Connect4 and TicTacToe

This project implements Connect4 and TicTacToe games with various player types, including human, random, algorithmic, minimax, and Q-learning agents.

Requirements

• Python 3.12 or later
• Required Python packages: readchar, tqdm

You can install the required packages using pip:

pip install readchar tqdm

Running the Games

Connect4

To run the Connect4 game, use the following command:

python connect4.py [options]

TicTacToe

To run the TicTacToe game, use the following command:

python tictactoe.py [options]

Options

• -p1 <player1>: Set the type of player 1. Default is minimax.
• -p2 <player2>: Set the type of player 2. Default is algo.
• -g <number of games>: Set the number of games to play. Default is 1.
• -d <max depth>: Set the maximum depth for the minimax algorithm. Default is the size of the board.
• -v: Enable visual mode.
• -w <width>: Set the width of the board (Connect4 only). Must be between 4 and 9. Default is 7.
• -h <height>: Set the height of the board (Connect4 only). Must be between 4 and 9. Default is 6.
• -train <batches>: Train Q-learning agents for the specified number of batches.
• -b <batch size>: Set the batch size for training. Default is 50000.
• -s <seed>: Set the random seed for training. Default is a random integer.
• -o <order>: Set the training order (first, second, or both). Default is both.

Player Types

• human: Human player.
• random: Player that chooses moves randomly.
• algo: Player that uses a custom algorithm to choose moves.
• minimax: Player that uses the minimax algorithm to choose moves.
• minimax_ab: Player that uses the minimax algorithm with alpha-beta pruning.
• qlearn: Player that uses Q-learning to choose moves.

Examples

Play a Connect4 Game

To play a game between a human and a minimax player:

python connect4.py -p1 human -p2 minimax

Play a TicTacToe Game

To play a game between a human and a minimax player:

python tictactoe.py -p1 human -p2 minimax -v

Train Q-learning Agents for Connect4

To train Q-learning agents for 10 batches with a batch size of 50000:

python connect4.py -train 10 -b 50000

Train Q-learning Agents for TicTacToe

To train Q-learning agents for 10 batches with a batch size of 50000:

python tictactoe.py -train 10 -b 50000

Notes

• The game will display instructions and the board if a human player is involved.
• The Q-learning agents will save their Q-tables after training.
• The Connect4 board size can be customized using the -w and -h options, but must be between 4 and 9 for both dimensions.