https://github.com/campiohe/breakthrough
SI202 : Resolução de Problemas I
https://github.com/campiohe/breakthrough
analysis chess-variant data-science reinforcement-learning
Last synced: 2 months ago
JSON representation
SI202 : Resolução de Problemas I
- Host: GitHub
- URL: https://github.com/campiohe/breakthrough
- Owner: campiohe
- License: mit
- Created: 2019-08-21T13:05:20.000Z (almost 7 years ago)
- Default Branch: master
- Last Pushed: 2019-12-01T18:05:06.000Z (over 6 years ago)
- Last Synced: 2025-12-26T19:42:55.996Z (6 months ago)
- Topics: analysis, chess-variant, data-science, reinforcement-learning
- Language: Jupyter Notebook
- Homepage:
- Size: 93.8 KB
- Stars: 2
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# py_breakthrough
SI202 : Resolução de Problemas I
## Trabalho 1 (individual)
Cada aluno será responsável por analisar o jogo de breakthrough e criar 10 estratégias para verificar se consegue vencer a partida contra um adversário que joga ao acaso. As estratégias podem ser simples e baseadas simplesmente em heurísticas, o seu objetivo é garantir que entende bem o que funciona e o que não funciona no jogo. Por exemplo, algumas estratégias podem ser:
1. Mover sempre o jogador da frente.
1. Mover sempre o jogador de trás.
1. Mover em blocos de jogadores.
1. Capturar sempre que possível.
O que você deverá fazer é implementar programas com alguma dessas estratégias e analisar esses programas. Você deverá implementar o seu programa em um kernel no jupyter e adicionar as suas análises no próprio kernel. Desse modo, ao compartilhar com o professor o seu kernel, ele poderá ver as suas análises e o seus códigos.
## Players (strategies.py)
* __dump_player__ : calculate the square values of the board, perform a move that tries to get a square with the highest value
* __evil_player__ : calculates the game score by the evaluation function, performing the move that gets the highest score in that state
* __forward_player__ : always move the piece that is in the rows with the highest index and try to capture whenever possible
* __mirror_player__ : whenever possible, try to make the same move as your opponent
* __team_player__ : moves a piece following the neighbors and capturing whenever possible
* __killer_palyer__ : calculates the shortest distance for an opponent in order to eliminate it
* __zigzag_player__ : moves only across the diagonals alternately, capturing whenever possible
* __dodge_player__ : checks if the destination square is occupied and defended, if not defended moves to it
* __conn_player__ : calculates the connectivity of the pieces by performing the movement that will assign a higher connectivity value
* __sup_player__ : check if the destination house is defended by allies, if it is, performes the move
### 100 match results
## Minimax
- [x] alpha-beta pruning
- [x] dynamic programming
- [ ] improve evaluation function