AlphaViergewinnt

Reimplementation of DeepMind's AlphaGo Zero for the game Viergewinnt (WIP)

WORK IN PROGRESS

Current state

• Viergewinnt game logic
• Basic Monte-Carlo tree search
• Reinforcement learning model as MCTS strategy

Usage

$ python play_match.py --help
Usage: play_match.py [OPTIONS]

  Play a match

Options:
  --game [tictactoe|viergewinnt]  Game to be played  [required]
  -x [random|human|mcts]          Strategy for player X  [required]
  -o [random|human|mcts]          Strategy for player O  [required]
  --help                          Show this message and exit.

$ python play_match.py -x human -o mcts --game viergewinnt

MCTS visualization

Visualization of MCTS search tree with networkx