gym-tic-tac-toe

Open AI Gym environment for Tic-Tac-Toe.

Installation

git clone https://github.com/LudwigStumpp/gym-tic-tac-toe.git
cd gym-tic-tac-toe
pip install -e .

Usage

Make sure to have a look at the demo gym-tic-tac-toe/demo_q_learning.py where Q-Learning is used to train an agent. Or take a look at gym-tic-tac-toe/demo_random.py where two random opponents play against each other.

Initialization

import gym

env = gym.make('gym_tictactoe:tictactoe-v0')
env.reset()

env.render()
# | | | |
# | | | |
# | | | |

Make a move

The board is indexed as follows:

# |0|1|2|
# |3|4|5|
# |6|7|8|

To make a move, call the step - function:

(observation, reward, done, info) = env.step([0, 3]) # 0 for player 1 and position 3
# (27, 0, False, 'normal move')

env.render()
# | | | |
# |O| | |
# | | | |

env.step([1, 2]) # 1 for player 2 and position 2
env.render()
# | | |X|
# |O| | |
# | | | |

State representation

preset = [[0, 1, 2], [0, 0, 0], [1, 0, 2]]
env.s = env._encode(preset)
env.render()
# | |O|X|
# | | | |
# |O| |X|

print(env.s) 
# 13872

board = env._decode(env.s)
# [[0, 1, 2], [0, 0, 0], [1, 0, 2]]

Quick demo

From gym-tic-tac-toe/demo_random.py:

import gym
import itertools

def main():
    env = gym.make('gym_tictactoe:tictactoe-v0')

    observation = env.reset()
    while True:
        player = next_player()
        position = env.action_space.sample()[1]
        observation, reward, done, info = env.step((player, position))
        env.render()

        if done:
            print(info)
            break

next_player = itertools.cycle([0, 1]).__next__
if __name__ == '__main__':
    main()

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

Please make sure to update tests (gym-tic-tac-toe/tests.py) as appropriate.

License

MIT