boardgame2

boardgame2 is an extension of OpenAI Gym that implements multiple two-player zero-sum 2-dimension board games, such as TicTacToe, Gomuko, and Reversi.

Environments

• Reversi-v0
• KInARow-v0, as well as Gomuku-v0 and TicTacToe-v0
• Go-v0 (Experimental, not fully implemented)

Install

pip install --upgrade boardgame2

We support Windows, macOS, Linux, and other operating systems.

Usage

See API docs for all classes and functions.

Create a Game

import gym
import boardgame2

env = gym.make('TicTacToe-v0') # 3x3, 3-in-a-row
env = gym.make('Gomuku-v0') # 15x15, 5-in-a-row
env = gym.make('KInARow-v0', board_shape=5, target_length=4) # 5x5, 4-in-a-row
env = gym.make('KInARow-v0', board_shape=(3, 5), target_length=4) # 3x5, 4-in-a-row
env = gym.make('Reversi-v0') # 8x8
env = gym.make('Reversi-v0', board_shape=6) # 6x6
env = gym.make('Go-v0') # 19x19
env = gym.make('Go-v0', board_shape=15) # 15x15

Play a Game

import gym
import boardgame2

env = gym.make('TicTacToe-v0')
print('observation space = {}'.format(env.observation_space))
print('action space = {}'.format(env.action_space))

observation = env.reset()
while True:
    action = env.action_space.sample()
    observation, reward, done, info = env.step(action)
    if done:
        break
env.close()

BibTeX

This package has been published in the following book:

@book{xiao2019,
 title     = {Reinforcement Learning: Theory and {Python} Implementation},
 author    = {Zhiqing Xiao}
 year      = 2019,
 month     = 8,
 publisher = {China Machine Press},
}