Tic-Tac-Toe with MCTS

A simple implementation of Tic-Tac-Toe using Monte Carlo Tree Search (MCTS) for AI moves. Includes two AI strategies: epsilon-greedy and UCB-based.

Files

MCTS.py

Implements the MCTS algorithm for AI decision-making. Provides:

• MCTS_epsilon_greedy: Balances exploration/exploitation with an epsilon parameter.
• MCTS_UCB: Uses UCB1 for action selection, balancing exploration and exploitation.

TicTacToe.py

Defines the Tic-Tac-Toe game logic:

• Board representation and state management.
• Move validation and state transitions.
• Win/loss/draw detection.

play.py

Allows playing Tic-Tac-Toe against the AI:

• Choose 'X' or 'O' to play as.
• AI uses either MCTS_epsilon_greedy or MCTS_UCB.

Parameters

Adjustable in play.py:

• USE_UCB: Toggle AI strategy (True for UCB, False for epsilon-greedy).
• MAX_ITERS: Number of MCTS iterations per move.
• EPSILON: Exploration rate for epsilon-greedy.
• C: UCB exploration-exploitation balance.
• SEED: Random seed for consistency.

Example Gameplay

Welcome to Tic-Tac-Toe!
Do you want to play as 'X' or 'O'? (X goes first): X
You are playing as 'X'. You go first!
. . .
. . .
. . .
Your turn!
Enter your move (1-9): 1
X . .
. . .
. . .
AI is thinking...
AI chose position 5.
X . .
. O .
. . .
Your turn!
Enter your move (1-9): 2
X X .
. O .
. . .
AI is thinking...
AI chose position 3.
X X O
. O .
. . .
Your turn!
Enter your move (1-9): 4
X X O
X O .
. . .
AI is thinking...
AI chose position 7.
X X O
X O .
O . .
AI wins!