours", 4, 8): {0: 23, 12: 18149, 16: 81828},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Fours", 5, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Fours", 5, 1): {0: 40215, 4: 40127, 8: 19658},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Fours", 5, 2): {0: 15946, 4: 35579, 8: 31158, 12: 17317},
("Category Fours", 5, 3): {0: 6479, 4: 23705, 8: 34575, 12: 35241},
("Category Fours", 5, 4): {0: 4987, 8: 25190, 12: 48849, 16: 20974},
("Category Fours", 5, 5): {0: 1553, 12: 39966, 16: 58481},
("Category Fours", 5, 6): {0: 843, 16: 99157},
("Category Fours", 5, 7): {16: 80514, 20: 19486},
("Category Fours", 5, 8): {16: 38393, 20: 61607},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Fours", 6, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Fours", 6, 1): {0: 33632, 4: 39856, 8: 26512},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Fours", 6, 2): {0: 11175, 4: 29824, 8: 32381, 12: 26620},
("Category Fours", 6, 3): {0: 3698, 4: 16329, 8: 29939, 12: 29071, 16: 20963},
("Category Fours", 6, 4): {0: 2326, 12: 28286, 16: 69388},
("Category Fours", 6, 5): {0: 1030, 16: 76056, 20: 22914},
("Category Fours", 6, 6): {0: 7, 16: 29753, 20: 70240},
("Category Fours", 6, 7): {20: 99999, 16: 1},
("Category Fours", 6, 8): {20: 79470, 24: 20530},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Fours", 7, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Fours", 7, 1): {0: 27821, 4: 39289, 8: 32890},
("Category Fours", 7, 2): {0: 7950, 4: 24026, 8: 31633, 12: 36391},
("Category Fours", 7, 3): {0: 1887, 12: 31108, 16: 67005},
("Category Fours", 7, 4): {0: 423, 16: 73837, 20: 25740},
("Category Fours", 7, 5): {0: 57, 16: 10063, 20: 74092, 24: 15788},
("Category Fours", 7, 6): {0: 6, 20: 31342, 24: 68652},
("Category Fours", 7, 7): {24: 99995, 20: 5},
("Category Fours", 7, 8): {24: 84330, 28: 15670},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Fours", 8, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Fours", 8, 1): {0: 23275, 4: 37161, 8: 39564},
("Category Fours", 8, 2): {0: 5421, 4: 19014, 8: 29259, 12: 25812, 16: 20494},
("Category Fours", 8, 3): {0: 649, 16: 78572, 20: 20779},
("Category Fours", 8, 4): {0: 15, 20: 80772, 24: 17355, 16: 1858},
("Category Fours", 8, 5): {20: 15615, 24: 84385},
("Category Fours", 8, 6): {24: 80655, 28: 19345},
("Category Fours", 8, 7): {24: 23969, 28: 76031},
("Category Fours", 8, 8): {28: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Fives", 0, 0): {0: 100000},
("Category Fives", 0, 1): {0: 100000},
("Category Fives", 0, 2): {0: 100000},
("Category Fives", 0, 3): {0: 100000},
("Category Fives", 0, 4): {0: 100000},
("Category Fives", 0, 5): {0: 100000},
("Category Fives", 0, 6): {0: 100000},
("Category Fives", 0, 7): {0: 100000},
("Category Fives", 0, 8): {0: 100000},
("Category Fives", 1, 0): {0: 100000},
("Category Fives", 1, 1): {0: 83338, 5: 16662},
("Category Fives", 1, 2): {0: 69499, 5: 30501},
("Category Fives", 1, 3): {0: 57799, 5: 42201},
("Category Fives", 1, 4): {0: 48311, 5: 51689},
("Category Fives", 1, 5): {0: 40084, 5: 59916},
("Category Fives", 1, 6): {0: 33440, 5: 66560},
("Category Fives", 1, 7): {0: 27730, 5: 72270},
("Category Fives", 1, 8): {0: 23210, 5: 76790},
("Category Fives", 2, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Fives", 2, 1): {0: 69299, 5: 30701},
("Category Fives", 2, 2): {0: 48156, 5: 51844},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Fives", 2, 3): {0: 33225, 5: 49153, 10: 17622},
("Category Fives", 2, 4): {0: 23218, 5: 50075, 10: 26707},
("Category Fives", 2, 5): {0: 15939, 5: 48313, 10: 35748},
("Category Fives", 2, 6): {0: 11340, 5: 44324, 10: 44336},
("Category Fives", 2, 7): {0: 7822, 5: 40388, 10: 51790},
("Category Fives", 2, 8): {0: 5386, 5: 35636, 10: 58978},
("Category Fives", 3, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Fives", 3, 1): {0: 58034, 5: 41966},
("Category Fives", 3, 2): {0: 33466, 5: 44227, 10: 22307},
("Category Fives", 3, 3): {0: 19231, 5: 42483, 10: 38286},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Fives", 3, 4): {0: 11196, 5: 36192, 10: 38673, 15: 13939},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Fives", 3, 5): {0: 6561, 5: 29163, 10: 43014, 15: 21262},
("Category Fives", 3, 6): {0: 3719, 5: 22181, 10: 44611, 15: 29489},
("Category Fives", 3, 7): {0: 2099, 5: 16817, 10: 43466, 15: 37618},
("Category Fives", 3, 8): {0: 1281, 5: 12473, 10: 40936, 15: 45310},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Fives", 4, 0): {0: 100000},
("Category Fives", 4, 1): {0: 48377, 5: 38345, 10: 13278},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Fives", 4, 2): {0: 23126, 5: 40940, 10: 35934},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Fives", 4, 3): {0: 11192, 5: 32597, 10: 35753, 15: 20458},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Fives", 4, 4): {0: 5362, 5: 23073, 10: 37379, 15: 34186},
("Category Fives", 4, 5): {0: 2655, 5: 15662, 10: 34602, 15: 34186, 20: 12895},
("Category Fives", 4, 6): {0: 2059, 10: 19678, 15: 48376, 20: 29887},
("Category Fives", 4, 7): {0: 1473, 15: 34402, 20: 64125},
("Category Fives", 4, 8): {0: 551, 20: 99449},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Fives", 5, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Fives", 5, 1): {0: 39911, 5: 40561, 10: 19528},
("Category Fives", 5, 2): {0: 16178, 5: 35517, 10: 31246, 15: 17059},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Fives", 5, 3): {0: 6526, 5: 23716, 10: 34430, 15: 35328},
("Category Fives", 5, 4): {0: 2615, 5: 13975, 10: 30133, 15: 32247, 20: 21030},
("Category Fives", 5, 5): {0: 1482, 10: 13532, 15: 37597, 20: 47389},
("Category Fives", 5, 6): {0: 477, 15: 14484, 20: 71985, 25: 13054},
("Category Fives", 5, 7): {0: 273, 20: 52865, 25: 46862},
("Category Fives", 5, 8): {20: 16822, 25: 83178},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Fives", 6, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Fives", 6, 1): {0: 33476, 5: 40167, 10: 26357},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Fives", 6, 2): {0: 11322, 5: 29613, 10: 32664, 15: 26401},
("Category Fives", 6, 3): {0: 3765, 5: 16288, 10: 29770, 15: 29233, 20: 20944},
("Category Fives", 6, 4): {0: 1889, 10: 13525, 15: 33731, 20: 38179, 25: 12676},
("Category Fives", 6, 5): {0: 53, 10: 11118, 20: 47588, 25: 41241},
("Category Fives", 6, 6): {0: 10, 20: 8876, 25: 91114},
("Category Fives", 6, 7): {0: 7, 25: 85815, 30: 14178},
("Category Fives", 6, 8): {25: 43072, 30: 56928},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Fives", 7, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Fives", 7, 1): {0: 27826, 5: 39154, 10: 33020},
("Category Fives", 7, 2): {0: 7609, 5: 24193, 10: 31722, 15: 23214, 20: 13262},
("Category Fives", 7, 3): {0: 1879, 15: 23021, 20: 75100},
("Category Fives", 7, 4): {0: 345, 20: 64636, 25: 35019},
("Category Fives", 7, 5): {0: 40, 20: 7522, 25: 76792, 30: 15646},
("Category Fives", 7, 6): {0: 8, 25: 26517, 30: 73475},
("Category Fives", 7, 7): {0: 2, 30: 99998},
("Category Fives", 7, 8): {30: 84211, 35: 15789},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Fives", 8, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Fives", 8, 1): {0: 23333, 5: 37259, 10: 25947, 15: 13461},
("Category Fives", 8, 2): {0: 5425, 5: 18915, 10: 29380, 15: 25994, 20: 20286},
("Category Fives", 8, 3): {0: 495, 20: 78726, 25: 20779},
("Category Fives", 8, 4): {20: 12998, 25: 70085, 30: 16917},
("Category Fives", 8, 5): {25: 15859, 30: 84141},
("Category Fives", 8, 6): {30: 80722, 35: 19278},
("Category Fives", 8, 7): {30: 23955, 35: 76045},
("Category Fives", 8, 8): {35: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Sixes", 0, 0): {0: 100000},
("Category Sixes", 0, 1): {0: 100000},
("Category Sixes", 0, 2): {0: 100000},
("Category Sixes", 0, 3): {0: 100000},
("Category Sixes", 0, 4): {0: 100000},
("Category Sixes", 0, 5): {0: 100000},
("Category Sixes", 0, 6): {0: 100000},
("Category Sixes", 0, 7): {0: 100000},
("Category Sixes", 0, 8): {0: 100000},
("Category Sixes", 1, 0): {0: 100000},
("Category Sixes", 1, 1): {0: 83168, 6: 16832},
("Category Sixes", 1, 2): {0: 69548, 6: 30452},
("Category Sixes", 1, 3): {0: 57697, 6: 42303},
("Category Sixes", 1, 4): {0: 48043, 6: 51957},
("Category Sixes", 1, 5): {0: 39912, 6: 60088},
("Category Sixes", 1, 6): {0: 33499, 6: 66501},
("Category Sixes", 1, 7): {0: 28251, 6: 71749},
("Category Sixes", 1, 8): {0: 23206, 6: 76794},
("Category Sixes", 2, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Sixes", 2, 1): {0: 69463, 6: 30537},
("Category Sixes", 2, 2): {0: 47896, 6: 52104},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Sixes", 2, 3): {0: 33394, 6: 48757, 12: 17849},
("Category Sixes", 2, 4): {0: 23552, 6: 49554, 12: 26894},
("Category Sixes", 2, 5): {0: 16090, 6: 48098, 12: 35812},
("Category Sixes", 2, 6): {0: 11073, 6: 44833, 12: 44094},
("Category Sixes", 2, 7): {0: 7737, 6: 40480, 12: 51783},
("Category Sixes", 2, 8): {0: 5379, 6: 35672, 12: 58949},
("Category Sixes", 3, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Sixes", 3, 1): {0: 57718, 6: 42282},
("Category Sixes", 3, 2): {0: 33610, 6: 44328, 12: 22062},
("Category Sixes", 3, 3): {0: 19366, 6: 42246, 12: 38388},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Sixes", 3, 4): {0: 11144, 6: 36281, 12: 38817, 18: 13758},
("Category Sixes", 3, 5): {0: 6414, 6: 28891, 12: 43114, 18: 21581},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Sixes", 3, 6): {0: 3870, 6: 22394, 12: 44318, 18: 29418},
("Category Sixes", 3, 7): {0: 2188, 6: 16803, 12: 43487, 18: 37522},
("Category Sixes", 3, 8): {0: 1289, 6: 12421, 12: 41082, 18: 45208},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Sixes", 4, 0): {0: 100000},
("Category Sixes", 4, 1): {0: 48197, 6: 38521, 12: 13282},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Sixes", 4, 2): {0: 23155, 6: 41179, 12: 35666},
("Category Sixes", 4, 3): {0: 11256, 6: 32609, 12: 35588, 18: 20547},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Sixes", 4, 4): {0: 5324, 6: 23265, 12: 37209, 18: 34202},
("Category Sixes", 4, 5): {0: 2658, 6: 15488, 12: 34685, 18: 34476, 24: 12693},
("Category Sixes", 4, 6): {0: 2045, 12: 19683, 18: 48559, 24: 29713},
("Category Sixes", 4, 7): {0: 1470, 18: 34646, 24: 63884},
("Category Sixes", 4, 8): {0: 22, 18: 12111, 24: 87867},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Sixes", 5, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Sixes", 5, 1): {0: 40393, 6: 39904, 12: 19703},
("Category Sixes", 5, 2): {0: 16202, 6: 35664, 12: 31241, 18: 16893},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Sixes", 5, 3): {0: 6456, 6: 23539, 12: 34585, 18: 25020, 24: 10400},
("Category Sixes", 5, 4): {0: 2581, 6: 13980, 12: 30355, 18: 32198, 24: 20886},
("Category Sixes", 5, 5): {0: 1472, 12: 13518, 18: 37752, 24: 47258},
("Category Sixes", 5, 6): {0: 476, 18: 14559, 24: 71856, 30: 13109},
("Category Sixes", 5, 7): {0: 275, 24: 52573, 30: 47152},
("Category Sixes", 5, 8): {24: 16500, 30: 83500},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Sixes", 6, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Sixes", 6, 1): {0: 33316, 6: 40218, 12: 26466},
("Category Sixes", 6, 2): {0: 11256, 6: 29444, 12: 32590, 18: 26710},
("Category Sixes", 6, 3): {0: 3787, 6: 16266, 12: 29873, 18: 29107, 24: 20967},
("Category Sixes", 6, 4): {0: 1875, 12: 13602, 18: 33731, 24: 38090, 30: 12702},
("Category Sixes", 6, 5): {0: 433, 18: 10665, 24: 47398, 30: 41504},
("Category Sixes", 6, 6): {0: 89, 24: 14905, 30: 85006},
("Category Sixes", 6, 7): {0: 19, 30: 85816, 36: 14165},
("Category Sixes", 6, 8): {30: 43219, 36: 56781},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Sixes", 7, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Sixes", 7, 1): {0: 27852, 6: 38984, 12: 33164},
("Category Sixes", 7, 2): {0: 7883, 6: 23846, 12: 31558, 18: 23295, 24: 13418},
("Category Sixes", 7, 3): {0: 2186, 6: 10928, 12: 24321, 18: 29650, 24: 21177, 30: 11738},
("Category Sixes", 7, 4): {0: 1034, 18: 12857, 24: 37227, 30: 48882},
("Category Sixes", 7, 5): {0: 300, 30: 83887, 36: 15813},
("Category Sixes", 7, 6): {30: 31359, 36: 68641},
("Category Sixes", 7, 7): {36: 89879, 42: 10121},
("Category Sixes", 7, 8): {36: 49549, 42: 50451},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Sixes", 8, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Sixes", 8, 1): {0: 23220, 6: 37213, 12: 25961, 18: 13606},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Sixes", 8, 2): {0: 5280, 6: 18943, 12: 29664, 18: 25777, 24: 20336},
("Category Sixes", 8, 3): {0: 2024, 12: 12586, 18: 28717, 24: 35860, 30: 20813},
("Category Sixes", 8, 4): {0: 175, 24: 10907, 30: 72017, 36: 16901},
("Category Sixes", 8, 5): {0: 1, 30: 23224, 36: 66215, 42: 10560},
("Category Sixes", 8, 6): {36: 29563, 42: 70437},
("Category Sixes", 8, 7): {42: 99990, 36: 10},
("Category Sixes", 8, 8): {42: 87843, 48: 12157},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Choice", 0, 0): {0: 100000},
("Category Choice", 0, 1): {0: 100000},
("Category Choice", 0, 2): {0: 100000},
("Category Choice", 0, 3): {0: 100000},
("Category Choice", 0, 4): {0: 100000},
("Category Choice", 0, 5): {0: 100000},
("Category Choice", 0, 6): {0: 100000},
("Category Choice", 0, 7): {0: 100000},
("Category Choice", 0, 8): {0: 100000},
("Category Choice", 1, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Choice", 1, 1): {1: 33315, 3: 66685},
("Category Choice", 1, 2): {1: 32981, 4: 67019},
("Category Choice", 1, 3): {1: 12312, 4: 25020, 5: 62668},
("Category Choice", 1, 4): {1: 11564, 5: 88436},
("Category Choice", 1, 5): {1: 2956, 5: 97044},
("Category Choice", 1, 6): {4: 1024, 6: 65357, 5: 33619},
("Category Choice", 1, 7): {6: 100000},
("Category Choice", 1, 8): {6: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Choice", 2, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Choice", 2, 1): {2: 27810, 6: 72190},
("Category Choice", 2, 2): {2: 10285, 6: 26698, 8: 63017},
("Category Choice", 2, 3): {2: 3965, 8: 96035},
("Category Choice", 2, 4): {2: 143, 8: 33731, 9: 66126},
("Category Choice", 2, 5): {8: 12687, 10: 62544, 9: 24769},
("Category Choice", 2, 6): {10: 100000},
("Category Choice", 2, 7): {11: 66194, 10: 33806},
("Category Choice", 2, 8): {11: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Choice", 3, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Choice", 3, 1): {3: 10461, 6: 27156, 10: 62383},
("Category Choice", 3, 2): {3: 3586, 6: 31281, 12: 65133},
("Category Choice", 3, 3): {3: 1491, 12: 33737, 13: 64772},
("Category Choice", 3, 4): {12: 13802, 14: 60820, 13: 25378},
("Category Choice", 3, 5): {14: 99999, 13: 1},
("Category Choice", 3, 6): {15: 64851, 14: 35149},
("Category Choice", 3, 7): {16: 62341, 15: 24422, 14: 13237},
("Category Choice", 3, 8): {16: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Choice", 4, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Choice", 4, 1): {4: 4748, 10: 28815, 13: 66437},
("Category Choice", 4, 2): {12: 12006, 16: 64226, 13: 23768},
("Category Choice", 4, 3): {16: 32567, 17: 67433},
("Category Choice", 4, 4): {16: 30845, 18: 69155},
("Category Choice", 4, 5): {16: 9568, 19: 68981, 18: 21451},
("Category Choice", 4, 6): {18: 10841, 20: 65051, 19: 24108},
("Category Choice", 4, 7): {18: 4198, 21: 61270, 20: 25195, 19: 9337},
("Category Choice", 4, 8): {21: 99999, 20: 1},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Choice", 5, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Choice", 5, 1): {5: 4485, 13: 35340, 17: 60175},
("Category Choice", 5, 2): {16: 35286, 20: 64714},
("Category Choice", 5, 3): {20: 39254, 22: 60746},
("Category Choice", 5, 4): {20: 35320, 23: 64680},
("Category Choice", 5, 5): {22: 33253, 24: 66747},
("Category Choice", 5, 6): {22: 11089, 25: 66653, 24: 22258},
("Category Choice", 5, 7): {24: 12216, 26: 63214, 22: 50, 25: 24520},
("Category Choice", 5, 8): {24: 4866, 27: 60314, 26: 25089, 25: 9731},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Choice", 6, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Choice", 6, 1): {6: 2276, 17: 33774, 20: 63950},
("Category Choice", 6, 2): {20: 34853, 24: 65147},
("Category Choice", 6, 3): {22: 12477, 26: 64201, 24: 23322},
("Category Choice", 6, 4): {24: 14073, 28: 60688, 26: 25239},
("Category Choice", 6, 5): {26: 35591, 29: 64409},
("Category Choice", 6, 6): {26: 33229, 30: 66771},
("Category Choice", 6, 7): {28: 33078, 31: 66922},
("Category Choice", 6, 8): {28: 12143, 31: 24567, 32: 63290},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Choice", 7, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Choice", 7, 1): {7: 1558, 20: 31716, 23: 66726},
("Category Choice", 7, 2): {23: 34663, 28: 65337},
("Category Choice", 7, 3): {28: 32932, 30: 67062, 23: 6},
("Category Choice", 7, 4): {28: 11163, 32: 66108, 30: 22729},
("Category Choice", 7, 5): {30: 12528, 34: 63034, 32: 24438},
("Category Choice", 7, 6): {30: 4270, 35: 65916, 34: 21485, 32: 8329},
("Category Choice", 7, 7): {32: 4014, 36: 68134, 35: 21006, 34: 6846},
("Category Choice", 7, 8): {34: 3434, 37: 68373, 36: 21550, 35: 6643},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Choice", 8, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Choice", 8, 1): {10: 1400, 23: 36664, 27: 61936},
("Category Choice", 8, 2): {27: 34595, 32: 65405},
("Category Choice", 8, 3): {30: 13097, 35: 62142, 32: 24761},
("Category Choice", 8, 4): {35: 36672, 37: 63328},
("Category Choice", 8, 5): {39: 61292, 35: 14195, 37: 24513},
("Category Choice", 8, 6): {40: 65672, 39: 21040, 35: 4873, 37: 8415},
("Category Choice", 8, 7): {39: 13561, 42: 60493, 40: 25946},
("Category Choice", 8, 8): {39: 5250, 43: 61284, 42: 23421, 40: 10045},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Inverse Choice", 0, 0): {0: 100000},
("Category Inverse Choice", 0, 1): {0: 100000},
("Category Inverse Choice", 0, 2): {0: 100000},
("Category Inverse Choice", 0, 3): {0: 100000},
("Category Inverse Choice", 0, 4): {0: 100000},
("Category Inverse Choice", 0, 5): {0: 100000},
("Category Inverse Choice", 0, 6): {0: 100000},
("Category Inverse Choice", 0, 7): {0: 100000},
("Category Inverse Choice", 0, 8): {0: 100000},
("Category Inverse Choice", 1, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Inverse Choice", 1, 1): {1: 33315, 3: 66685},
("Category Inverse Choice", 1, 2): {1: 32981, 4: 67019},
("Category Inverse Choice", 1, 3): {1: 12312, 4: 25020, 5: 62668},
("Category Inverse Choice", 1, 4): {1: 11564, 5: 88436},
("Category Inverse Choice", 1, 5): {1: 2956, 5: 97044},
("Category Inverse Choice", 1, 6): {4: 1024, 6: 65357, 5: 33619},
("Category Inverse Choice", 1, 7): {6: 100000},
("Category Inverse Choice", 1, 8): {6: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Inverse Choice", 2, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Inverse Choice", 2, 1): {2: 27810, 6: 72190},
("Category Inverse Choice", 2, 2): {2: 10285, 6: 26698, 8: 63017},
("Category Inverse Choice", 2, 3): {2: 3965, 8: 96035},
("Category Inverse Choice", 2, 4): {2: 143, 8: 33731, 9: 66126},
("Category Inverse Choice", 2, 5): {8: 12687, 10: 62544, 9: 24769},
("Category Inverse Choice", 2, 6): {10: 100000},
("Category Inverse Choice", 2, 7): {11: 66194, 10: 33806},
("Category Inverse Choice", 2, 8): {11: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Inverse Choice", 3, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Inverse Choice", 3, 1): {3: 10461, 6: 27156, 10: 62383},
("Category Inverse Choice", 3, 2): {3: 3586, 6: 31281, 12: 65133},
("Category Inverse Choice", 3, 3): {3: 1491, 12: 33737, 13: 64772},
("Category Inverse Choice", 3, 4): {12: 13802, 14: 60820, 13: 25378},
("Category Inverse Choice", 3, 5): {14: 99999, 13: 1},
("Category Inverse Choice", 3, 6): {15: 64851, 14: 35149},
("Category Inverse Choice", 3, 7): {16: 62341, 15: 24422, 14: 13237},
("Category Inverse Choice", 3, 8): {16: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Inverse Choice", 4, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Inverse Choice", 4, 1): {4: 4748, 10: 28815, 13: 66437},
("Category Inverse Choice", 4, 2): {12: 12006, 16: 64226, 13: 23768},
("Category Inverse Choice", 4, 3): {16: 32567, 17: 67433},
("Category Inverse Choice", 4, 4): {16: 30845, 18: 69155},
("Category Inverse Choice", 4, 5): {16: 9568, 19: 68981, 18: 21451},
("Category Inverse Choice", 4, 6): {18: 10841, 20: 65051, 19: 24108},
("Category Inverse Choice", 4, 7): {18: 4198, 21: 61270, 20: 25195, 19: 9337},
("Category Inverse Choice", 4, 8): {21: 99999, 20: 1},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Inverse Choice", 5, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Inverse Choice", 5, 1): {5: 4485, 13: 35340, 17: 60175},
("Category Inverse Choice", 5, 2): {16: 35286, 20: 64714},
("Category Inverse Choice", 5, 3): {20: 39254, 22: 60746},
("Category Inverse Choice", 5, 4): {20: 35320, 23: 64680},
("Category Inverse Choice", 5, 5): {22: 33253, 24: 66747},
("Category Inverse Choice", 5, 6): {22: 11089, 25: 66653, 24: 22258},
("Category Inverse Choice", 5, 7): {24: 12216, 26: 63214, 22: 50, 25: 24520},
("Category Inverse Choice", 5, 8): {24: 4866, 27: 60314, 26: 25089, 25: 9731},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Inverse Choice", 6, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Inverse Choice", 6, 1): {6: 2276, 17: 33774, 20: 63950},
("Category Inverse Choice", 6, 2): {20: 34853, 24: 65147},
("Category Inverse Choice", 6, 3): {22: 12477, 26: 64201, 24: 23322},
("Category Inverse Choice", 6, 4): {24: 14073, 28: 60688, 26: 25239},
("Category Inverse Choice", 6, 5): {26: 35591, 29: 64409},
("Category Inverse Choice", 6, 6): {26: 33229, 30: 66771},
("Category Inverse Choice", 6, 7): {28: 33078, 31: 66922},
("Category Inverse Choice", 6, 8): {28: 12143, 31: 24567, 32: 63290},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Inverse Choice", 7, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Inverse Choice", 7, 1): {7: 1558, 20: 31716, 23: 66726},
("Category Inverse Choice", 7, 2): {23: 34663, 28: 65337},
("Category Inverse Choice", 7, 3): {28: 32932, 30: 67062, 23: 6},
("Category Inverse Choice", 7, 4): {28: 11163, 32: 66108, 30: 22729},
("Category Inverse Choice", 7, 5): {30: 12528, 34: 63034, 32: 24438},
("Category Inverse Choice", 7, 6): {30: 4270, 35: 65916, 34: 21485, 32: 8329},
("Category Inverse Choice", 7, 7): {32: 4014, 36: 68134, 35: 21006, 34: 6846},
("Category Inverse Choice", 7, 8): {34: 3434, 37: 68373, 36: 21550, 35: 6643},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Inverse Choice", 8, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Inverse Choice", 8, 1): {10: 1400, 23: 36664, 27: 61936},
("Category Inverse Choice", 8, 2): {27: 34595, 32: 65405},
("Category Inverse Choice", 8, 3): {30: 13097, 35: 62142, 32: 24761},
("Category Inverse Choice", 8, 4): {35: 36672, 37: 63328},
("Category Inverse Choice", 8, 5): {39: 61292, 35: 14195, 37: 24513},
("Category Inverse Choice", 8, 6): {40: 65672, 39: 21040, 35: 4873, 37: 8415},
("Category Inverse Choice", 8, 7): {39: 13561, 42: 60493, 40: 25946},
("Category Inverse Choice", 8, 8): {39: 5250, 43: 61284, 42: 23421, 40: 10045},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Pair", 0, 0): {0: 100000},
("Category Pair", 0, 1): {0: 100000},
("Category Pair", 0, 2): {0: 100000},
("Category Pair", 0, 3): {0: 100000},
("Category Pair", 0, 4): {0: 100000},
("Category Pair", 0, 5): {0: 100000},
("Category Pair", 0, 6): {0: 100000},
("Category Pair", 0, 7): {0: 100000},
("Category Pair", 0, 8): {0: 100000},
("Category Pair", 1, 0): {0: 100000},
("Category Pair", 1, 1): {0: 100000},
("Category Pair", 1, 2): {0: 100000},
("Category Pair", 1, 3): {0: 100000},
("Category Pair", 1, 4): {0: 100000},
("Category Pair", 1, 5): {0: 100000},
("Category Pair", 1, 6): {0: 100000},
("Category Pair", 1, 7): {0: 100000},
("Category Pair", 1, 8): {0: 100000},
("Category Pair", 2, 0): {0: 100000},
("Category Pair", 2, 1): {0: 83388, 10: 16612},
("Category Pair", 2, 2): {0: 69422, 10: 30578},
("Category Pair", 2, 3): {0: 57830, 10: 42170},
("Category Pair", 2, 4): {0: 48195, 10: 51805},
("Category Pair", 2, 5): {0: 40117, 10: 59883},
("Category Pair", 2, 6): {0: 33286, 10: 66714},
("Category Pair", 2, 7): {0: 27917, 10: 72083},
("Category Pair", 2, 8): {0: 23354, 10: 76646},
("Category Pair", 3, 0): {0: 100000},
("Category Pair", 3, 1): {0: 55518, 10: 44482},
("Category Pair", 3, 2): {0: 30904, 10: 69096},
("Category Pair", 3, 3): {0: 17242, 10: 82758},
("Category Pair", 3, 4): {0: 9486, 10: 90514},
("Category Pair", 3, 5): {0: 5362, 10: 94638},
("Category Pair", 3, 6): {0: 2909, 10: 97091},
("Category Pair", 3, 7): {0: 1574, 10: 98426},
("Category Pair", 3, 8): {0: 902, 10: 99098},
("Category Pair", 4, 0): {0: 100000},
("Category Pair", 4, 1): {0: 27789, 10: 72211},
("Category Pair", 4, 2): {0: 7799, 10: 92201},
("Category Pair", 4, 3): {0: 2113, 10: 97887},
("Category Pair", 4, 4): {0: 601, 10: 99399},
("Category Pair", 4, 5): {0: 155, 10: 99845},
("Category Pair", 4, 6): {0: 43, 10: 99957},
("Category Pair", 4, 7): {0: 10, 10: 99990},
("Category Pair", 4, 8): {0: 3, 10: 99997},
("Category Pair", 5, 0): {0: 100000},
("Category Pair", 5, 1): {0: 9298, 10: 90702},
("Category Pair", 5, 2): {0: 863, 10: 99137},
("Category Pair", 5, 3): {0: 79, 10: 99921},
("Category Pair", 5, 4): {0: 2, 10: 99998},
("Category Pair", 5, 5): {0: 2, 10: 99998},
("Category Pair", 5, 6): {10: 100000},
("Category Pair", 5, 7): {10: 100000},
("Category Pair", 5, 8): {10: 100000},
("Category Pair", 6, 0): {0: 100000},
("Category Pair", 6, 1): {0: 1541, 10: 98459},
("Category Pair", 6, 2): {0: 23, 10: 99977},
("Category Pair", 6, 3): {10: 100000},
("Category Pair", 6, 4): {10: 100000},
("Category Pair", 6, 5): {10: 100000},
("Category Pair", 6, 6): {10: 100000},
("Category Pair", 6, 7): {10: 100000},
("Category Pair", 6, 8): {10: 100000},
("Category Pair", 7, 0): {0: 100000},
("Category Pair", 7, 1): {10: 100000},
("Category Pair", 7, 2): {10: 100000},
("Category Pair", 7, 3): {10: 100000},
("Category Pair", 7, 4): {10: 100000},
("Category Pair", 7, 5): {10: 100000},
("Category Pair", 7, 6): {10: 100000},
("Category Pair", 7, 7): {10: 100000},
("Category Pair", 7, 8): {10: 100000},
("Category Pair", 8, 0): {0: 100000},
("Category Pair", 8, 1): {10: 100000},
("Category Pair", 8, 2): {10: 100000},
("Category Pair", 8, 3): {10: 100000},
("Category Pair", 8, 4): {10: 100000},
("Category Pair", 8, 5): {10: 100000},
("Category Pair", 8, 6): {10: 100000},
("Category Pair", 8, 7): {10: 100000},
("Category Pair", 8, 8): {10: 100000},
("Category Three of a Kind", 0, 0): {0: 100000},
("Category Three of a Kind", 0, 1): {0: 100000},
("Category Three of a Kind", 0, 2): {0: 100000},
("Category Three of a Kind", 0, 3): {0: 100000},
("Category Three of a Kind", 0, 4): {0: 100000},
("Category Three of a Kind", 0, 5): {0: 100000},
("Category Three of a Kind", 0, 6): {0: 100000},
("Category Three of a Kind", 0, 7): {0: 100000},
("Category Three of a Kind", 0, 8): {0: 100000},
("Category Three of a Kind", 1, 0): {0: 100000},
("Category Three of a Kind", 1, 1): {0: 100000},
("Category Three of a Kind", 1, 2): {0: 100000},
("Category Three of a Kind", 1, 3): {0: 100000},
("Category Three of a Kind", 1, 4): {0: 100000},
("Category Three of a Kind", 1, 5): {0: 100000},
("Category Three of a Kind", 1, 6): {0: 100000},
("Category Three of a Kind", 1, 7): {0: 100000},
("Category Three of a Kind", 1, 8): {0: 100000},
("Category Three of a Kind", 2, 0): {0: 100000},
("Category Three of a Kind", 2, 1): {0: 100000},
("Category Three of a Kind", 2, 2): {0: 100000},
("Category Three of a Kind", 2, 3): {0: 100000},
("Category Three of a Kind", 2, 4): {0: 100000},
("Category Three of a Kind", 2, 5): {0: 100000},
("Category Three of a Kind", 2, 6): {0: 100000},
("Category Three of a Kind", 2, 7): {0: 100000},
("Category Three of a Kind", 2, 8): {0: 100000},
("Category Three of a Kind", 3, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Three of a Kind", 3, 1): {0: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Three of a Kind", 3, 2): {0: 88880, 20: 11120},
("Category Three of a Kind", 3, 3): {0: 78187, 20: 21813},
("Category Three of a Kind", 3, 4): {0: 67476, 20: 32524},
("Category Three of a Kind", 3, 5): {0: 57476, 20: 42524},
("Category Three of a Kind", 3, 6): {0: 48510, 20: 51490},
("Category Three of a Kind", 3, 7): {0: 40921, 20: 59079},
("Category Three of a Kind", 3, 8): {0: 34533, 20: 65467},
("Category Three of a Kind", 4, 0): {0: 100000},
("Category Three of a Kind", 4, 1): {0: 90316, 20: 9684},
("Category Three of a Kind", 4, 2): {0: 68401, 20: 31599},
("Category Three of a Kind", 4, 3): {0: 49383, 20: 50617},
("Category Three of a Kind", 4, 4): {0: 34399, 20: 65601},
("Category Three of a Kind", 4, 5): {0: 24154, 20: 75846},
("Category Three of a Kind", 4, 6): {0: 16802, 20: 83198},
("Category Three of a Kind", 4, 7): {0: 11623, 20: 88377},
("Category Three of a Kind", 4, 8): {0: 8105, 20: 91895},
("Category Three of a Kind", 5, 0): {0: 100000},
("Category Three of a Kind", 5, 1): {0: 78629, 20: 21371},
("Category Three of a Kind", 5, 2): {0: 46013, 20: 53987},
("Category Three of a Kind", 5, 3): {0: 25698, 20: 74302},
("Category Three of a Kind", 5, 4): {0: 14205, 20: 85795},
("Category Three of a Kind", 5, 5): {0: 7932, 20: 92068},
("Category Three of a Kind", 5, 6): {0: 4357, 20: 95643},
("Category Three of a Kind", 5, 7): {0: 2432, 20: 97568},
("Category Three of a Kind", 5, 8): {0: 1378, 20: 98622},
("Category Three of a Kind", 6, 0): {0: 100000},
("Category Three of a Kind", 6, 1): {0: 63231, 20: 36769},
("Category Three of a Kind", 6, 2): {0: 26818, 20: 73182},
("Category Three of a Kind", 6, 3): {0: 11075, 20: 88925},
("Category Three of a Kind", 6, 4): {0: 4749, 20: 95251},
("Category Three of a Kind", 6, 5): {0: 1982, 20: 98018},
("Category Three of a Kind", 6, 6): {0: 827, 20: 99173},
("Category Three of a Kind", 6, 7): {0: 358, 20: 99642},
("Category Three of a Kind", 6, 8): {0: 146, 20: 99854},
("Category Three of a Kind", 7, 0): {0: 100000},
("Category Three of a Kind", 7, 1): {0: 45975, 20: 54025},
("Category Three of a Kind", 7, 2): {0: 13207, 20: 86793},
("Category Three of a Kind", 7, 3): {0: 3727, 20: 96273},
("Category Three of a Kind", 7, 4): {0: 1097, 20: 98903},
("Category Three of a Kind", 7, 5): {0: 313, 20: 99687},
("Category Three of a Kind", 7, 6): {0: 96, 20: 99904},
("Category Three of a Kind", 7, 7): {0: 22, 20: 99978},
("Category Three of a Kind", 7, 8): {0: 8, 20: 99992},
("Category Three of a Kind", 8, 0): {0: 100000},
("Category Three of a Kind", 8, 1): {0: 29316, 20: 70684},
("Category Three of a Kind", 8, 2): {0: 5027, 20: 94973},
("Category Three of a Kind", 8, 3): {0: 857, 20: 99143},
("Category Three of a Kind", 8, 4): {0: 162, 20: 99838},
("Category Three of a Kind", 8, 5): {0: 25, 20: 99975},
("Category Three of a Kind", 8, 6): {0: 4, 20: 99996},
("Category Three of a Kind", 8, 7): {0: 1, 20: 99999},
("Category Three of a Kind", 8, 8): {20: 100000},
("Category Four of a Kind", 0, 0): {0: 100000},
("Category Four of a Kind", 0, 1): {0: 100000},
("Category Four of a Kind", 0, 2): {0: 100000},
("Category Four of a Kind", 0, 3): {0: 100000},
("Category Four of a Kind", 0, 4): {0: 100000},
("Category Four of a Kind", 0, 5): {0: 100000},
("Category Four of a Kind", 0, 6): {0: 100000},
("Category Four of a Kind", 0, 7): {0: 100000},
("Category Four of a Kind", 0, 8): {0: 100000},
("Category Four of a Kind", 1, 0): {0: 100000},
("Category Four of a Kind", 1, 1): {0: 100000},
("Category Four of a Kind", 1, 2): {0: 100000},
("Category Four of a Kind", 1, 3): {0: 100000},
("Category Four of a Kind", 1, 4): {0: 100000},
("Category Four of a Kind", 1, 5): {0: 100000},
("Category Four of a Kind", 1, 6): {0: 100000},
("Category Four of a Kind", 1, 7): {0: 100000},
("Category Four of a Kind", 1, 8): {0: 100000},
("Category Four of a Kind", 2, 0): {0: 100000},
("Category Four of a Kind", 2, 1): {0: 100000},
("Category Four of a Kind", 2, 2): {0: 100000},
("Category Four of a Kind", 2, 3): {0: 100000},
("Category Four of a Kind", 2, 4): {0: 100000},
("Category Four of a Kind", 2, 5): {0: 100000},
("Category Four of a Kind", 2, 6): {0: 100000},
("Category Four of a Kind", 2, 7): {0: 100000},
("Category Four of a Kind", 2, 8): {0: 100000},
("Category Four of a Kind", 3, 0): {0: 100000},
("Category Four of a Kind", 3, 1): {0: 100000},
("Category Four of a Kind", 3, 2): {0: 100000},
("Category Four of a Kind", 3, 3): {0: 100000},
("Category Four of a Kind", 3, 4): {0: 100000},
("Category Four of a Kind", 3, 5): {0: 100000},
("Category Four of a Kind", 3, 6): {0: 100000},
("Category Four of a Kind", 3, 7): {0: 100000},
("Category Four of a Kind", 3, 8): {0: 100000},
("Category Four of a Kind", 4, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Four of a Kind", 4, 1): {0: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Four of a Kind", 4, 2): {0: 96122, 30: 3878},
("Category Four of a Kind", 4, 3): {0: 89867, 30: 10133},
("Category Four of a Kind", 4, 4): {0: 81771, 30: 18229},
("Category Four of a Kind", 4, 5): {0: 72893, 30: 27107},
("Category Four of a Kind", 4, 6): {0: 64000, 30: 36000},
("Category Four of a Kind", 4, 7): {0: 55921, 30: 44079},
("Category Four of a Kind", 4, 8): {0: 48175, 30: 51825},
("Category Four of a Kind", 5, 0): {0: 100000},
("Category Four of a Kind", 5, 1): {0: 97938, 30: 2062},
("Category Four of a Kind", 5, 2): {0: 86751, 30: 13249},
("Category Four of a Kind", 5, 3): {0: 70886, 30: 29114},
("Category Four of a Kind", 5, 4): {0: 54807, 30: 45193},
("Category Four of a Kind", 5, 5): {0: 41729, 30: 58271},
("Category Four of a Kind", 5, 6): {0: 30960, 30: 69040},
("Category Four of a Kind", 5, 7): {0: 22207, 30: 77793},
("Category Four of a Kind", 5, 8): {0: 16027, 30: 83973},
("Category Four of a Kind", 6, 0): {0: 100000},
("Category Four of a Kind", 6, 1): {0: 94810, 30: 5190},
("Category Four of a Kind", 6, 2): {0: 73147, 30: 26853},
("Category Four of a Kind", 6, 3): {0: 49873, 30: 50127},
("Category Four of a Kind", 6, 4): {0: 31913, 30: 68087},
("Category Four of a Kind", 6, 5): {0: 19877, 30: 80123},
("Category Four of a Kind", 6, 6): {0: 11973, 30: 88027},
("Category Four of a Kind", 6, 7): {0: 7324, 30: 92676},
("Category Four of a Kind", 6, 8): {0: 4221, 30: 95779},
("Category Four of a Kind", 7, 0): {0: 100000},
("Category Four of a Kind", 7, 1): {0: 89422, 30: 10578},
("Category Four of a Kind", 7, 2): {0: 57049, 30: 42951},
("Category Four of a Kind", 7, 3): {0: 30903, 30: 69097},
("Category Four of a Kind", 7, 4): {0: 15962, 30: 84038},
("Category Four of a Kind", 7, 5): {0: 8148, 30: 91852},
("Category Four of a Kind", 7, 6): {0: 3943, 30: 96057},
("Category Four of a Kind", 7, 7): {0: 1933, 30: 98067},
("Category Four of a Kind", 7, 8): {0: 912, 30: 99088},
("Category Four of a Kind", 8, 0): {0: 100000},
("Category Four of a Kind", 8, 1): {0: 81614, 30: 18386},
("Category Four of a Kind", 8, 2): {0: 40524, 30: 59476},
("Category Four of a Kind", 8, 3): {0: 17426, 30: 82574},
("Category Four of a Kind", 8, 4): {0: 6958, 30: 93042},
("Category Four of a Kind", 8, 5): {0: 2862, 30: 97138},
("Category Four of a Kind", 8, 6): {0: 1049, 30: 98951},
("Category Four of a Kind", 8, 7): {0: 401, 30: 99599},
("Category Four of a Kind", 8, 8): {0: 156, 30: 99844},
("Category Tiny Straight", 0, 0): {0: 100000},
("Category Tiny Straight", 0, 1): {0: 100000},
("Category Tiny Straight", 0, 2): {0: 100000},
("Category Tiny Straight", 0, 3): {0: 100000},
("Category Tiny Straight", 0, 4): {0: 100000},
("Category Tiny Straight", 0, 5): {0: 100000},
("Category Tiny Straight", 0, 6): {0: 100000},
("Category Tiny Straight", 0, 7): {0: 100000},
("Category Tiny Straight", 0, 8): {0: 100000},
("Category Tiny Straight", 1, 0): {0: 100000},
("Category Tiny Straight", 1, 1): {0: 100000},
("Category Tiny Straight", 1, 2): {0: 100000},
("Category Tiny Straight", 1, 3): {0: 100000},
("Category Tiny Straight", 1, 4): {0: 100000},
("Category Tiny Straight", 1, 5): {0: 100000},
("Category Tiny Straight", 1, 6): {0: 100000},
("Category Tiny Straight", 1, 7): {0: 100000},
("Category Tiny Straight", 1, 8): {0: 100000},
("Category Tiny Straight", 2, 0): {0: 100000},
("Category Tiny Straight", 2, 1): {0: 100000},
("Category Tiny Straight", 2, 2): {0: 100000},
("Category Tiny Straight", 2, 3): {0: 100000},
("Category Tiny Straight", 2, 4): {0: 100000},
("Category Tiny Straight", 2, 5): {0: 100000},
("Category Tiny Straight", 2, 6): {0: 100000},
("Category Tiny Straight", 2, 7): {0: 100000},
("Category Tiny Straight", 2, 8): {0: 100000},
("Category Tiny Straight", 3, 0): {0: 100000},
("Category Tiny Straight", 3, 1): {0: 91672, 20: 8328},
("Category Tiny Straight", 3, 2): {0: 79082, 20: 20918},
("Category Tiny Straight", 3, 3): {0: 66490, 20: 33510},
("Category Tiny Straight", 3, 4): {0: 55797, 20: 44203},
("Category Tiny Straight", 3, 5): {0: 46967, 20: 53033},
("Category Tiny Straight", 3, 6): {0: 39595, 20: 60405},
("Category Tiny Straight", 3, 7): {0: 33384, 20: 66616},
("Category Tiny Straight", 3, 8): {0: 28747, 20: 71253},
("Category Tiny Straight", 4, 0): {0: 100000},
("Category Tiny Straight", 4, 1): {0: 78812, 20: 21188},
("Category Tiny Straight", 4, 2): {0: 55525, 20: 44475},
("Category Tiny Straight", 4, 3): {0: 38148, 20: 61852},
("Category Tiny Straight", 4, 4): {0: 26432, 20: 73568},
("Category Tiny Straight", 4, 5): {0: 18225, 20: 81775},
("Category Tiny Straight", 4, 6): {0: 12758, 20: 87242},
("Category Tiny Straight", 4, 7): {0: 8991, 20: 91009},
("Category Tiny Straight", 4, 8): {0: 6325, 20: 93675},
("Category Tiny Straight", 5, 0): {0: 100000},
("Category Tiny Straight", 5, 1): {0: 64979, 20: 35021},
("Category Tiny Straight", 5, 2): {0: 36509, 20: 63491},
("Category Tiny Straight", 5, 3): {0: 20576, 20: 79424},
("Category Tiny Straight", 5, 4): {0: 11585, 20: 88415},
("Category Tiny Straight", 5, 5): {0: 6874, 20: 93126},
("Category Tiny Straight", 5, 6): {0: 3798, 20: 96202},
("Category Tiny Straight", 5, 7): {0: 2214, 20: 97786},
("Category Tiny Straight", 5, 8): {0: 1272, 20: 98728},
("Category Tiny Straight", 6, 0): {0: 100000},
("Category Tiny Straight", 6, 1): {0: 52157, 20: 47843},
("Category Tiny Straight", 6, 2): {0: 23641, 20: 76359},
("Category Tiny Straight", 6, 3): {0: 10883, 20: 89117},
("Category Tiny Straight", 6, 4): {0: 5127, 20: 94873},
("Category Tiny Straight", 6, 5): {0: 2442, 20: 97558},
("Category Tiny Straight", 6, 6): {0: 1158, 20: 98842},
("Category Tiny Straight", 6, 7): {0: 542, 20: 99458},
("Category Tiny Straight", 6, 8): {0: 252, 20: 99748},
("Category Tiny Straight", 7, 0): {0: 100000},
("Category Tiny Straight", 7, 1): {0: 41492, 20: 58508},
("Category Tiny Straight", 7, 2): {0: 15072, 20: 84928},
("Category Tiny Straight", 7, 3): {0: 5905, 20: 94095},
("Category Tiny Straight", 7, 4): {0: 2246, 20: 97754},
("Category Tiny Straight", 7, 5): {0: 942, 20: 99058},
("Category Tiny Straight", 7, 6): {0: 337, 20: 99663},
("Category Tiny Straight", 7, 7): {0: 155, 20: 99845},
("Category Tiny Straight", 7, 8): {0: 61, 20: 99939},
("Category Tiny Straight", 8, 0): {0: 100000},
("Category Tiny Straight", 8, 1): {0: 32993, 20: 67007},
("Category Tiny Straight", 8, 2): {0: 10074, 20: 89926},
("Category Tiny Straight", 8, 3): {0: 3158, 20: 96842},
("Category Tiny Straight", 8, 4): {0: 1060, 20: 98940},
("Category Tiny Straight", 8, 5): {0: 356, 20: 99644},
("Category Tiny Straight", 8, 6): {0: 117, 20: 99883},
("Category Tiny Straight", 8, 7): {0: 32, 20: 99968},
("Category Tiny Straight", 8, 8): {0: 10, 20: 99990},
("Category Small Straight", 0, 0): {0: 100000},
("Category Small Straight", 0, 1): {0: 100000},
("Category Small Straight", 0, 2): {0: 100000},
("Category Small Straight", 0, 3): {0: 100000},
("Category Small Straight", 0, 4): {0: 100000},
("Category Small Straight", 0, 5): {0: 100000},
("Category Small Straight", 0, 6): {0: 100000},
("Category Small Straight", 0, 7): {0: 100000},
("Category Small Straight", 0, 8): {0: 100000},
("Category Small Straight", 1, 0): {0: 100000},
("Category Small Straight", 1, 1): {0: 100000},
("Category Small Straight", 1, 2): {0: 100000},
("Category Small Straight", 1, 3): {0: 100000},
("Category Small Straight", 1, 4): {0: 100000},
("Category Small Straight", 1, 5): {0: 100000},
("Category Small Straight", 1, 6): {0: 100000},
("Category Small Straight", 1, 7): {0: 100000},
("Category Small Straight", 1, 8): {0: 100000},
("Category Small Straight", 2, 0): {0: 100000},
("Category Small Straight", 2, 1): {0: 100000},
("Category Small Straight", 2, 2): {0: 100000},
("Category Small Straight", 2, 3): {0: 100000},
("Category Small Straight", 2, 4): {0: 100000},
("Category Small Straight", 2, 5): {0: 100000},
("Category Small Straight", 2, 6): {0: 100000},
("Category Small Straight", 2, 7): {0: 100000},
("Category Small Straight", 2, 8): {0: 100000},
("Category Small Straight", 3, 0): {0: 100000},
("Category Small Straight", 3, 1): {0: 100000},
("Category Small Straight", 3, 2): {0: 100000},
("Category Small Straight", 3, 3): {0: 100000},
("Category Small Straight", 3, 4): {0: 100000},
("Category Small Straight", 3, 5): {0: 100000},
("Category Small Straight", 3, 6): {0: 100000},
("Category Small Straight", 3, 7): {0: 100000},
("Category Small Straight", 3, 8): {0: 100000},
("Category Small Straight", 4, 0): {0: 100000},
("Category Small Straight", 4, 1): {0: 94516, 30: 5484},
("Category Small Straight", 4, 2): {0: 82700, 30: 17300},
("Category Small Straight", 4, 3): {0: 67926, 30: 32074},
("Category Small Straight", 4, 4): {0: 54265, 30: 45735},
("Category Small Straight", 4, 5): {0: 42130, 30: 57870},
("Category Small Straight", 4, 6): {0: 32536, 30: 67464},
("Category Small Straight", 4, 7): {0: 25008, 30: 74992},
("Category Small Straight", 4, 8): {0: 19595, 30: 80405},
("Category Small Straight", 5, 0): {0: 100000},
("Category Small Straight", 5, 1): {0: 84528, 30: 15472},
("Category Small Straight", 5, 2): {0: 60775, 30: 39225},
("Category Small Straight", 5, 3): {0: 39543, 30: 60457},
("Category Small Straight", 5, 4): {0: 24760, 30: 75240},
("Category Small Straight", 5, 5): {0: 15713, 30: 84287},
("Category Small Straight", 5, 6): {0: 10199, 30: 89801},
("Category Small Straight", 5, 7): {0: 6618, 30: 93382},
("Category Small Straight", 5, 8): {0: 4205, 30: 95795},
("Category Small Straight", 6, 0): {0: 100000},
("Category Small Straight", 6, 1): {0: 73121, 30: 26879},
("Category Small Straight", 6, 2): {0: 41832, 30: 58168},
("Category Small Straight", 6, 3): {0: 21949, 30: 78051},
("Category Small Straight", 6, 4): {0: 11304, 30: 88696},
("Category Small Straight", 6, 5): {0: 6063, 30: 93937},
("Category Small Straight", 6, 6): {0: 3362, 30: 96638},
("Category Small Straight", 6, 7): {0: 1799, 30: 98201},
("Category Small Straight", 6, 8): {0: 1069, 30: 98931},
("Category Small Straight", 7, 0): {0: 100000},
("Category Small Straight", 7, 1): {0: 61837, 30: 38163},
("Category Small Straight", 7, 2): {0: 28202, 30: 71798},
("Category Small Straight", 7, 3): {0: 12187, 30: 87813},
("Category Small Straight", 7, 4): {0: 5427, 30: 94573},
("Category Small Straight", 7, 5): {0: 2444, 30: 97556},
("Category Small Straight", 7, 6): {0: 1144, 30: 98856},
("Category Small Straight", 7, 7): {0: 588, 30: 99412},
("Category Small Straight", 7, 8): {0: 258, 30: 99742},
("Category Small Straight", 8, 0): {0: 100000},
("Category Small Straight", 8, 1): {0: 51394, 30: 48606},
("Category Small Straight", 8, 2): {0: 19090, 30: 80910},
("Category Small Straight", 8, 3): {0: 7104, 30: 92896},
("Category Small Straight", 8, 4): {0: 2645, 30: 97355},
("Category Small Straight", 8, 5): {0: 1010, 30: 98990},
("Category Small Straight", 8, 6): {0: 408, 30: 99592},
("Category Small Straight", 8, 7): {0: 153, 30: 99847},
("Category Small Straight", 8, 8): {0: 78, 30: 99922},
("Category Large Straight", 0, 0): {0: 100000},
("Category Large Straight", 0, 1): {0: 100000},
("Category Large Straight", 0, 2): {0: 100000},
("Category Large Straight", 0, 3): {0: 100000},
("Category Large Straight", 0, 4): {0: 100000},
("Category Large Straight", 0, 5): {0: 100000},
("Category Large Straight", 0, 6): {0: 100000},
("Category Large Straight", 0, 7): {0: 100000},
("Category Large Straight", 0, 8): {0: 100000},
("Category Large Straight", 1, 0): {0: 100000},
("Category Large Straight", 1, 1): {0: 100000},
("Category Large Straight", 1, 2): {0: 100000},
("Category Large Straight", 1, 3): {0: 100000},
("Category Large Straight", 1, 4): {0: 100000},
("Category Large Straight", 1, 5): {0: 100000},
("Category Large Straight", 1, 6): {0: 100000},
("Category Large Straight", 1, 7): {0: 100000},
("Category Large Straight", 1, 8): {0: 100000},
("Category Large Straight", 2, 0): {0: 100000},
("Category Large Straight", 2, 1): {0: 100000},
("Category Large Straight", 2, 2): {0: 100000},
("Category Large Straight", 2, 3): {0: 100000},
("Category Large Straight", 2, 4): {0: 100000},
("Category Large Straight", 2, 5): {0: 100000},
("Category Large Straight", 2, 6): {0: 100000},
("Category Large Straight", 2, 7): {0: 100000},
("Category Large Straight", 2, 8): {0: 100000},
("Category Large Straight", 3, 0): {0: 100000},
("Category Large Straight", 3, 1): {0: 100000},
("Category Large Straight", 3, 2): {0: 100000},
("Category Large Straight", 3, 3): {0: 100000},
("Category Large Straight", 3, 4): {0: 100000},
("Category Large Straight", 3, 5): {0: 100000},
("Category Large Straight", 3, 6): {0: 100000},
("Category Large Straight", 3, 7): {0: 100000},
("Category Large Straight", 3, 8): {0: 100000},
("Category Large Straight", 4, 0): {0: 100000},
("Category Large Straight", 4, 1): {0: 100000},
("Category Large Straight", 4, 2): {0: 100000},
("Category Large Straight", 4, 3): {0: 100000},
("Category Large Straight", 4, 4): {0: 100000},
("Category Large Straight", 4, 5): {0: 100000},
("Category Large Straight", 4, 6): {0: 100000},
("Category Large Straight", 4, 7): {0: 100000},
("Category Large Straight", 4, 8): {0: 100000},
("Category Large Straight", 5, 0): {0: 100000},
("Category Large Straight", 5, 1): {0: 96929, 40: 3071},
("Category Large Straight", 5, 2): {0: 87056, 40: 12944},
("Category Large Straight", 5, 3): {0: 75101, 40: 24899},
("Category Large Straight", 5, 4): {0: 63617, 40: 36383},
("Category Large Straight", 5, 5): {0: 53149, 40: 46851},
("Category Large Straight", 5, 6): {0: 44321, 40: 55679},
("Category Large Straight", 5, 7): {0: 36948, 40: 63052},
("Category Large Straight", 5, 8): {0: 30661, 40: 69339},
("Category Large Straight", 6, 0): {0: 100000},
("Category Large Straight", 6, 1): {0: 90756, 40: 9244},
("Category Large Straight", 6, 2): {0: 69805, 40: 30195},
("Category Large Straight", 6, 3): {0: 49814, 40: 50186},
("Category Large Straight", 6, 4): {0: 35102, 40: 64898},
("Category Large Straight", 6, 5): {0: 24385, 40: 75615},
("Category Large Straight", 6, 6): {0: 17018, 40: 82982},
("Category Large Straight", 6, 7): {0: 11739, 40: 88261},
("Category Large Straight", 6, 8): {0: 7972, 40: 92028},
("Category Large Straight", 7, 0): {0: 100000},
("Category Large Straight", 7, 1): {0: 82840, 40: 17160},
("Category Large Straight", 7, 2): {0: 52821, 40: 47179},
("Category Large Straight", 7, 3): {0: 31348, 40: 68652},
("Category Large Straight", 7, 4): {0: 18166, 40: 81834},
("Category Large Straight", 7, 5): {0: 10690, 40: 89310},
("Category Large Straight", 7, 6): {0: 6051, 40: 93949},
("Category Large Straight", 7, 7): {0: 3617, 40: 96383},
("Category Large Straight", 7, 8): {0: 1941, 40: 98059},
("Category Large Straight", 8, 0): {0: 100000},
("Category Large Straight", 8, 1): {0: 73520, 40: 26480},
("Category Large Straight", 8, 2): {0: 39031, 40: 60969},
("Category Large Straight", 8, 3): {0: 19156, 40: 80844},
("Category Large Straight", 8, 4): {0: 9304, 40: 90696},
("Category Large Straight", 8, 5): {0: 4420, 40: 95580},
("Category Large Straight", 8, 6): {0: 2141, 40: 97859},
("Category Large Straight", 8, 7): {0: 1037, 40: 98963},
("Category Large Straight", 8, 8): {0: 511, 40: 99489},
("Category Full House", 0, 0): {0: 100000},
("Category Full House", 0, 1): {0: 100000},
("Category Full House", 0, 2): {0: 100000},
("Category Full House", 0, 3): {0: 100000},
("Category Full House", 0, 4): {0: 100000},
("Category Full House", 0, 5): {0: 100000},
("Category Full House", 0, 6): {0: 100000},
("Category Full House", 0, 7): {0: 100000},
("Category Full House", 0, 8): {0: 100000},
("Category Full House", 1, 0): {0: 100000},
("Category Full House", 1, 1): {0: 100000},
("Category Full House", 1, 2): {0: 100000},
("Category Full House", 1, 3): {0: 100000},
("Category Full House", 1, 4): {0: 100000},
("Category Full House", 1, 5): {0: 100000},
("Category Full House", 1, 6): {0: 100000},
("Category Full House", 1, 7): {0: 100000},
("Category Full House", 1, 8): {0: 100000},
("Category Full House", 2, 0): {0: 100000},
("Category Full House", 2, 1): {0: 100000},
("Category Full House", 2, 2): {0: 100000},
("Category Full House", 2, 3): {0: 100000},
("Category Full House", 2, 4): {0: 100000},
("Category Full House", 2, 5): {0: 100000},
("Category Full House", 2, 6): {0: 100000},
("Category Full House", 2, 7): {0: 100000},
("Category Full House", 2, 8): {0: 100000},
("Category Full House", 3, 0): {0: 100000},
("Category Full House", 3, 1): {0: 100000},
("Category Full House", 3, 2): {0: 100000},
("Category Full House", 3, 3): {0: 100000},
("Category Full House", 3, 4): {0: 100000},
("Category Full House", 3, 5): {0: 100000},
("Category Full House", 3, 6): {0: 100000},
("Category Full House", 3, 7): {0: 100000},
("Category Full House", 3, 8): {0: 100000},
("Category Full House", 4, 0): {0: 100000},
("Category Full House", 4, 1): {0: 100000},
("Category Full House", 4, 2): {0: 100000},
("Category Full House", 4, 3): {0: 100000},
("Category Full House", 4, 4): {0: 100000},
("Category Full House", 4, 5): {0: 100000},
("Category Full House", 4, 6): {0: 100000},
("Category Full House", 4, 7): {0: 100000},
("Category Full House", 4, 8): {0: 100000},
("Category Full House", 5, 0): {0: 100000},
("Category Full House", 5, 1): {0: 96155, 25: 3845},
("Category Full House", 5, 2): {0: 81391, 25: 18609},
("Category Full House", 5, 3): {0: 64300, 25: 35700},
("Category Full House", 5, 4): {0: 49669, 25: 50331},
("Category Full House", 5, 5): {0: 38019, 25: 61981},
("Category Full House", 5, 6): {0: 29751, 25: 70249},
("Category Full House", 5, 7): {0: 22960, 25: 77040},
("Category Full House", 5, 8): {0: 18650, 25: 81350},
("Category Full House", 6, 0): {0: 100000},
("Category Full House", 6, 1): {0: 82989, 25: 17011},
("Category Full House", 6, 2): {0: 47153, 25: 52847},
("Category Full House", 6, 3): {0: 24151, 25: 75849},
("Category Full House", 6, 4): {0: 12519, 25: 87481},
("Category Full House", 6, 5): {0: 6524, 25: 93476},
("Category Full House", 6, 6): {0: 3606, 25: 96394},
("Category Full House", 6, 7): {0: 1959, 25: 98041},
("Category Full House", 6, 8): {0: 1026, 25: 98974},
("Category Full House", 7, 0): {0: 100000},
("Category Full House", 7, 1): {0: 60232, 25: 39768},
("Category Full House", 7, 2): {0: 18894, 25: 81106},
("Category Full House", 7, 3): {0: 5682, 25: 94318},
("Category Full House", 7, 4): {0: 1706, 25: 98294},
("Category Full House", 7, 5): {0: 522, 25: 99478},
("Category Full House", 7, 6): {0: 146, 25: 99854},
("Category Full House", 7, 7): {0: 54, 25: 99946},
("Category Full House", 7, 8): {0: 18, 25: 99982},
("Category Full House", 8, 0): {0: 100000},
("Category Full House", 8, 1): {0: 35909, 25: 64091},
("Category Full House", 8, 2): {0: 5712, 25: 94288},
("Category Full House", 8, 3): {0: 930, 25: 99070},
("Category Full House", 8, 4): {0: 165, 25: 99835},
("Category Full House", 8, 5): {0: 19, 25: 99981},
("Category Full House", 8, 6): {0: 6, 25: 99994},
("Category Full House", 8, 7): {25: 100000},
("Category Full House", 8, 8): {25: 100000},
("Category Yacht", 0, 0): {0: 100000},
("Category Yacht", 0, 1): {0: 100000},
("Category Yacht", 0, 2): {0: 100000},
("Category Yacht", 0, 3): {0: 100000},
("Category Yacht", 0, 4): {0: 100000},
("Category Yacht", 0, 5): {0: 100000},
("Category Yacht", 0, 6): {0: 100000},
("Category Yacht", 0, 7): {0: 100000},
("Category Yacht", 0, 8): {0: 100000},
("Category Yacht", 1, 0): {0: 100000},
("Category Yacht", 1, 1): {0: 100000},
("Category Yacht", 1, 2): {0: 100000},
("Category Yacht", 1, 3): {0: 100000},
("Category Yacht", 1, 4): {0: 100000},
("Category Yacht", 1, 5): {0: 100000},
("Category Yacht", 1, 6): {0: 100000},
("Category Yacht", 1, 7): {0: 100000},
("Category Yacht", 1, 8): {0: 100000},
("Category Yacht", 2, 0): {0: 100000},
("Category Yacht", 2, 1): {0: 100000},
("Category Yacht", 2, 2): {0: 100000},
("Category Yacht", 2, 3): {0: 100000},
("Category Yacht", 2, 4): {0: 100000},
("Category Yacht", 2, 5): {0: 100000},
("Category Yacht", 2, 6): {0: 100000},
("Category Yacht", 2, 7): {0: 100000},
("Category Yacht", 2, 8): {0: 100000},
("Category Yacht", 3, 0): {0: 100000},
("Category Yacht", 3, 1): {0: 100000},
("Category Yacht", 3, 2): {0: 100000},
("Category Yacht", 3, 3): {0: 100000},
("Category Yacht", 3, 4): {0: 100000},
("Category Yacht", 3, 5): {0: 100000},
("Category Yacht", 3, 6): {0: 100000},
("Category Yacht", 3, 7): {0: 100000},
("Category Yacht", 3, 8): {0: 100000},
("Category Yacht", 4, 0): {0: 100000},
("Category Yacht", 4, 1): {0: 100000},
("Category Yacht", 4, 2): {0: 100000},
("Category Yacht", 4, 3): {0: 100000},
("Category Yacht", 4, 4): {0: 100000},
("Category Yacht", 4, 5): {0: 100000},
("Category Yacht", 4, 6): {0: 100000},
("Category Yacht", 4, 7): {0: 100000},
("Category Yacht", 4, 8): {0: 100000},
("Category Yacht", 5, 0): {0: 100000},
("Category Yacht", 5, 1): {0: 100000},
("Category Yacht", 5, 2): {0: 98727, 50: 1273},
("Category Yacht", 5, 3): {0: 95347, 50: 4653},
("Category Yacht", 5, 4): {0: 89969, 50: 10031},
("Category Yacht", 5, 5): {0: 83124, 50: 16876},
("Category Yacht", 5, 6): {0: 75023, 50: 24977},
("Category Yacht", 5, 7): {0: 67007, 50: 32993},
("Category Yacht", 5, 8): {0: 58618, 50: 41382},
("Category Yacht", 6, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Yacht", 6, 1): {0: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Yacht", 6, 2): {0: 94726, 50: 5274},
("Category Yacht", 6, 3): {0: 84366, 50: 15634},
("Category Yacht", 6, 4): {0: 70782, 50: 29218},
("Category Yacht", 6, 5): {0: 56573, 50: 43427},
("Category Yacht", 6, 6): {0: 44206, 50: 55794},
("Category Yacht", 6, 7): {0: 33578, 50: 66422},
("Category Yacht", 6, 8): {0: 25079, 50: 74921},
("Category Yacht", 7, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Yacht", 7, 1): {0: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Yacht", 7, 2): {0: 87511, 50: 12489},
("Category Yacht", 7, 3): {0: 68252, 50: 31748},
("Category Yacht", 7, 4): {0: 49065, 50: 50935},
("Category Yacht", 7, 5): {0: 33364, 50: 66636},
("Category Yacht", 7, 6): {0: 21483, 50: 78517},
("Category Yacht", 7, 7): {0: 13597, 50: 86403},
("Category Yacht", 7, 8): {0: 8483, 50: 91517},
("Category Yacht", 8, 0): {0: 100000},
("Category Yacht", 8, 1): {0: 97212, 50: 2788},
("Category Yacht", 8, 2): {0: 76962, 50: 23038},
("Category Yacht", 8, 3): {0: 50533, 50: 49467},
("Category Yacht", 8, 4): {0: 29981, 50: 70019},
("Category Yacht", 8, 5): {0: 16776, 50: 83224},
("Category Yacht", 8, 6): {0: 9079, 50: 90921},
("Category Yacht", 8, 7): {0: 4705, 50: 95295},
("Category Yacht", 8, 8): {0: 2363, 50: 97637},
("Category Distincts", 1, 1): {1: 100000},
("Category Distincts", 1, 2): {1: 100000},
("Category Distincts", 1, 3): {1: 100000},
("Category Distincts", 1, 4): {1: 100000},
("Category Distincts", 1, 5): {1: 100000},
("Category Distincts", 1, 6): {1: 100000},
("Category Distincts", 1, 7): {1: 100000},
("Category Distincts", 1, 8): {1: 100000},
("Category Distincts", 2, 1): {1: 16804, 2: 83196},
("Category Distincts", 2, 2): {1: 2686, 2: 97314},
("Category Distincts", 2, 3): {1: 463, 2: 99537},
("Category Distincts", 2, 4): {1: 66, 2: 99934},
("Category Distincts", 2, 5): {1: 11, 2: 99989},
("Category Distincts", 2, 6): {1: 1, 2: 99999},
("Category Distincts", 2, 7): {2: 100000},
("Category Distincts", 2, 8): {2: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Distincts", 3, 1): {1: 2760, 2: 97240},
("Category Distincts", 3, 2): {1: 414, 3: 84996, 2: 14590},
("Category Distincts", 3, 3): {1: 109, 3: 99891},
("Category Distincts", 3, 4): {2: 11, 3: 99989},
("Category Distincts", 3, 5): {3: 100000},
("Category Distincts", 3, 6): {3: 100000},
("Category Distincts", 3, 7): {3: 100000},
("Category Distincts", 3, 8): {3: 100000},
("Category Distincts", 4, 1): {1: 458, 3: 83376, 2: 16166},
("Category Distincts", 4, 2): {1: 26, 4: 61232, 3: 37802, 2: 940},
("Category Distincts", 4, 3): {2: 3, 4: 97020, 3: 2977},
("Category Distincts", 4, 4): {4: 100000},
("Category Distincts", 4, 5): {4: 100000},
("Category Distincts", 4, 6): {4: 100000},
("Category Distincts", 4, 7): {4: 100000},
("Category Distincts", 4, 8): {4: 100000},
("Category Distincts", 5, 1): {1: 159, 3: 99841},
("Category Distincts", 5, 2): {2: 18, 4: 88167, 3: 11815},
("Category Distincts", 5, 3): {4: 100000},
("Category Distincts", 5, 4): {5: 67650, 4: 32350},
("Category Distincts", 5, 5): {5: 100000},
("Category Distincts", 5, 6): {5: 100000},
("Category Distincts", 5, 7): {5: 100000},
("Category Distincts", 5, 8): {5: 100000},
("Category Distincts", 6, 1): {1: 39, 4: 74998, 3: 24963},
("Category Distincts", 6, 2): {2: 1, 5: 61568, 4: 37296, 3: 1135},
("Category Distincts", 6, 3): {5: 93157, 4: 6843},
("Category Distincts", 6, 4): {5: 100000},
("Category Distincts", 6, 5): {5: 100000},
("Category Distincts", 6, 6): {5: 100000},
("Category Distincts", 6, 7): {5: 100000},
("Category Distincts", 6, 8): {6: 65828, 5: 34172},
("Category Distincts", 7, 1): {1: 13, 4: 99987},
("Category Distincts", 7, 2): {5: 99580, 4: 420},
("Category Distincts", 7, 3): {5: 100000},
("Category Distincts", 7, 4): {5: 100000},
("Category Distincts", 7, 5): {6: 71744, 5: 28256},
("Category Distincts", 7, 6): {6: 100000},
("Category Distincts", 7, 7): {6: 100000},
("Category Distincts", 7, 8): {6: 100000},
("Category Distincts", 8, 1): {4: 100000},
("Category Distincts", 8, 2): {5: 99981, 4: 19},
("Category Distincts", 8, 3): {6: 63291, 5: 36709},
("Category Distincts", 8, 4): {6: 99994, 5: 6},
("Category Distincts", 8, 5): {6: 100000},
("Category Distincts", 8, 6): {6: 100000},
("Category Distincts", 8, 7): {6: 100000},
("Category Distincts", 8, 8): {6: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Two times Ones", 0, 0): {0: 100000},
("Category Two times Ones", 0, 1): {0: 100000},
("Category Two times Ones", 0, 2): {0: 100000},
("Category Two times Ones", 0, 3): {0: 100000},
("Category Two times Ones", 0, 4): {0: 100000},
("Category Two times Ones", 0, 5): {0: 100000},
("Category Two times Ones", 0, 6): {0: 100000},
("Category Two times Ones", 0, 7): {0: 100000},
("Category Two times Ones", 0, 8): {0: 100000},
("Category Two times Ones", 1, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Two times Ones", 1, 1): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Two times Ones", 1, 2): {0: 69690, 2: 30310},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Two times Ones", 1, 3): {0: 57818, 2: 42182},
("Category Two times Ones", 1, 4): {0: 48418, 2: 51582},
("Category Two times Ones", 1, 5): {0: 40301, 2: 59699},
("Category Two times Ones", 1, 6): {0: 33558, 2: 66442},
("Category Two times Ones", 1, 7): {0: 28182, 2: 71818},
("Category Two times Ones", 1, 8): {0: 23406, 2: 76594},
("Category Two times Ones", 2, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Two times Ones", 2, 1): {0: 69724, 2: 30276},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Two times Ones", 2, 2): {0: 48238, 2: 51762},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Two times Ones", 2, 3): {0: 33290, 2: 66710},
("Category Two times Ones", 2, 4): {0: 23136, 2: 76864},
("Category Two times Ones", 2, 5): {0: 16146, 2: 48200, 4: 35654},
("Category Two times Ones", 2, 6): {0: 11083, 2: 44497, 4: 44420},
("Category Two times Ones", 2, 7): {0: 7662, 2: 40343, 4: 51995},
("Category Two times Ones", 2, 8): {0: 5354, 2: 35526, 4: 59120},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Two times Ones", 3, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Two times Ones", 3, 1): {0: 58021, 2: 41979},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Two times Ones", 3, 2): {0: 33548, 2: 66452},
("Category Two times Ones", 3, 3): {0: 19375, 2: 42372, 4: 38253},
("Category Two times Ones", 3, 4): {0: 10998, 2: 36435, 4: 52567},
("Category Two times Ones", 3, 5): {0: 7954, 4: 92046},
("Category Two times Ones", 3, 6): {0: 347, 4: 99653},
("Category Two times Ones", 3, 7): {0: 2, 4: 62851, 6: 37147},
("Category Two times Ones", 3, 8): {6: 99476, 4: 524},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Two times Ones", 4, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Two times Ones", 4, 1): {0: 48235, 2: 51765},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Two times Ones", 4, 2): {0: 23289, 2: 40678, 4: 36033},
("Category Two times Ones", 4, 3): {0: 11177, 2: 32677, 4: 56146},
("Category Two times Ones", 4, 4): {0: 5522, 4: 60436, 6: 34042},
("Category Two times Ones", 4, 5): {0: 4358, 6: 95642},
("Category Two times Ones", 4, 6): {0: 20, 6: 99980},
("Category Two times Ones", 4, 7): {6: 100000},
("Category Two times Ones", 4, 8): {6: 65250, 8: 34750},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Two times Ones", 5, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Two times Ones", 5, 1): {0: 40028, 2: 59972},
("Category Two times Ones", 5, 2): {0: 16009, 2: 35901, 4: 48090},
("Category Two times Ones", 5, 3): {0: 6820, 4: 57489, 6: 35691},
("Category Two times Ones", 5, 4): {0: 5285, 6: 94715},
("Category Two times Ones", 5, 5): {0: 18, 6: 66613, 8: 33369},
("Category Two times Ones", 5, 6): {8: 99073, 6: 927},
("Category Two times Ones", 5, 7): {8: 100000},
("Category Two times Ones", 5, 8): {8: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Two times Ones", 6, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Two times Ones", 6, 1): {0: 33502, 2: 66498},
("Category Two times Ones", 6, 2): {0: 13681, 4: 59162, 2: 27157},
("Category Two times Ones", 6, 3): {0: 5486, 6: 94514},
("Category Two times Ones", 6, 4): {0: 190, 6: 62108, 8: 37702},
("Category Two times Ones", 6, 5): {8: 99882, 6: 118},
("Category Two times Ones", 6, 6): {8: 65144, 10: 34856},
("Category Two times Ones", 6, 7): {10: 99524, 8: 476},
("Category Two times Ones", 6, 8): {10: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Two times Ones", 7, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Two times Ones", 7, 1): {0: 27683, 2: 39060, 4: 33257},
("Category Two times Ones", 7, 2): {0: 8683, 4: 54932, 6: 36385},
("Category Two times Ones", 7, 3): {0: 373, 6: 66572, 8: 33055},
("Category Two times Ones", 7, 4): {8: 99816, 6: 184},
("Category Two times Ones", 7, 5): {8: 58124, 10: 41876},
("Category Two times Ones", 7, 6): {10: 99948, 8: 52},
("Category Two times Ones", 7, 7): {10: 62549, 12: 37451},
("Category Two times Ones", 7, 8): {12: 99818, 10: 182},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Two times Ones", 8, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Two times Ones", 8, 1): {0: 23378, 2: 37157, 4: 39465},
("Category Two times Ones", 8, 2): {0: 5602, 6: 94398},
("Category Two times Ones", 8, 3): {0: 8, 6: 10911, 8: 89081},
("Category Two times Ones", 8, 4): {8: 59809, 10: 40191},
("Category Two times Ones", 8, 5): {10: 68808, 12: 31114, 8: 78},
("Category Two times Ones", 8, 6): {12: 98712, 10: 1287, 8: 1},
("Category Two times Ones", 8, 7): {12: 100000},
("Category Two times Ones", 8, 8): {12: 59018, 14: 40982},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Half of Sixes", 0, 0): {0: 100000},
("Category Half of Sixes", 0, 1): {0: 100000},
("Category Half of Sixes", 0, 2): {0: 100000},
("Category Half of Sixes", 0, 3): {0: 100000},
("Category Half of Sixes", 0, 4): {0: 100000},
("Category Half of Sixes", 0, 5): {0: 100000},
("Category Half of Sixes", 0, 6): {0: 100000},
("Category Half of Sixes", 0, 7): {0: 100000},
("Category Half of Sixes", 0, 8): {0: 100000},
("Category Half of Sixes", 1, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Half of Sixes", 1, 1): {0: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Half of Sixes", 1, 2): {0: 69569, 3: 30431},
("Category Half of Sixes", 1, 3): {0: 57872, 3: 42128},
("Category Half of Sixes", 1, 4): {0: 48081, 3: 51919},
("Category Half of Sixes", 1, 5): {0: 40271, 3: 59729},
("Category Half of Sixes", 1, 6): {0: 33201, 3: 66799},
("Category Half of Sixes", 1, 7): {0: 27903, 3: 72097},
("Category Half of Sixes", 1, 8): {0: 23240, 3: 76760},
("Category Half of Sixes", 2, 0): {0: 100000},
("Category Half of Sixes", 2, 1): {0: 69419, 3: 30581},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Half of Sixes", 2, 2): {0: 48202, 3: 51798},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Half of Sixes", 2, 3): {0: 33376, 3: 66624},
("Category Half of Sixes", 2, 4): {0: 23276, 3: 49810, 6: 26914},
("Category Half of Sixes", 2, 5): {0: 16092, 3: 47718, 6: 36190},
("Category Half of Sixes", 2, 6): {0: 11232, 3: 44515, 6: 44253},
("Category Half of Sixes", 2, 7): {0: 7589, 3: 40459, 6: 51952},
("Category Half of Sixes", 2, 8): {0: 5447, 3: 35804, 6: 58749},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Half of Sixes", 3, 0): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Half of Sixes", 3, 1): {0: 57964, 3: 42036},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Half of Sixes", 3, 2): {0: 33637, 3: 44263, 6: 22100},
("Category Half of Sixes", 3, 3): {0: 19520, 3: 42382, 6: 38098},
("Category Half of Sixes", 3, 4): {0: 11265, 3: 35772, 6: 52963},
("Category Half of Sixes", 3, 5): {0: 6419, 3: 28916, 6: 43261, 9: 21404},
("Category Half of Sixes", 3, 6): {0: 3810, 3: 22496, 6: 44388, 9: 29306},
("Category Half of Sixes", 3, 7): {0: 1317, 6: 30047, 9: 68636},
("Category Half of Sixes", 3, 8): {0: 750, 9: 99250},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Half of Sixes", 4, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Half of Sixes", 4, 1): {0: 48121, 3: 51879},
("Category Half of Sixes", 4, 2): {0: 23296, 3: 40989, 6: 35715},
("Category Half of Sixes", 4, 3): {0: 11233, 3: 32653, 6: 35710, 9: 20404},
("Category Half of Sixes", 4, 4): {0: 5463, 3: 23270, 6: 37468, 9: 33799},
("Category Half of Sixes", 4, 5): {0: 5225, 6: 29678, 9: 65097},
("Category Half of Sixes", 4, 6): {0: 3535, 9: 96465},
("Category Half of Sixes", 4, 7): {0: 6, 9: 72939, 12: 27055},
("Category Half of Sixes", 4, 8): {9: 25326, 12: 74674},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Half of Sixes", 5, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Half of Sixes", 5, 1): {0: 40183, 3: 59817},
("Category Half of Sixes", 5, 2): {0: 16197, 3: 35494, 6: 48309},
("Category Half of Sixes", 5, 3): {0: 6583, 3: 23394, 6: 34432, 9: 35591},
("Category Half of Sixes", 5, 4): {0: 5007, 6: 25159, 9: 49038, 12: 20796},
("Category Half of Sixes", 5, 5): {0: 2900, 9: 38935, 12: 58165},
("Category Half of Sixes", 5, 6): {0: 2090, 12: 97910},
("Category Half of Sixes", 5, 7): {12: 99994, 9: 6},
("Category Half of Sixes", 5, 8): {12: 73524, 15: 26476},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Half of Sixes", 6, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Half of Sixes", 6, 1): {0: 33473, 3: 40175, 6: 26352},
("Category Half of Sixes", 6, 2): {0: 11147, 3: 29592, 6: 32630, 9: 26631},
("Category Half of Sixes", 6, 3): {0: 2460, 6: 21148, 9: 55356, 12: 21036},
("Category Half of Sixes", 6, 4): {0: 997, 9: 29741, 12: 69262},
("Category Half of Sixes", 6, 5): {0: 831, 12: 76328, 15: 22841},
("Category Half of Sixes", 6, 6): {12: 29960, 15: 70040},
("Category Half of Sixes", 6, 7): {15: 100000},
("Category Half of Sixes", 6, 8): {15: 79456, 18: 20544},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Half of Sixes", 7, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Half of Sixes", 7, 1): {0: 27933, 3: 39105, 6: 32962},
("Category Half of Sixes", 7, 2): {0: 7794, 3: 23896, 6: 31832, 9: 36478},
("Category Half of Sixes", 7, 3): {0: 1321, 9: 40251, 12: 58428},
("Category Half of Sixes", 7, 4): {0: 370, 12: 74039, 15: 25591},
("Category Half of Sixes", 7, 5): {0: 6, 15: 98660, 12: 1334},
("Category Half of Sixes", 7, 6): {15: 73973, 18: 26027},
("Category Half of Sixes", 7, 7): {18: 100000},
("Category Half of Sixes", 7, 8): {18: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Half of Sixes", 8, 0): {0: 100000},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Half of Sixes", 8, 1): {0: 23337, 3: 37232, 6: 39431},
("Category Half of Sixes", 8, 2): {0: 4652, 6: 29310, 9: 45517, 12: 20521},
("Category Half of Sixes", 8, 3): {0: 1300, 12: 77919, 15: 20781},
("Category Half of Sixes", 8, 4): {0: 21, 15: 98678, 12: 1301},
("Category Half of Sixes", 8, 5): {15: 68893, 18: 31107},
("Category Half of Sixes", 8, 6): {18: 100000},
("Category Half of Sixes", 8, 7): {18: 69986, 21: 30014},
("Category Half of Sixes", 8, 8): {21: 98839, 18: 1161},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category Twos and Threes", 1, 1): {0: 66466, 2: 33534},
("Category Twos and Threes", 1, 2): {0: 55640, 2: 44360},
Yacht Dice: Fix logic (again) so that score doesn't drop when receiving item (#4044) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. * add filler item name * Textual fixes and changes * Remove Victory item and use event instead. * Revert "Remove Victory item and use event instead." This reverts commit c2f7d674d392a3acbc1db8614411164ba3b28bff. * Revert "Textual fixes and changes" This reverts commit e9432f92454979fcd5a31f8517586585362a7ab7. * Remove Victory item and make it an event instead * Yacht Dice logic fix, no decreasing score when obtain item take 2 * Logic fix: Revert max_tries and mults, change ordering * Remove spaces :^) * Updated weights that are stochastically ordered by dice/roll In the trimming of the weights, sometimes it having 4 rolls would be better than having 5 rolls. I did a check that this does not happen for any dice increment or roll increment * Swap for-loops to increase performance This method is faster if the first for-loop contains fewer items. Since the function is called with, typically, `dist2` having less items, let's loop over `dist2` first. This makes the entire program 10% faster. * Remove options with 0 chance from list --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-10-22 19:07:44 +00:00
("Category Twos and Threes", 1, 3): {0: 46223, 2: 53777},
("Category Twos and Threes", 1, 4): {0: 38552, 2: 61448},
("Category Twos and Threes", 1, 5): {0: 32320, 2: 67680},
("Category Twos and Threes", 1, 6): {0: 10797, 3: 66593, 2: 22610},
("Category Twos and Threes", 1, 7): {0: 9307, 3: 90693},
("Category Twos and Threes", 1, 8): {0: 2173, 3: 97827},
("Category Twos and Threes", 2, 1): {0: 44565, 2: 55435},
("Category Twos and Threes", 2, 2): {0: 30855, 2: 69145},
("Category Twos and Threes", 2, 3): {0: 9977, 3: 67663, 2: 22360},
("Category Twos and Threes", 2, 4): {0: 7252, 3: 92748},
("Category Twos and Threes", 2, 5): {0: 1135, 3: 98865},
("Category Twos and Threes", 2, 6): {0: 121, 3: 99879},
("Category Twos and Threes", 2, 7): {2: 48, 5: 60169, 3: 39783},
("Category Twos and Threes", 2, 8): {5: 99998, 3: 2},
("Category Twos and Threes", 3, 1): {0: 29892, 2: 70108},
("Category Twos and Threes", 3, 2): {0: 8977, 3: 69968, 2: 21055},
("Category Twos and Threes", 3, 3): {0: 5237, 3: 94763},
("Category Twos and Threes", 3, 4): {2: 1781, 5: 65980, 3: 32239},
("Category Twos and Threes", 3, 5): {2: 609, 6: 65803, 5: 22563, 3: 11025},
("Category Twos and Threes", 3, 6): {6: 100000},
("Category Twos and Threes", 3, 7): {6: 100000},
("Category Twos and Threes", 3, 8): {6: 100000},
("Category Twos and Threes", 4, 1): {0: 11769, 3: 60627, 2: 27604},
("Category Twos and Threes", 4, 2): {2: 15639, 4: 60280, 3: 24081},
("Category Twos and Threes", 4, 3): {5: 72517, 2: 4298, 4: 16567, 3: 6618},
("Category Twos and Threes", 4, 4): {6: 73910, 5: 18921, 2: 1121, 4: 4322, 3: 1726},
("Category Twos and Threes", 4, 5): {2: 430, 7: 61608, 6: 28377, 5: 7264, 4: 1659, 3: 662},
("Category Twos and Threes", 4, 6): {9: 60343, 7: 24434, 6: 15223},
("Category Twos and Threes", 4, 7): {9: 100000},
("Category Twos and Threes", 4, 8): {9: 100000},
("Category Twos and Threes", 5, 1): {0: 11610, 3: 88390},
("Category Twos and Threes", 5, 2): {5: 70562, 3: 11158, 2: 534, 4: 17746},
("Category Twos and Threes", 5, 3): {6: 74716, 5: 23240, 3: 774, 2: 37, 4: 1233},
("Category Twos and Threes", 5, 4): {8: 68531, 6: 29461, 5: 1962, 3: 18, 4: 28},
("Category Twos and Threes", 5, 5): {9: 70635, 8: 26461, 6: 2860, 5: 44},
("Category Twos and Threes", 5, 6): {9: 100000},
("Category Twos and Threes", 5, 7): {11: 67606, 9: 32394},
("Category Twos and Threes", 5, 8): {12: 68354, 11: 21395, 9: 10251},
("Category Twos and Threes", 6, 1): {2: 4096, 4: 64713, 3: 31191},
("Category Twos and Threes", 6, 2): {2: 169, 6: 68210, 5: 22433, 3: 3547, 4: 5641},
("Category Twos and Threes", 6, 3): {2: 11, 8: 68425, 6: 23593, 5: 7338, 3: 244, 4: 389},
("Category Twos and Threes", 6, 4): {9: 73054, 8: 26109, 6: 787, 5: 50},
("Category Twos and Threes", 6, 5): {8: 8568, 11: 68223, 9: 23209},
("Category Twos and Threes", 6, 6): {12: 70373, 11: 20213, 9: 9414},
("Category Twos and Threes", 6, 7): {12: 100000},
("Category Twos and Threes", 6, 8): {14: 68062, 12: 31938},
("Category Twos and Threes", 7, 1): {2: 1390, 5: 66048, 4: 21972, 3: 10590},
("Category Twos and Threes", 7, 2): {2: 22, 8: 60665, 5: 11253, 6: 26834, 3: 473, 4: 753},
("Category Twos and Threes", 7, 3): {9: 70126, 8: 26169, 5: 909, 6: 2772, 3: 9, 4: 15},
("Category Twos and Threes", 7, 4): {11: 70543, 9: 28824, 8: 633},
("Category Twos and Threes", 7, 5): {12: 74745, 11: 22893, 9: 2173, 8: 189},
("Category Twos and Threes", 7, 6): {11: 7636, 14: 69766, 12: 22598},
("Category Twos and Threes", 7, 7): {15: 71620, 14: 19800, 12: 8580},
("Category Twos and Threes", 7, 8): {14: 10952, 16: 61407, 15: 27641},
("Category Twos and Threes", 8, 1): {2: 555, 6: 60067, 5: 26375, 4: 8774, 3: 4229},
("Category Twos and Threes", 8, 2): {8: 99967, 2: 13, 6: 20},
("Category Twos and Threes", 8, 3): {8: 10167, 11: 65964, 9: 23869},
("Category Twos and Threes", 8, 4): {11: 37966, 13: 62034},
("Category Twos and Threes", 8, 5): {11: 9059, 15: 64126, 12: 26815},
("Category Twos and Threes", 8, 6): {14: 14139, 17: 60581, 11: 2, 15: 25278},
("Category Twos and Threes", 8, 7): {14: 5173, 18: 63415, 17: 22164, 15: 9248},
("Category Twos and Threes", 8, 8): {18: 100000},
("Category Sum of Odds", 1, 1): {0: 66572, 3: 33428},
("Category Sum of Odds", 1, 2): {0: 44489, 3: 55511},
("Category Sum of Odds", 1, 3): {0: 26778, 3: 33412, 5: 39810},
("Category Sum of Odds", 1, 4): {0: 18191, 5: 81809},
("Category Sum of Odds", 1, 5): {0: 2299, 5: 97701},
("Category Sum of Odds", 1, 6): {0: 101, 5: 99899},
("Category Sum of Odds", 1, 7): {5: 100000},
("Category Sum of Odds", 1, 8): {5: 100000},
("Category Sum of Odds", 2, 1): {0: 66571, 3: 33429},
("Category Sum of Odds", 2, 2): {0: 38206, 4: 61794},
("Category Sum of Odds", 2, 3): {3: 15100, 8: 34337, 4: 24422, 5: 26141},
("Category Sum of Odds", 2, 4): {3: 4389, 8: 75870, 5: 19741},
("Category Sum of Odds", 2, 5): {8: 66180, 10: 33820},
("Category Sum of Odds", 2, 6): {10: 99075, 8: 925},
("Category Sum of Odds", 2, 7): {10: 100000},
("Category Sum of Odds", 2, 8): {10: 100000},
("Category Sum of Odds", 3, 1): {0: 19440, 3: 80560},
("Category Sum of Odds", 3, 2): {0: 3843, 3: 30607, 6: 65550},
("Category Sum of Odds", 3, 3): {8: 99451, 3: 126, 4: 204, 5: 219},
("Category Sum of Odds", 3, 4): {8: 39493, 9: 60507},
("Category Sum of Odds", 3, 5): {8: 25186, 13: 36226, 9: 38588},
("Category Sum of Odds", 3, 6): {13: 99387, 8: 242, 9: 371},
("Category Sum of Odds", 3, 7): {13: 63989, 15: 36011},
("Category Sum of Odds", 3, 8): {15: 99350, 13: 650},
("Category Sum of Odds", 4, 1): {0: 7100, 3: 29425, 5: 63475},
("Category Sum of Odds", 4, 2): {0: 1227, 3: 30702, 8: 68071},
("Category Sum of Odds", 4, 3): {8: 34941, 10: 65059},
("Category Sum of Odds", 4, 4): {8: 30671, 11: 69329},
("Category Sum of Odds", 4, 5): {8: 20766, 13: 79234},
("Category Sum of Odds", 4, 6): {13: 67313, 18: 32687},
("Category Sum of Odds", 4, 7): {13: 12063, 18: 87937},
("Category Sum of Odds", 4, 8): {18: 66936, 20: 33064},
("Category Sum of Odds", 5, 1): {0: 2404, 3: 31470, 6: 66126},
("Category Sum of Odds", 5, 2): {6: 12689, 11: 60256, 8: 27055},
("Category Sum of Odds", 5, 3): {10: 36853, 13: 63147},
("Category Sum of Odds", 5, 4): {13: 38005, 15: 61994, 10: 1},
("Category Sum of Odds", 5, 5): {13: 33747, 16: 66253},
("Category Sum of Odds", 5, 6): {13: 23587, 18: 76413},
("Category Sum of Odds", 5, 7): {18: 67776, 23: 32224},
("Category Sum of Odds", 5, 8): {23: 99176, 18: 824},
("Category Sum of Odds", 6, 1): {0: 791, 3: 32146, 7: 67063},
("Category Sum of Odds", 6, 2): {11: 38567, 13: 61432, 8: 1},
("Category Sum of Odds", 6, 3): {15: 65880, 11: 5075, 13: 29045},
("Category Sum of Odds", 6, 4): {15: 37367, 18: 62633},
("Category Sum of Odds", 6, 5): {18: 38038, 20: 61948, 15: 14},
("Category Sum of Odds", 6, 6): {18: 33838, 21: 66162},
("Category Sum of Odds", 6, 7): {18: 16130, 23: 83870},
("Category Sum of Odds", 6, 8): {23: 66748, 28: 33252},
("Category Sum of Odds", 7, 1): {5: 12019, 9: 63507, 7: 24474},
("Category Sum of Odds", 7, 2): {11: 37365, 15: 62635},
("Category Sum of Odds", 7, 3): {15: 36250, 18: 63750},
("Category Sum of Odds", 7, 4): {18: 37627, 21: 62373},
("Category Sum of Odds", 7, 5): {20: 35127, 23: 64873},
("Category Sum of Odds", 7, 6): {20: 12629, 25: 64047, 23: 23324},
("Category Sum of Odds", 7, 7): {23: 32409, 26: 67591},
("Category Sum of Odds", 7, 8): {23: 22322, 28: 77678},
("Category Sum of Odds", 8, 1): {5: 4088, 10: 65985, 9: 21602, 7: 8325},
("Category Sum of Odds", 8, 2): {13: 35686, 17: 64314},
("Category Sum of Odds", 8, 3): {17: 13770, 21: 62013, 18: 24217},
("Category Sum of Odds", 8, 4): {21: 37763, 24: 62237},
("Category Sum of Odds", 8, 5): {23: 12631, 26: 66541, 21: 4, 24: 20824},
("Category Sum of Odds", 8, 6): {23: 4929, 29: 60982, 26: 25964, 24: 8125},
("Category Sum of Odds", 8, 7): {23: 1608, 30: 67370, 29: 19899, 26: 8472, 24: 2651},
("Category Sum of Odds", 8, 8): {28: 4861, 32: 61811, 30: 25729, 29: 7599},
("Category Sum of Evens", 1, 1): {0: 66318, 4: 33682},
("Category Sum of Evens", 1, 2): {0: 44331, 4: 55669},
("Category Sum of Evens", 1, 3): {0: 29576, 4: 35040, 6: 35384},
("Category Sum of Evens", 1, 4): {0: 22612, 6: 77388},
("Category Sum of Evens", 1, 5): {0: 3566, 6: 96434},
("Category Sum of Evens", 1, 6): {0: 209, 6: 99791},
("Category Sum of Evens", 1, 7): {0: 3, 6: 99997},
("Category Sum of Evens", 1, 8): {6: 100000},
("Category Sum of Evens", 2, 1): {0: 25229, 2: 36083, 6: 38688},
("Category Sum of Evens", 2, 2): {0: 57, 4: 38346, 8: 37232, 2: 81, 6: 24284},
("Category Sum of Evens", 2, 3): {6: 39504, 10: 37060, 4: 1, 8: 23435},
("Category Sum of Evens", 2, 4): {10: 99495, 6: 317, 8: 188},
("Category Sum of Evens", 2, 5): {10: 69597, 12: 30403},
("Category Sum of Evens", 2, 6): {12: 98377, 10: 1623},
("Category Sum of Evens", 2, 7): {12: 100000},
("Category Sum of Evens", 2, 8): {12: 100000},
("Category Sum of Evens", 3, 1): {0: 76, 4: 38332, 8: 37178, 2: 109, 6: 24305},
("Category Sum of Evens", 3, 2): {8: 67248, 12: 32556, 4: 196},
("Category Sum of Evens", 3, 3): {10: 44843, 14: 33195, 8: 213, 12: 21749},
("Category Sum of Evens", 3, 4): {10: 37288, 14: 62712},
("Category Sum of Evens", 3, 5): {14: 61196, 16: 38802, 10: 2},
("Category Sum of Evens", 3, 6): {16: 99621, 14: 379},
("Category Sum of Evens", 3, 7): {16: 67674, 18: 32326},
("Category Sum of Evens", 3, 8): {18: 100000},
("Category Sum of Evens", 4, 1): {6: 37636, 10: 40039, 4: 32, 8: 22293},
("Category Sum of Evens", 4, 2): {10: 57689, 14: 42258, 6: 53},
("Category Sum of Evens", 4, 3): {14: 67801, 18: 32152, 10: 47},
("Category Sum of Evens", 4, 4): {18: 98878, 14: 1122},
("Category Sum of Evens", 4, 5): {18: 60401, 20: 39599},
("Category Sum of Evens", 4, 6): {20: 64396, 22: 35186, 18: 418},
("Category Sum of Evens", 4, 7): {22: 99697, 20: 302, 18: 1},
("Category Sum of Evens", 4, 8): {22: 100000},
("Category Sum of Evens", 5, 1): {8: 35338, 12: 41027, 6: 22, 10: 23613},
("Category Sum of Evens", 5, 2): {12: 37027, 18: 35856, 10: 10, 14: 27107},
("Category Sum of Evens", 5, 3): {18: 68230, 22: 31735, 14: 35},
("Category Sum of Evens", 5, 4): {18: 14880, 22: 53608, 24: 31512},
("Category Sum of Evens", 5, 5): {24: 98732, 18: 275, 22: 993},
("Category Sum of Evens", 5, 6): {24: 61498, 26: 38502},
("Category Sum of Evens", 5, 7): {26: 65201, 28: 34488, 24: 311},
("Category Sum of Evens", 5, 8): {28: 99648, 26: 351, 24: 1},
("Category Sum of Evens", 6, 1): {10: 34538, 14: 41426, 8: 4, 12: 24032},
("Category Sum of Evens", 6, 2): {16: 43552, 22: 31546, 14: 235, 12: 121, 18: 24546},
("Category Sum of Evens", 6, 3): {22: 68714, 26: 31239, 18: 47},
("Category Sum of Evens", 6, 4): {26: 59168, 28: 33835, 22: 4791, 18: 1, 24: 2205},
("Category Sum of Evens", 6, 5): {26: 44386, 30: 32920, 28: 22694},
("Category Sum of Evens", 6, 6): {30: 98992, 26: 667, 28: 341},
("Category Sum of Evens", 6, 7): {30: 60806, 32: 39194},
("Category Sum of Evens", 6, 8): {32: 64584, 34: 35252, 30: 164},
("Category Sum of Evens", 7, 1): {12: 40703, 18: 30507, 10: 1, 14: 28789},
("Category Sum of Evens", 7, 2): {22: 60249, 24: 38366, 12: 1, 18: 767, 16: 614, 14: 3},
("Category Sum of Evens", 7, 3): {24: 47964, 30: 30240, 22: 4, 26: 21792},
("Category Sum of Evens", 7, 4): {30: 63108, 32: 35114, 24: 1778},
("Category Sum of Evens", 7, 5): {32: 62062, 34: 37406, 30: 523, 26: 6, 28: 3},
("Category Sum of Evens", 7, 6): {32: 40371, 36: 35507, 34: 24122},
("Category Sum of Evens", 7, 7): {34: 44013, 38: 31749, 32: 4, 36: 24234},
("Category Sum of Evens", 7, 8): {38: 99116, 34: 570, 36: 314},
("Category Sum of Evens", 8, 1): {18: 66673, 20: 31528, 12: 1054, 14: 745},
("Category Sum of Evens", 8, 2): {22: 40918, 28: 33610, 24: 25472},
("Category Sum of Evens", 8, 3): {28: 40893, 32: 41346, 24: 17, 30: 17737, 26: 7},
("Category Sum of Evens", 8, 4): {32: 63665, 36: 36316, 28: 19},
("Category Sum of Evens", 8, 5): {36: 58736, 38: 40234, 32: 1030},
("Category Sum of Evens", 8, 6): {36: 57946, 40: 42054},
("Category Sum of Evens", 8, 7): {38: 34984, 42: 39622, 36: 2, 40: 25392},
("Category Sum of Evens", 8, 8): {42: 65137, 44: 34611, 38: 146, 40: 106},
("Category Double Threes and Fours", 1, 1): {0: 66749, 6: 33251},
("Category Double Threes and Fours", 1, 2): {0: 44675, 6: 55325},
("Category Double Threes and Fours", 1, 3): {0: 29592, 6: 35261, 8: 35147},
("Category Double Threes and Fours", 1, 4): {0: 24601, 6: 29406, 8: 45993},
("Category Double Threes and Fours", 1, 5): {0: 20499, 6: 24420, 8: 55081},
("Category Double Threes and Fours", 1, 6): {0: 17116, 6: 20227, 8: 62657},
("Category Double Threes and Fours", 1, 7): {0: 14193, 6: 17060, 8: 68747},
("Category Double Threes and Fours", 1, 8): {0: 11977, 6: 13924, 8: 74099},
("Category Double Threes and Fours", 2, 1): {0: 44382, 6: 22191, 8: 33427},
("Category Double Threes and Fours", 2, 2): {0: 5, 6: 46088, 12: 30763, 8: 23144},
("Category Double Threes and Fours", 2, 3): {0: 5, 6: 30159, 12: 32725, 14: 37111},
("Category Double Threes and Fours", 2, 4): {6: 20533, 14: 79467},
("Category Double Threes and Fours", 2, 5): {14: 69789, 16: 30211},
("Category Double Threes and Fours", 2, 6): {16: 99978, 14: 22},
("Category Double Threes and Fours", 2, 7): {16: 100000},
("Category Double Threes and Fours", 2, 8): {16: 100000},
("Category Double Threes and Fours", 3, 1): {0: 8, 6: 49139, 12: 26176, 8: 24677},
("Category Double Threes and Fours", 3, 2): {0: 5, 6: 24942, 12: 27065, 14: 47988},
("Category Double Threes and Fours", 3, 3): {6: 12743, 14: 56776, 20: 30481},
("Category Double Threes and Fours", 3, 4): {14: 9753, 20: 90247},
("Category Double Threes and Fours", 3, 5): {20: 61293, 22: 38707},
("Category Double Threes and Fours", 3, 6): {22: 99615, 20: 385},
("Category Double Threes and Fours", 3, 7): {22: 67267, 24: 32733},
("Category Double Threes and Fours", 3, 8): {24: 100000},
("Category Double Threes and Fours", 4, 1): {6: 26819, 12: 39789, 14: 33392},
("Category Double Threes and Fours", 4, 2): {14: 63726, 20: 36011, 6: 106, 12: 157},
("Category Double Threes and Fours", 4, 3): {20: 69628, 24: 30158, 14: 214},
("Category Double Threes and Fours", 4, 4): {20: 11409, 24: 57067, 26: 31524},
("Category Double Threes and Fours", 4, 5): {20: 6566, 26: 57047, 28: 36387},
("Category Double Threes and Fours", 4, 6): {28: 63694, 30: 35203, 20: 113, 26: 990},
("Category Double Threes and Fours", 4, 7): {30: 98893, 28: 1092, 26: 15},
("Category Double Threes and Fours", 4, 8): {30: 100000},
("Category Double Threes and Fours", 5, 1): {6: 16042, 14: 83958},
("Category Double Threes and Fours", 5, 2): {14: 44329, 20: 24912, 24: 30759},
("Category Double Threes and Fours", 5, 3): {24: 57603, 28: 42155, 20: 242},
("Category Double Threes and Fours", 5, 4): {26: 32446, 30: 43875, 24: 21, 28: 23658},
("Category Double Threes and Fours", 5, 5): {30: 69209, 34: 30672, 26: 69, 28: 50},
("Category Double Threes and Fours", 5, 6): {34: 63882, 36: 35323, 30: 795},
("Category Double Threes and Fours", 5, 7): {36: 65178, 38: 34598, 34: 222, 30: 2},
("Category Double Threes and Fours", 5, 8): {38: 99654, 36: 345, 34: 1},
("Category Double Threes and Fours", 6, 1): {14: 68079, 18: 31921},
("Category Double Threes and Fours", 6, 2): {14: 14542, 24: 48679, 28: 36779},
("Category Double Threes and Fours", 6, 3): {28: 62757, 34: 36962, 24: 281},
("Category Double Threes and Fours", 6, 4): {34: 68150, 38: 30771, 28: 604, 26: 1, 30: 474},
("Category Double Threes and Fours", 6, 5): {38: 68332, 40: 30833, 34: 823, 28: 12},
("Category Double Threes and Fours", 6, 6): {40: 67631, 42: 31174, 38: 1181, 34: 14},
("Category Double Threes and Fours", 6, 7): {42: 63245, 44: 35699, 40: 1038, 38: 18},
("Category Double Threes and Fours", 6, 8): {44: 64056, 46: 35162, 42: 770, 40: 12},
("Category Double Threes and Fours", 7, 1): {14: 14976, 18: 54685, 22: 30339},
("Category Double Threes and Fours", 7, 2): {14: 10532, 28: 55372, 32: 34096},
("Category Double Threes and Fours", 7, 3): {32: 42786, 40: 32123, 28: 2, 34: 25089},
("Category Double Threes and Fours", 7, 4): {38: 46172, 44: 31648, 32: 226, 40: 21954},
("Category Double Threes and Fours", 7, 5): {44: 64883, 46: 34437, 38: 460, 32: 2, 40: 218},
("Category Double Threes and Fours", 7, 6): {44: 43458, 48: 33715, 46: 22827},
("Category Double Threes and Fours", 7, 7): {46: 44472, 50: 32885, 44: 15, 48: 22628},
("Category Double Threes and Fours", 7, 8): {48: 41682, 52: 37868, 46: 18, 50: 20432},
("Category Double Threes and Fours", 8, 1): {14: 14227, 22: 85773},
("Category Double Threes and Fours", 8, 2): {22: 7990, 32: 56319, 36: 35691},
("Category Double Threes and Fours", 8, 3): {32: 19914, 40: 43585, 44: 36501},
("Category Double Threes and Fours", 8, 4): {44: 63232, 48: 36613, 32: 48, 40: 107},
("Category Double Threes and Fours", 8, 5): {48: 62939, 52: 36798, 44: 263},
("Category Double Threes and Fours", 8, 6): {52: 60756, 54: 38851, 48: 392, 44: 1},
("Category Double Threes and Fours", 8, 7): {54: 62281, 56: 37262, 52: 455, 48: 2},
("Category Double Threes and Fours", 8, 8): {56: 67295, 60: 32064, 54: 637, 52: 4},
("Category Quadruple Ones and Twos", 1, 1): {0: 66567, 4: 16803, 8: 16630},
("Category Quadruple Ones and Twos", 1, 2): {0: 44809, 4: 27448, 8: 27743},
("Category Quadruple Ones and Twos", 1, 3): {0: 37100, 4: 23184, 8: 39716},
("Category Quadruple Ones and Twos", 1, 4): {0: 30963, 4: 19221, 8: 49816},
("Category Quadruple Ones and Twos", 1, 5): {0: 25316, 4: 16079, 8: 58605},
("Category Quadruple Ones and Twos", 1, 6): {0: 14381, 8: 85619},
("Category Quadruple Ones and Twos", 1, 7): {0: 4137, 8: 95863},
("Category Quadruple Ones and Twos", 1, 8): {0: 1004, 8: 98996},
("Category Quadruple Ones and Twos", 2, 1): {0: 44566, 4: 22273, 8: 33161},
("Category Quadruple Ones and Twos", 2, 2): {0: 19963, 4: 24890, 8: 32262, 12: 22885},
("Category Quadruple Ones and Twos", 2, 3): {0: 13766, 4: 17158, 8: 34907, 12: 18539, 16: 15630},
("Category Quadruple Ones and Twos", 2, 4): {0: 6655, 8: 30200, 12: 26499, 16: 36646},
("Category Quadruple Ones and Twos", 2, 5): {0: 982, 8: 16426, 12: 24307, 16: 58285},
("Category Quadruple Ones and Twos", 2, 6): {0: 68, 8: 9887, 16: 90045},
("Category Quadruple Ones and Twos", 2, 7): {0: 11, 16: 99989},
("Category Quadruple Ones and Twos", 2, 8): {16: 100000},
("Category Quadruple Ones and Twos", 3, 1): {0: 29440, 4: 22574, 8: 27747, 12: 20239},
("Category Quadruple Ones and Twos", 3, 2): {0: 8857, 4: 16295, 8: 26434, 12: 22986, 16: 25428},
("Category Quadruple Ones and Twos", 3, 3): {0: 3649, 8: 15314, 12: 24619, 16: 38944, 20: 17474},
("Category Quadruple Ones and Twos", 3, 4): {0: 11, 8: 8430, 16: 41259, 20: 50300},
("Category Quadruple Ones and Twos", 3, 5): {20: 80030, 24: 19902, 8: 11, 16: 57},
("Category Quadruple Ones and Twos", 3, 6): {20: 23895, 24: 76105},
("Category Quadruple Ones and Twos", 3, 7): {24: 100000},
("Category Quadruple Ones and Twos", 3, 8): {24: 100000},
("Category Quadruple Ones and Twos", 4, 1): {0: 19691, 4: 19657, 8: 27288, 12: 16126, 16: 17238},
("Category Quadruple Ones and Twos", 4, 2): {0: 1222, 4: 15703, 12: 24015, 16: 34944, 20: 24116},
("Category Quadruple Ones and Twos", 4, 3): {0: 227, 12: 14519, 20: 62257, 24: 22997},
("Category Quadruple Ones and Twos", 4, 4): {0: 11, 20: 17266, 24: 67114, 28: 15609},
("Category Quadruple Ones and Twos", 4, 5): {24: 27365, 28: 72632, 20: 3},
("Category Quadruple Ones and Twos", 4, 6): {28: 81782, 32: 18215, 24: 3},
("Category Quadruple Ones and Twos", 4, 7): {28: 22319, 32: 77681},
("Category Quadruple Ones and Twos", 4, 8): {32: 100000},
("Category Quadruple Ones and Twos", 5, 1): {0: 13112, 4: 16534, 8: 24718, 12: 18558, 16: 27078},
("Category Quadruple Ones and Twos", 5, 2): {0: 21, 4: 15200, 16: 28784, 20: 32131, 24: 23864},
("Category Quadruple Ones and Twos", 5, 3): {0: 4, 16: 8475, 24: 66718, 28: 24803},
("Category Quadruple Ones and Twos", 5, 4): {28: 76149, 32: 23289, 24: 550, 20: 12},
("Category Quadruple Ones and Twos", 5, 5): {32: 81110, 36: 16222, 28: 2663, 24: 5},
("Category Quadruple Ones and Twos", 5, 6): {32: 18542, 36: 81458},
("Category Quadruple Ones and Twos", 5, 7): {36: 82036, 40: 17964},
("Category Quadruple Ones and Twos", 5, 8): {36: 27864, 40: 72136},
("Category Quadruple Ones and Twos", 6, 1): {0: 6419, 8: 16963, 12: 22116, 16: 33903, 20: 20599},
("Category Quadruple Ones and Twos", 6, 2): {0: 5, 16: 8913, 24: 67749, 28: 23333},
("Category Quadruple Ones and Twos", 6, 3): {28: 71779, 32: 27514, 16: 82, 24: 625},
("Category Quadruple Ones and Twos", 6, 4): {32: 72333, 36: 27328, 28: 337, 24: 2},
("Category Quadruple Ones and Twos", 6, 5): {36: 73993, 40: 25138, 32: 865, 28: 4},
("Category Quadruple Ones and Twos", 6, 6): {40: 80918, 44: 17126, 36: 1934, 32: 22},
("Category Quadruple Ones and Twos", 6, 7): {40: 20298, 44: 79702},
("Category Quadruple Ones and Twos", 6, 8): {44: 81077, 48: 18923},
("Category Quadruple Ones and Twos", 7, 1): {0: 508, 8: 10298, 16: 41828, 20: 30853, 24: 16513},
("Category Quadruple Ones and Twos", 7, 2): {16: 7429, 28: 69817, 32: 22754},
("Category Quadruple Ones and Twos", 7, 3): {32: 82871, 40: 16531, 16: 57, 28: 541},
("Category Quadruple Ones and Twos", 7, 4): {36: 67601, 44: 17916, 32: 909, 40: 13569, 28: 5},
("Category Quadruple Ones and Twos", 7, 5): {40: 67395, 48: 17447, 36: 364, 44: 14790, 32: 4},
("Category Quadruple Ones and Twos", 7, 6): {48: 91242, 40: 7151, 36: 38, 44: 1569},
("Category Quadruple Ones and Twos", 7, 7): {48: 80854, 52: 19146},
("Category Quadruple Ones and Twos", 7, 8): {48: 25334, 52: 74666},
("Category Quadruple Ones and Twos", 8, 1): {0: 119, 16: 17496, 20: 26705, 24: 55680},
("Category Quadruple Ones and Twos", 8, 2): {24: 569, 32: 72257, 36: 21817, 28: 5357},
("Category Quadruple Ones and Twos", 8, 3): {36: 66654, 44: 18473, 32: 1396, 40: 13477},
("Category Quadruple Ones and Twos", 8, 4): {44: 73954, 48: 22240, 36: 3178, 40: 628},
("Category Quadruple Ones and Twos", 8, 5): {48: 76082, 52: 22415, 44: 1500, 36: 3},
("Category Quadruple Ones and Twos", 8, 6): {52: 74901, 56: 21332, 48: 3766, 44: 1},
("Category Quadruple Ones and Twos", 8, 7): {56: 96171, 52: 3640, 48: 189},
("Category Quadruple Ones and Twos", 8, 8): {56: 78035, 60: 21965},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category Micro Straight", 1, 1): {0: 100000},
("Category Micro Straight", 1, 2): {0: 100000},
("Category Micro Straight", 1, 3): {0: 100000},
("Category Micro Straight", 1, 4): {0: 100000},
("Category Micro Straight", 1, 5): {0: 100000},
("Category Micro Straight", 1, 6): {0: 100000},
("Category Micro Straight", 1, 7): {0: 100000},
("Category Micro Straight", 1, 8): {0: 100000},
("Category Micro Straight", 2, 1): {0: 72326, 10: 27674},
("Category Micro Straight", 2, 2): {0: 48546, 10: 51454},
("Category Micro Straight", 2, 3): {0: 32619, 10: 67381},
("Category Micro Straight", 2, 4): {0: 21659, 10: 78341},
("Category Micro Straight", 2, 5): {0: 14288, 10: 85712},
("Category Micro Straight", 2, 6): {0: 9882, 10: 90118},
("Category Micro Straight", 2, 7): {0: 6502, 10: 93498},
("Category Micro Straight", 2, 8): {0: 4161, 10: 95839},
("Category Micro Straight", 3, 1): {0: 41943, 10: 58057},
("Category Micro Straight", 3, 2): {0: 15524, 10: 84476},
("Category Micro Straight", 3, 3): {0: 5700, 10: 94300},
("Category Micro Straight", 3, 4): {0: 2127, 10: 97873},
("Category Micro Straight", 3, 5): {0: 744, 10: 99256},
("Category Micro Straight", 3, 6): {0: 260, 10: 99740},
("Category Micro Straight", 3, 7): {0: 115, 10: 99885},
("Category Micro Straight", 3, 8): {0: 34, 10: 99966},
("Category Micro Straight", 4, 1): {0: 22307, 10: 77693},
("Category Micro Straight", 4, 2): {0: 4420, 10: 95580},
("Category Micro Straight", 4, 3): {0: 806, 10: 99194},
("Category Micro Straight", 4, 4): {0: 205, 10: 99795},
("Category Micro Straight", 4, 5): {0: 20, 10: 99980},
("Category Micro Straight", 4, 6): {0: 5, 10: 99995},
("Category Micro Straight", 4, 7): {0: 1, 10: 99999},
("Category Micro Straight", 4, 8): {0: 1, 10: 99999},
("Category Micro Straight", 5, 1): {0: 11685, 10: 88315},
("Category Micro Straight", 5, 2): {0: 1141, 10: 98859},
("Category Micro Straight", 5, 3): {0: 119, 10: 99881},
("Category Micro Straight", 5, 4): {0: 11, 10: 99989},
("Category Micro Straight", 5, 5): {0: 1, 10: 99999},
("Category Micro Straight", 5, 6): {10: 100000},
("Category Micro Straight", 5, 7): {10: 100000},
("Category Micro Straight", 5, 8): {10: 100000},
("Category Micro Straight", 6, 1): {0: 5937, 10: 94063},
("Category Micro Straight", 6, 2): {0: 307, 10: 99693},
("Category Micro Straight", 6, 3): {0: 9, 10: 99991},
("Category Micro Straight", 6, 4): {0: 1, 10: 99999},
("Category Micro Straight", 6, 5): {10: 100000},
("Category Micro Straight", 6, 6): {10: 100000},
("Category Micro Straight", 6, 7): {10: 100000},
("Category Micro Straight", 6, 8): {10: 100000},
("Category Micro Straight", 7, 1): {0: 3072, 10: 96928},
("Category Micro Straight", 7, 2): {0: 85, 10: 99915},
("Category Micro Straight", 7, 3): {0: 2, 10: 99998},
("Category Micro Straight", 7, 4): {10: 100000},
("Category Micro Straight", 7, 5): {10: 100000},
("Category Micro Straight", 7, 6): {10: 100000},
("Category Micro Straight", 7, 7): {10: 100000},
("Category Micro Straight", 7, 8): {10: 100000},
("Category Micro Straight", 8, 1): {0: 1544, 10: 98456},
("Category Micro Straight", 8, 2): {0: 15, 10: 99985},
("Category Micro Straight", 8, 3): {10: 100000},
("Category Micro Straight", 8, 4): {10: 100000},
("Category Micro Straight", 8, 5): {10: 100000},
("Category Micro Straight", 8, 6): {10: 100000},
("Category Micro Straight", 8, 7): {10: 100000},
("Category Micro Straight", 8, 8): {10: 100000},
("Category Three Odds", 1, 1): {0: 100000},
("Category Three Odds", 1, 2): {0: 100000},
("Category Three Odds", 1, 3): {0: 100000},
("Category Three Odds", 1, 4): {0: 100000},
("Category Three Odds", 1, 5): {0: 100000},
("Category Three Odds", 1, 6): {0: 100000},
("Category Three Odds", 1, 7): {0: 100000},
("Category Three Odds", 1, 8): {0: 100000},
("Category Three Odds", 2, 1): {0: 100000},
("Category Three Odds", 2, 2): {0: 100000},
("Category Three Odds", 2, 3): {0: 100000},
("Category Three Odds", 2, 4): {0: 100000},
("Category Three Odds", 2, 5): {0: 100000},
("Category Three Odds", 2, 6): {0: 100000},
("Category Three Odds", 2, 7): {0: 100000},
("Category Three Odds", 2, 8): {0: 100000},
("Category Three Odds", 3, 1): {0: 87592, 20: 12408},
("Category Three Odds", 3, 2): {0: 57855, 20: 42145},
("Category Three Odds", 3, 3): {0: 32668, 20: 67332},
("Category Three Odds", 3, 4): {0: 17508, 20: 82492},
("Category Three Odds", 3, 5): {0: 9156, 20: 90844},
("Category Three Odds", 3, 6): {0: 4572, 20: 95428},
("Category Three Odds", 3, 7): {0: 2325, 20: 97675},
("Category Three Odds", 3, 8): {0: 1116, 20: 98884},
("Category Three Odds", 4, 1): {0: 68669, 20: 31331},
("Category Three Odds", 4, 2): {0: 26140, 20: 73860},
("Category Three Odds", 4, 3): {0: 7837, 20: 92163},
("Category Three Odds", 4, 4): {0: 2169, 20: 97831},
("Category Three Odds", 4, 5): {0: 516, 20: 99484},
("Category Three Odds", 4, 6): {0: 156, 20: 99844},
("Category Three Odds", 4, 7): {0: 40, 20: 99960},
("Category Three Odds", 4, 8): {0: 12, 20: 99988},
("Category Three Odds", 5, 1): {0: 49908, 20: 50092},
("Category Three Odds", 5, 2): {0: 10373, 20: 89627},
("Category Three Odds", 5, 3): {0: 1640, 20: 98360},
("Category Three Odds", 5, 4): {0: 223, 20: 99777},
("Category Three Odds", 5, 5): {0: 24, 20: 99976},
("Category Three Odds", 5, 6): {0: 3, 20: 99997},
("Category Three Odds", 5, 7): {0: 1, 20: 99999},
("Category Three Odds", 5, 8): {20: 100000},
("Category Three Odds", 6, 1): {0: 34566, 20: 65434},
("Category Three Odds", 6, 2): {0: 3766, 20: 96234},
("Category Three Odds", 6, 3): {0: 291, 20: 99709},
("Category Three Odds", 6, 4): {0: 22, 20: 99978},
("Category Three Odds", 6, 5): {20: 100000},
("Category Three Odds", 6, 6): {20: 100000},
("Category Three Odds", 6, 7): {20: 100000},
("Category Three Odds", 6, 8): {20: 100000},
("Category Three Odds", 7, 1): {0: 22722, 20: 77278},
("Category Three Odds", 7, 2): {0: 1291, 20: 98709},
("Category Three Odds", 7, 3): {0: 38, 20: 99962},
("Category Three Odds", 7, 4): {0: 2, 20: 99998},
("Category Three Odds", 7, 5): {20: 100000},
("Category Three Odds", 7, 6): {20: 100000},
("Category Three Odds", 7, 7): {20: 100000},
("Category Three Odds", 7, 8): {20: 100000},
("Category Three Odds", 8, 1): {0: 14556, 20: 85444},
("Category Three Odds", 8, 2): {0: 430, 20: 99570},
("Category Three Odds", 8, 3): {0: 3, 20: 99997},
("Category Three Odds", 8, 4): {20: 100000},
("Category Three Odds", 8, 5): {20: 100000},
("Category Three Odds", 8, 6): {20: 100000},
("Category Three Odds", 8, 7): {20: 100000},
("Category Three Odds", 8, 8): {20: 100000},
("Category 1-2-1 Consecutive", 1, 1): {0: 100000},
("Category 1-2-1 Consecutive", 1, 2): {0: 100000},
("Category 1-2-1 Consecutive", 1, 3): {0: 100000},
("Category 1-2-1 Consecutive", 1, 4): {0: 100000},
("Category 1-2-1 Consecutive", 1, 5): {0: 100000},
("Category 1-2-1 Consecutive", 1, 6): {0: 100000},
("Category 1-2-1 Consecutive", 1, 7): {0: 100000},
("Category 1-2-1 Consecutive", 1, 8): {0: 100000},
("Category 1-2-1 Consecutive", 2, 1): {0: 100000},
("Category 1-2-1 Consecutive", 2, 2): {0: 100000},
("Category 1-2-1 Consecutive", 2, 3): {0: 100000},
("Category 1-2-1 Consecutive", 2, 4): {0: 100000},
("Category 1-2-1 Consecutive", 2, 5): {0: 100000},
("Category 1-2-1 Consecutive", 2, 6): {0: 100000},
("Category 1-2-1 Consecutive", 2, 7): {0: 100000},
("Category 1-2-1 Consecutive", 2, 8): {0: 100000},
("Category 1-2-1 Consecutive", 3, 1): {0: 100000},
("Category 1-2-1 Consecutive", 3, 2): {0: 100000},
("Category 1-2-1 Consecutive", 3, 3): {0: 100000},
("Category 1-2-1 Consecutive", 3, 4): {0: 100000},
("Category 1-2-1 Consecutive", 3, 5): {0: 100000},
("Category 1-2-1 Consecutive", 3, 6): {0: 100000},
("Category 1-2-1 Consecutive", 3, 7): {0: 100000},
("Category 1-2-1 Consecutive", 3, 8): {0: 100000},
("Category 1-2-1 Consecutive", 4, 1): {0: 96371, 30: 3629},
("Category 1-2-1 Consecutive", 4, 2): {0: 86605, 30: 13395},
("Category 1-2-1 Consecutive", 4, 3): {0: 75037, 30: 24963},
("Category 1-2-1 Consecutive", 4, 4): {0: 63656, 30: 36344},
("Category 1-2-1 Consecutive", 4, 5): {0: 53869, 30: 46131},
("Category 1-2-1 Consecutive", 4, 6): {0: 45131, 30: 54869},
("Category 1-2-1 Consecutive", 4, 7): {0: 37535, 30: 62465},
("Category 1-2-1 Consecutive", 4, 8): {0: 31425, 30: 68575},
("Category 1-2-1 Consecutive", 5, 1): {0: 86632, 30: 13368},
("Category 1-2-1 Consecutive", 5, 2): {0: 62779, 30: 37221},
("Category 1-2-1 Consecutive", 5, 3): {0: 46034, 30: 53966},
("Category 1-2-1 Consecutive", 5, 4): {0: 34983, 30: 65017},
("Category 1-2-1 Consecutive", 5, 5): {0: 28056, 30: 71944},
("Category 1-2-1 Consecutive", 5, 6): {0: 23150, 30: 76850},
("Category 1-2-1 Consecutive", 5, 7): {0: 19577, 30: 80423},
("Category 1-2-1 Consecutive", 5, 8): {0: 17613, 30: 82387},
("Category 1-2-1 Consecutive", 6, 1): {0: 71928, 30: 28072},
("Category 1-2-1 Consecutive", 6, 2): {0: 40724, 30: 59276},
("Category 1-2-1 Consecutive", 6, 3): {0: 26723, 30: 73277},
("Category 1-2-1 Consecutive", 6, 4): {0: 19685, 30: 80315},
("Category 1-2-1 Consecutive", 6, 5): {0: 15460, 30: 84540},
("Category 1-2-1 Consecutive", 6, 6): {0: 12526, 30: 87474},
("Category 1-2-1 Consecutive", 6, 7): {0: 10014, 30: 89986},
("Category 1-2-1 Consecutive", 6, 8): {0: 8251, 30: 91749},
("Category 1-2-1 Consecutive", 7, 1): {0: 55544, 30: 44456},
("Category 1-2-1 Consecutive", 7, 2): {0: 24840, 30: 75160},
("Category 1-2-1 Consecutive", 7, 3): {0: 15102, 30: 84898},
("Category 1-2-1 Consecutive", 7, 4): {0: 10541, 30: 89459},
("Category 1-2-1 Consecutive", 7, 5): {0: 7720, 30: 92280},
("Category 1-2-1 Consecutive", 7, 6): {0: 5554, 30: 94446},
("Category 1-2-1 Consecutive", 7, 7): {0: 4106, 30: 95894},
("Category 1-2-1 Consecutive", 7, 8): {0: 3025, 30: 96975},
("Category 1-2-1 Consecutive", 8, 1): {0: 40693, 30: 59307},
("Category 1-2-1 Consecutive", 8, 2): {0: 14827, 30: 85173},
("Category 1-2-1 Consecutive", 8, 3): {0: 8195, 30: 91805},
("Category 1-2-1 Consecutive", 8, 4): {0: 5383, 30: 94617},
("Category 1-2-1 Consecutive", 8, 5): {0: 3395, 30: 96605},
("Category 1-2-1 Consecutive", 8, 6): {0: 2299, 30: 97701},
("Category 1-2-1 Consecutive", 8, 7): {0: 1412, 30: 98588},
("Category 1-2-1 Consecutive", 8, 8): {0: 872, 30: 99128},
("Category Three Distinct Dice", 1, 1): {0: 100000},
("Category Three Distinct Dice", 1, 2): {0: 100000},
("Category Three Distinct Dice", 1, 3): {0: 100000},
("Category Three Distinct Dice", 1, 4): {0: 100000},
("Category Three Distinct Dice", 1, 5): {0: 100000},
("Category Three Distinct Dice", 1, 6): {0: 100000},
("Category Three Distinct Dice", 1, 7): {0: 100000},
("Category Three Distinct Dice", 1, 8): {0: 100000},
("Category Three Distinct Dice", 2, 1): {0: 100000},
("Category Three Distinct Dice", 2, 2): {0: 100000},
("Category Three Distinct Dice", 2, 3): {0: 100000},
("Category Three Distinct Dice", 2, 4): {0: 100000},
("Category Three Distinct Dice", 2, 5): {0: 100000},
("Category Three Distinct Dice", 2, 6): {0: 100000},
("Category Three Distinct Dice", 2, 7): {0: 100000},
("Category Three Distinct Dice", 2, 8): {0: 100000},
("Category Three Distinct Dice", 3, 1): {0: 44707, 20: 55293},
("Category Three Distinct Dice", 3, 2): {0: 15078, 20: 84922},
("Category Three Distinct Dice", 3, 3): {0: 5056, 20: 94944},
("Category Three Distinct Dice", 3, 4): {0: 1688, 20: 98312},
("Category Three Distinct Dice", 3, 5): {0: 516, 20: 99484},
("Category Three Distinct Dice", 3, 6): {0: 182, 20: 99818},
("Category Three Distinct Dice", 3, 7): {0: 56, 20: 99944},
("Category Three Distinct Dice", 3, 8): {0: 15, 20: 99985},
("Category Three Distinct Dice", 4, 1): {0: 16721, 20: 83279},
("Category Three Distinct Dice", 4, 2): {0: 1826, 20: 98174},
("Category Three Distinct Dice", 4, 3): {0: 203, 20: 99797},
("Category Three Distinct Dice", 4, 4): {0: 18, 20: 99982},
("Category Three Distinct Dice", 4, 5): {0: 3, 20: 99997},
("Category Three Distinct Dice", 4, 6): {20: 100000},
("Category Three Distinct Dice", 4, 7): {20: 100000},
("Category Three Distinct Dice", 4, 8): {20: 100000},
("Category Three Distinct Dice", 5, 1): {0: 5904, 20: 94096},
("Category Three Distinct Dice", 5, 2): {0: 236, 20: 99764},
("Category Three Distinct Dice", 5, 3): {0: 12, 20: 99988},
("Category Three Distinct Dice", 5, 4): {20: 100000},
("Category Three Distinct Dice", 5, 5): {20: 100000},
("Category Three Distinct Dice", 5, 6): {20: 100000},
("Category Three Distinct Dice", 5, 7): {20: 100000},
("Category Three Distinct Dice", 5, 8): {20: 100000},
("Category Three Distinct Dice", 6, 1): {0: 1992, 20: 98008},
("Category Three Distinct Dice", 6, 2): {0: 21, 20: 99979},
("Category Three Distinct Dice", 6, 3): {20: 100000},
("Category Three Distinct Dice", 6, 4): {20: 100000},
("Category Three Distinct Dice", 6, 5): {20: 100000},
("Category Three Distinct Dice", 6, 6): {20: 100000},
("Category Three Distinct Dice", 6, 7): {20: 100000},
("Category Three Distinct Dice", 6, 8): {20: 100000},
("Category Three Distinct Dice", 7, 1): {0: 692, 20: 99308},
("Category Three Distinct Dice", 7, 2): {0: 4, 20: 99996},
("Category Three Distinct Dice", 7, 3): {20: 100000},
("Category Three Distinct Dice", 7, 4): {20: 100000},
("Category Three Distinct Dice", 7, 5): {20: 100000},
("Category Three Distinct Dice", 7, 6): {20: 100000},
("Category Three Distinct Dice", 7, 7): {20: 100000},
("Category Three Distinct Dice", 7, 8): {20: 100000},
("Category Three Distinct Dice", 8, 1): {0: 243, 20: 99757},
("Category Three Distinct Dice", 8, 2): {0: 1, 20: 99999},
("Category Three Distinct Dice", 8, 3): {20: 100000},
("Category Three Distinct Dice", 8, 4): {20: 100000},
("Category Three Distinct Dice", 8, 5): {20: 100000},
("Category Three Distinct Dice", 8, 6): {20: 100000},
("Category Three Distinct Dice", 8, 7): {20: 100000},
("Category Three Distinct Dice", 8, 8): {20: 100000},
("Category Two Pair", 1, 1): {0: 100000},
("Category Two Pair", 1, 2): {0: 100000},
("Category Two Pair", 1, 3): {0: 100000},
("Category Two Pair", 1, 4): {0: 100000},
("Category Two Pair", 1, 5): {0: 100000},
("Category Two Pair", 1, 6): {0: 100000},
("Category Two Pair", 1, 7): {0: 100000},
("Category Two Pair", 1, 8): {0: 100000},
("Category Two Pair", 2, 1): {0: 100000},
("Category Two Pair", 2, 2): {0: 100000},
("Category Two Pair", 2, 3): {0: 100000},
("Category Two Pair", 2, 4): {0: 100000},
("Category Two Pair", 2, 5): {0: 100000},
("Category Two Pair", 2, 6): {0: 100000},
("Category Two Pair", 2, 7): {0: 100000},
("Category Two Pair", 2, 8): {0: 100000},
("Category Two Pair", 3, 1): {0: 100000},
("Category Two Pair", 3, 2): {0: 100000},
("Category Two Pair", 3, 3): {0: 100000},
("Category Two Pair", 3, 4): {0: 100000},
("Category Two Pair", 3, 5): {0: 100000},
("Category Two Pair", 3, 6): {0: 100000},
("Category Two Pair", 3, 7): {0: 100000},
("Category Two Pair", 3, 8): {0: 100000},
("Category Two Pair", 4, 1): {0: 93065, 30: 6935},
("Category Two Pair", 4, 2): {0: 82102, 30: 17898},
("Category Two Pair", 4, 3): {0: 71209, 30: 28791},
("Category Two Pair", 4, 4): {0: 61609, 30: 38391},
("Category Two Pair", 4, 5): {0: 53036, 30: 46964},
("Category Two Pair", 4, 6): {0: 45705, 30: 54295},
("Category Two Pair", 4, 7): {0: 39398, 30: 60602},
("Category Two Pair", 4, 8): {0: 33673, 30: 66327},
("Category Two Pair", 5, 1): {0: 72847, 30: 27153},
("Category Two Pair", 5, 2): {0: 46759, 30: 53241},
("Category Two Pair", 5, 3): {0: 29462, 30: 70538},
("Category Two Pair", 5, 4): {0: 18351, 30: 81649},
("Category Two Pair", 5, 5): {0: 11793, 30: 88207},
("Category Two Pair", 5, 6): {0: 7385, 30: 92615},
("Category Two Pair", 5, 7): {0: 4610, 30: 95390},
("Category Two Pair", 5, 8): {0: 2938, 30: 97062},
("Category Two Pair", 6, 1): {0: 44431, 30: 55569},
("Category Two Pair", 6, 2): {0: 17183, 30: 82817},
("Category Two Pair", 6, 3): {0: 6759, 30: 93241},
("Category Two Pair", 6, 4): {0: 2562, 30: 97438},
("Category Two Pair", 6, 5): {0: 948, 30: 99052},
("Category Two Pair", 6, 6): {0: 375, 30: 99625},
("Category Two Pair", 6, 7): {0: 138, 30: 99862},
("Category Two Pair", 6, 8): {0: 57, 30: 99943},
("Category Two Pair", 7, 1): {0: 19888, 30: 80112},
("Category Two Pair", 7, 2): {0: 3935, 30: 96065},
("Category Two Pair", 7, 3): {0: 801, 30: 99199},
("Category Two Pair", 7, 4): {0: 175, 30: 99825},
("Category Two Pair", 7, 5): {0: 31, 30: 99969},
("Category Two Pair", 7, 6): {0: 7, 30: 99993},
("Category Two Pair", 7, 7): {0: 2, 30: 99998},
("Category Two Pair", 7, 8): {30: 100000},
("Category Two Pair", 8, 1): {0: 6791, 30: 93209},
("Category Two Pair", 8, 2): {0: 588, 30: 99412},
("Category Two Pair", 8, 3): {0: 61, 30: 99939},
("Category Two Pair", 8, 4): {0: 6, 30: 99994},
("Category Two Pair", 8, 5): {30: 100000},
("Category Two Pair", 8, 6): {30: 100000},
("Category Two Pair", 8, 7): {30: 100000},
("Category Two Pair", 8, 8): {30: 100000},
("Category 2-1-2 Consecutive", 1, 1): {0: 100000},
("Category 2-1-2 Consecutive", 1, 2): {0: 100000},
("Category 2-1-2 Consecutive", 1, 3): {0: 100000},
("Category 2-1-2 Consecutive", 1, 4): {0: 100000},
("Category 2-1-2 Consecutive", 1, 5): {0: 100000},
("Category 2-1-2 Consecutive", 1, 6): {0: 100000},
("Category 2-1-2 Consecutive", 1, 7): {0: 100000},
("Category 2-1-2 Consecutive", 1, 8): {0: 100000},
("Category 2-1-2 Consecutive", 2, 1): {0: 100000},
("Category 2-1-2 Consecutive", 2, 2): {0: 100000},
("Category 2-1-2 Consecutive", 2, 3): {0: 100000},
("Category 2-1-2 Consecutive", 2, 4): {0: 100000},
("Category 2-1-2 Consecutive", 2, 5): {0: 100000},
("Category 2-1-2 Consecutive", 2, 6): {0: 100000},
("Category 2-1-2 Consecutive", 2, 7): {0: 100000},
("Category 2-1-2 Consecutive", 2, 8): {0: 100000},
("Category 2-1-2 Consecutive", 3, 1): {0: 100000},
("Category 2-1-2 Consecutive", 3, 2): {0: 100000},
("Category 2-1-2 Consecutive", 3, 3): {0: 100000},
("Category 2-1-2 Consecutive", 3, 4): {0: 100000},
("Category 2-1-2 Consecutive", 3, 5): {0: 100000},
("Category 2-1-2 Consecutive", 3, 6): {0: 100000},
("Category 2-1-2 Consecutive", 3, 7): {0: 100000},
("Category 2-1-2 Consecutive", 3, 8): {0: 100000},
("Category 2-1-2 Consecutive", 4, 1): {0: 100000},
("Category 2-1-2 Consecutive", 4, 2): {0: 100000},
("Category 2-1-2 Consecutive", 4, 3): {0: 100000},
("Category 2-1-2 Consecutive", 4, 4): {0: 100000},
("Category 2-1-2 Consecutive", 4, 5): {0: 100000},
("Category 2-1-2 Consecutive", 4, 6): {0: 100000},
("Category 2-1-2 Consecutive", 4, 7): {0: 100000},
("Category 2-1-2 Consecutive", 4, 8): {0: 100000},
("Category 2-1-2 Consecutive", 5, 1): {0: 98403, 40: 1597},
("Category 2-1-2 Consecutive", 5, 2): {0: 90651, 40: 9349},
("Category 2-1-2 Consecutive", 5, 3): {0: 80100, 40: 19900},
("Category 2-1-2 Consecutive", 5, 4): {0: 69131, 40: 30869},
("Category 2-1-2 Consecutive", 5, 5): {0: 58252, 40: 41748},
("Category 2-1-2 Consecutive", 5, 6): {0: 49405, 40: 50595},
("Category 2-1-2 Consecutive", 5, 7): {0: 41585, 40: 58415},
("Category 2-1-2 Consecutive", 5, 8): {0: 34952, 40: 65048},
("Category 2-1-2 Consecutive", 6, 1): {0: 93465, 40: 6535},
("Category 2-1-2 Consecutive", 6, 2): {0: 73416, 40: 26584},
("Category 2-1-2 Consecutive", 6, 3): {0: 54041, 40: 45959},
("Category 2-1-2 Consecutive", 6, 4): {0: 38535, 40: 61465},
("Category 2-1-2 Consecutive", 6, 5): {0: 27366, 40: 72634},
("Category 2-1-2 Consecutive", 6, 6): {0: 18924, 40: 81076},
("Category 2-1-2 Consecutive", 6, 7): {0: 13387, 40: 86613},
("Category 2-1-2 Consecutive", 6, 8): {0: 9134, 40: 90866},
("Category 2-1-2 Consecutive", 7, 1): {0: 84168, 40: 15832},
("Category 2-1-2 Consecutive", 7, 2): {0: 52659, 40: 47341},
("Category 2-1-2 Consecutive", 7, 3): {0: 30435, 40: 69565},
("Category 2-1-2 Consecutive", 7, 4): {0: 17477, 40: 82523},
("Category 2-1-2 Consecutive", 7, 5): {0: 9782, 40: 90218},
("Category 2-1-2 Consecutive", 7, 6): {0: 5316, 40: 94684},
("Category 2-1-2 Consecutive", 7, 7): {0: 2995, 40: 97005},
("Category 2-1-2 Consecutive", 7, 8): {0: 1689, 40: 98311},
("Category 2-1-2 Consecutive", 8, 1): {0: 71089, 40: 28911},
("Category 2-1-2 Consecutive", 8, 2): {0: 33784, 40: 66216},
("Category 2-1-2 Consecutive", 8, 3): {0: 14820, 40: 85180},
("Category 2-1-2 Consecutive", 8, 4): {0: 6265, 40: 93735},
("Category 2-1-2 Consecutive", 8, 5): {0: 2600, 40: 97400},
("Category 2-1-2 Consecutive", 8, 6): {0: 1155, 40: 98845},
("Category 2-1-2 Consecutive", 8, 7): {0: 487, 40: 99513},
("Category 2-1-2 Consecutive", 8, 8): {0: 190, 40: 99810},
("Category Five Distinct Dice", 1, 1): {0: 100000},
("Category Five Distinct Dice", 1, 2): {0: 100000},
("Category Five Distinct Dice", 1, 3): {0: 100000},
("Category Five Distinct Dice", 1, 4): {0: 100000},
("Category Five Distinct Dice", 1, 5): {0: 100000},
("Category Five Distinct Dice", 1, 6): {0: 100000},
("Category Five Distinct Dice", 1, 7): {0: 100000},
("Category Five Distinct Dice", 1, 8): {0: 100000},
("Category Five Distinct Dice", 2, 1): {0: 100000},
("Category Five Distinct Dice", 2, 2): {0: 100000},
("Category Five Distinct Dice", 2, 3): {0: 100000},
("Category Five Distinct Dice", 2, 4): {0: 100000},
("Category Five Distinct Dice", 2, 5): {0: 100000},
("Category Five Distinct Dice", 2, 6): {0: 100000},
("Category Five Distinct Dice", 2, 7): {0: 100000},
("Category Five Distinct Dice", 2, 8): {0: 100000},
("Category Five Distinct Dice", 3, 1): {0: 100000},
("Category Five Distinct Dice", 3, 2): {0: 100000},
("Category Five Distinct Dice", 3, 3): {0: 100000},
("Category Five Distinct Dice", 3, 4): {0: 100000},
("Category Five Distinct Dice", 3, 5): {0: 100000},
("Category Five Distinct Dice", 3, 6): {0: 100000},
("Category Five Distinct Dice", 3, 7): {0: 100000},
("Category Five Distinct Dice", 3, 8): {0: 100000},
("Category Five Distinct Dice", 4, 1): {0: 100000},
("Category Five Distinct Dice", 4, 2): {0: 100000},
("Category Five Distinct Dice", 4, 3): {0: 100000},
("Category Five Distinct Dice", 4, 4): {0: 100000},
("Category Five Distinct Dice", 4, 5): {0: 100000},
("Category Five Distinct Dice", 4, 6): {0: 100000},
("Category Five Distinct Dice", 4, 7): {0: 100000},
("Category Five Distinct Dice", 4, 8): {0: 100000},
("Category Five Distinct Dice", 5, 1): {0: 90907, 25: 9093},
("Category Five Distinct Dice", 5, 2): {0: 68020, 25: 31980},
("Category Five Distinct Dice", 5, 3): {0: 47692, 25: 52308},
("Category Five Distinct Dice", 5, 4): {0: 32383, 25: 67617},
("Category Five Distinct Dice", 5, 5): {0: 21631, 25: 78369},
("Category Five Distinct Dice", 5, 6): {0: 14366, 25: 85634},
("Category Five Distinct Dice", 5, 7): {0: 9568, 25: 90432},
("Category Five Distinct Dice", 5, 8): {0: 6360, 25: 93640},
("Category Five Distinct Dice", 6, 1): {0: 75051, 25: 24949},
("Category Five Distinct Dice", 6, 2): {0: 38409, 25: 61591},
("Category Five Distinct Dice", 6, 3): {0: 17505, 25: 82495},
("Category Five Distinct Dice", 6, 4): {0: 7862, 25: 92138},
("Category Five Distinct Dice", 6, 5): {0: 3538, 25: 96462},
("Category Five Distinct Dice", 6, 6): {0: 1645, 25: 98355},
("Category Five Distinct Dice", 6, 7): {0: 714, 25: 99286},
("Category Five Distinct Dice", 6, 8): {0: 341, 25: 99659},
("Category Five Distinct Dice", 7, 1): {0: 58588, 25: 41412},
("Category Five Distinct Dice", 7, 2): {0: 19487, 25: 80513},
("Category Five Distinct Dice", 7, 3): {0: 6043, 25: 93957},
("Category Five Distinct Dice", 7, 4): {0: 1799, 25: 98201},
("Category Five Distinct Dice", 7, 5): {0: 544, 25: 99456},
("Category Five Distinct Dice", 7, 6): {0: 169, 25: 99831},
("Category Five Distinct Dice", 7, 7): {0: 59, 25: 99941},
("Category Five Distinct Dice", 7, 8): {0: 11, 25: 99989},
("Category Five Distinct Dice", 8, 1): {0: 43586, 25: 56414},
("Category Five Distinct Dice", 8, 2): {0: 9615, 25: 90385},
("Category Five Distinct Dice", 8, 3): {0: 1944, 25: 98056},
("Category Five Distinct Dice", 8, 4): {0: 383, 25: 99617},
("Category Five Distinct Dice", 8, 5): {0: 77, 25: 99923},
("Category Five Distinct Dice", 8, 6): {0: 18, 25: 99982},
("Category Five Distinct Dice", 8, 7): {0: 3, 25: 99997},
("Category Five Distinct Dice", 8, 8): {0: 2, 25: 99998},
("Category 4&5 Full House", 1, 1): {0: 100000},
("Category 4&5 Full House", 1, 2): {0: 100000},
("Category 4&5 Full House", 1, 3): {0: 100000},
("Category 4&5 Full House", 1, 4): {0: 100000},
("Category 4&5 Full House", 1, 5): {0: 100000},
("Category 4&5 Full House", 1, 6): {0: 100000},
("Category 4&5 Full House", 1, 7): {0: 100000},
("Category 4&5 Full House", 1, 8): {0: 100000},
("Category 4&5 Full House", 2, 1): {0: 100000},
("Category 4&5 Full House", 2, 2): {0: 100000},
("Category 4&5 Full House", 2, 3): {0: 100000},
("Category 4&5 Full House", 2, 4): {0: 100000},
("Category 4&5 Full House", 2, 5): {0: 100000},
("Category 4&5 Full House", 2, 6): {0: 100000},
("Category 4&5 Full House", 2, 7): {0: 100000},
("Category 4&5 Full House", 2, 8): {0: 100000},
("Category 4&5 Full House", 3, 1): {0: 100000},
("Category 4&5 Full House", 3, 2): {0: 100000},
("Category 4&5 Full House", 3, 3): {0: 100000},
("Category 4&5 Full House", 3, 4): {0: 100000},
("Category 4&5 Full House", 3, 5): {0: 100000},
("Category 4&5 Full House", 3, 6): {0: 100000},
("Category 4&5 Full House", 3, 7): {0: 100000},
("Category 4&5 Full House", 3, 8): {0: 100000},
("Category 4&5 Full House", 4, 1): {0: 100000},
("Category 4&5 Full House", 4, 2): {0: 100000},
("Category 4&5 Full House", 4, 3): {0: 100000},
("Category 4&5 Full House", 4, 4): {0: 100000},
("Category 4&5 Full House", 4, 5): {0: 100000},
("Category 4&5 Full House", 4, 6): {0: 100000},
("Category 4&5 Full House", 4, 7): {0: 100000},
("Category 4&5 Full House", 4, 8): {0: 100000},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category 4&5 Full House", 5, 1): {0: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category 4&5 Full House", 5, 2): {0: 96607, 50: 3393},
("Category 4&5 Full House", 5, 3): {0: 88788, 50: 11212},
("Category 4&5 Full House", 5, 4): {0: 77799, 50: 22201},
("Category 4&5 Full House", 5, 5): {0: 65797, 50: 34203},
("Category 4&5 Full House", 5, 6): {0: 54548, 50: 45452},
("Category 4&5 Full House", 5, 7): {0: 44898, 50: 55102},
("Category 4&5 Full House", 5, 8): {0: 36881, 50: 63119},
Yacht Dice: logic fix and several other fixes (#3878) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions * Yacht Dice: setup: change release-link to latest On the installation page, link to the latest release, instead of the page with all releases * Several fixes and changes -change apworld version -Removed the extra roll (this was not intended) -change extra_points_added to a mutable list to that it actually does something -removed variables multipliers_added and items_added -Rules, don't order by quantity, just by mean_score -Changed the weights in general to make it faster * :dog: * Revert setup to what it was (latest, without S) * remove temp weights file, shouldn't be here * Made sure that there is not too many step score multipliers. Too many step score multipliers lead to gen fails too, probably because you need many categories for them to actually help a lot. So it's hard to use them at the start of the game. --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-09-06 20:50:57 +00:00
("Category 4&5 Full House", 6, 1): {0: 100000},
YachtDice: implement new game (#3482) * Add the yacht dice (from other git) world to the yacht dice fork * Update .gitignore * Removed zillion because it doesn't work * Update .gitignore * added zillion again... * Now you can have 0 extra fragments * Added alt categories, also options * Added item categories * Extra categories are now working! :dog: * changed options and added exceptions * Testing if I change the generate.py * Revert "Testing if I change the generate.py" This reverts commit 7c2b3df6170dcf8d8f36a1de9fcbc9dccdec81f8. * ignore gitignore * Delete .gitignore * Update .gitignore * Update .gitignore * Update logic, added multiplicative categories * Changed difficulties * Update offline mode so that it works again * Adjusted difficulty * New version of the apworld, with 1000 as final score, always Will still need to check difficulty and weights of adding items. Website is not ready yet, so this version is not usable yet :) * Changed yaml and small bug fixes Fix when goal and max are same Options: changed chance to weight * no changes, just whitespaces * changed how logic works Now you put an array of mults and the cpu gets a couple of tries * Changed logic, tweaked a bit too * Preparation for 2.0 * logic tweak * Logic for alt categories properly now * Update setup_en.md * Update en_YachtDice.md * Improve performance of add_distributions * Formatting style * restore gitignore to APMW * Tweaked generation parameters and methods * Version 2.0.3 manual input option max score in logic always 2.0.3 faster gen * Comments and editing * Renamed setup guide * Improved create_items code * init of locations: remove self.event line * Moved setting early items to generate_early * Add my name to CODEOWNERS * Added Yacht Dice to the readme in list of games * Improve performance of Yacht Dice * newline * Improve typing * This is actually just slower lol * Update worlds/yachtdice/Items.py Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com> * Update Options.py * Styling * finished text whichstory option * removed roll and rollfragments; not used * import; worlds not world :) * Option groups! * ruff styling, fix * ruff format styling! * styling and capitalization of options * small comment * Cleaned up the "state_is_a_list" a little bit * RUFF :dog: * Changed filling the itempool for efficiency Now, we start with 17 extra items in the item pool, it's quite likely you need at least 17 items (~80%?). And then afterwards, we delete items if we overshoot the target of 1000, and add items if we haven't reached an achievable score of 1000 yet. Also, no need to recompute the entire logic when adding points. * :dog: * Removed plando "fix" * Changed indent of score multiplier * faster location function * Comments to docstrings * fixed making location closest to goal_score be goal_score * options format * iterate keys and values of a dict together * small optimization ListState * faster collection of categories * return arguments instead of making a list (will :dog: later) * Instead of turning it into a tuple, you can just make a tuple literal * remove .keys() * change .random and used enumerate * some readability improvements * Remove location "0", we don't use that one * Remove lookup_id_to_name entirely I for sure don't use it, and as far as I know it's not one of the mandatory functions for AP, these are item_name_to_id and location_name_to_id. * .append instead of += for single items, percentile function changed Also an extra comment for location ids. * remove ) too many * Removed sorted from category list * Hash categories (which makes it slower :( ) Maybe I messed up or misunderstood... I'll revert this right away since it is 2x slower, probably because of sorted instead of sort? * Revert "Hash categories (which makes it slower :( )" This reverts commit 34f2c1aed8c8813b2d9c58896650b82a810d3578. * temporary push: 40% faster generation test Small changes in logic make the generation 40% faster. I'll have to think about how big the changes are. I suspect they are rather limited. If this is the way to go, I'll remove the temp file and redo the YachtWeights file, I'll remove the functions there and just put the new weights here. * Add Points item category * Reverse changes of bad idea :) * ruff :dog: * Use numpy and pmf function to speed up gen Numpy has a built-in way to sum probability mass functions (pmf). This shaves of 60% of the generation time :D * Revert "Use numpy and pmf function to speed up gen" This reverts commit 9290191cb323ae92321d6c2cfcfe8c27370f439b. * Step inbetween to change the weights * Changed the weights to make it faster 135 -> 81 seconds on 100 random yamls * Adjusted max_dist, split dice_simulation function * Removed nonlocal and pass arguments instead * Change "weight-lists" to Dict[str, float] * Removed the return from ini_locations. Also added explanations to cat_weights * Choice options; dont'use .value (will ruff later) * Only put important options in slotdata * :dog: * Add Dict import * Split the cache per player, limit size to 400. * :dog: * added , because of style * Update apworld version to 2.0.6 2.0.5 is the apworld I released on github to be tested I never separately released 2.0.4. * Multiple smaller code improvements - changed names in YachtWeights so we don't need to translate them in Rules anymore - we now remember which categories are present in the game, and also put this in slotdata. This we do because only one of two categories is present in a game. If for some reason both are present (plando/getitem/startinventory), we now know which category to ignore - * :dog: ruff * Mostly minimize_extra_items improvements - Change logic, generation is now even faster (0.6s per default yaml). - Made the option 'minimize_extra_items' do a lot more, hopefully this makes the impact of Yacht Dice a little bit less, if you want that. Here's what is also does now: - you start with 2 dice and 2 rolls - there will be less locations/items at the start of you game * ruff :dog: * Removed printing options * Reworded some option descriptions --------- Co-authored-by: NewSoupVi <57900059+NewSoupVi@users.noreply.github.com> Co-authored-by: Exempt-Medic <60412657+Exempt-Medic@users.noreply.github.com>
2024-08-21 17:59:21 +00:00
("Category 4&5 Full House", 6, 2): {0: 88680, 50: 11320},
("Category 4&5 Full House", 6, 3): {0: 70215, 50: 29785},
("Category 4&5 Full House", 6, 4): {0: 50801, 50: 49199},
("Category 4&5 Full House", 6, 5): {0: 35756, 50: 64244},
("Category 4&5 Full House", 6, 6): {0: 24698, 50: 75302},
("Category 4&5 Full House", 6, 7): {0: 17145, 50: 82855},
("Category 4&5 Full House", 6, 8): {0: 11846, 50: 88154},
("Category 4&5 Full House", 7, 1): {0: 97090, 50: 2910},
("Category 4&5 Full House", 7, 2): {0: 77440, 50: 22560},
("Category 4&5 Full House", 7, 3): {0: 51372, 50: 48628},
("Category 4&5 Full House", 7, 4): {0: 30566, 50: 69434},
("Category 4&5 Full House", 7, 5): {0: 17866, 50: 82134},
("Category 4&5 Full House", 7, 6): {0: 10521, 50: 89479},
("Category 4&5 Full House", 7, 7): {0: 6204, 50: 93796},
("Category 4&5 Full House", 7, 8): {0: 3670, 50: 96330},
("Category 4&5 Full House", 8, 1): {0: 94172, 50: 5828},
("Category 4&5 Full House", 8, 2): {0: 64693, 50: 35307},
("Category 4&5 Full House", 8, 3): {0: 35293, 50: 64707},
("Category 4&5 Full House", 8, 4): {0: 17749, 50: 82251},
("Category 4&5 Full House", 8, 5): {0: 8740, 50: 91260},
("Category 4&5 Full House", 8, 6): {0: 4550, 50: 95450},
("Category 4&5 Full House", 8, 7): {0: 2218, 50: 97782},
("Category 4&5 Full House", 8, 8): {0: 1084, 50: 98916},
}