summaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
2019-06-26Implemented select move rulesJustin Worthe
2019-06-24New command types for select and bombsJustin Worthe
2019-05-26Update build instructions to match current processesJustin Worthe
2019-05-26Filesync on submission zipJustin Worthe
2019-05-26A bit faster through caching set of occupied spotsJustin Worthe
2019-05-26Setting up makefile for submissionJustin Worthe
2019-05-26Score based MCTSJustin Worthe
2019-05-25Implemented tracking of the tiebreaking scoreJustin Worthe
2019-05-22Added a strategy cache for reusing previous round dataJustin Worthe
2019-05-21More robust game logic and reasoningJustin Worthe
2019-05-20Extracted functionality for clearing out dead wormsJustin Worthe
2019-05-18Started breaking up state for easier unit testingJustin Worthe
2019-05-17Strategy to focus mctsJustin Worthe
2019-05-14Better performance for finding things to shoot atJustin Worthe
2019-05-14Filled in the rest of the MCTSJustin Worthe
Problem: The current random things isn't actually finding any victorious end states. This game easily meanders if it's played without purpose.
2019-05-13Building up unsimulated moves listJustin Worthe
2019-05-12Outline of MCTSJustin Worthe
2019-04-26Refactored game map to use less memoryJustin Worthe
2019-04-25Strategy that starts building exhaustive game state treeJustin Worthe
This falls over (and takes the host machine with it) because its memory usage grows catastrophically. The main use of time, reported by perf, was cloning the map vector.
2019-04-25Made the state object avoid any heap allocationsJustin Worthe
2019-04-25Implement move and dig commandsJustin Worthe
2019-04-25Test that match replay matches my simulationJustin Worthe
2019-04-25Arranged the types to start writing the replay based property testJustin Worthe
2019-04-25Call update vs create after the first moveJustin Worthe
2019-04-22More minimal game stateJustin Worthe
I'd prefer to start with just the state that I need, and progressively readd the bits that I've skipped as I find I need them, or as the competition evolves.
2019-04-22Structures representing game stateJustin Worthe
2019-04-22Start the project from the starter botJustin Worthe
2018-12-08Renamed variable to be more conciseJustin Worthe
2018-11-15More concise feature flags for variablesJustin Worthe
2018-09-19Added debug flags into release buildJustin Worthe
I had these turned off for the final submission, but it's necessary for showing profiling.
2018-09-17Added readme for completed botJustin Worthe
I also had to remove the readme that had been hanging around from the sample bot.
2018-09-08Update compile flags for submissionJustin Worthe
2018-09-08Moved metric calculations to reuse per rowJustin Worthe
2018-09-08Optimized away unnecessary sumJustin Worthe
2018-09-08Cached more values in tree exploration calcsJustin Worthe
2018-09-08Feature-flaggified weighting win ratiosJustin Worthe
2018-09-08Cleaned up compile warningsJustin Worthe
2018-09-08Added compile as a default build targetJustin Worthe
2018-09-08Added minimum to weighted victory / defeat scoreJustin Worthe
2018-09-08Added a static set of opening moves to get energy up and runningJustin Worthe
2018-09-08Put selection of full monte carlo tree behind a feature flagJustin Worthe
2018-09-06Added benchmarking for number of explored nodesJustin Worthe
2018-09-06Filled in the remaining TODOs on the tree searchJustin Worthe
2018-09-06Disabled detailed debuggingJustin Worthe
2018-09-05Filled in high level outline of monte carlo treeJustin Worthe
2018-09-05Data structures for proper monte carlo searchJustin Worthe
2018-09-05Tweaking weighting in heuristic searchJustin Worthe
2018-09-05Fixed config that hadn't been tested in a whileJustin Worthe
2018-09-05Added weighting of victory / defeat board positionsJustin Worthe
2018-09-02Improved performance of heuristic calcs by caching common calcsJustin Worthe