Connect4Zero

A Connect 4 game featuring two AI opponents with different strategies and difficulty levels

About • Features • AI Algorithms • Installation • Usage • Project Structure

  

---

About

This project implements the classic Connect 4 game in Python with Pygame, featuring two AI opponents of varying difficulty. The game showcases fundamental game theory concepts including the Minimax algorithm with position evaluation heuristics.

Core Concept

Optimal Play = Minimax(game_tree, depth) + Evaluate(board_position) + Heuristics(threats)

The Hard AI uses a depth-limited Minimax search combined with strategic position evaluation to find optimal moves, while the Easy AI provides a more casual experience with randomized play.

---

Features

Game Features

Feature	Description
Two AI Modes	Easy (random) and Hard (Minimax) opponents
Visual Effects	Gradient backgrounds, 3D token shading
Sound System	Audio feedback for moves, wins, and menu
Responsive UI	Clean menus with keyboard controls

Controls

Key	Action
1	Select Easy AI / Play again
2	Select Hard AI / Play again
ESC	Quit game
Mouse	Drop token in column

---

AI Algorithms

Easy AI - Random Strategy

The Easy AI implements a reactive strategy with three priority levels:

def get_move(board, player):
    1. If winning move exists     → Play it
    2. If opponent can win next   → Block it
    3. Otherwise                  → Random column

Characteristics:

• Predictable but not trivial
• Good for beginners
• Fast computation

Hard AI - Minimax Algorithm

The Hard AI uses a sophisticated Minimax search with evaluation heuristics:

def minimax(node, depth, maximizing_player):
    if depth == 0 or terminal_state:
        return evaluate(position)

    if maximizing:
        return max(minimax(child) for child in children)
    else:
        return min(minimax(child) for child in children)

Evaluation Function Components:

Component	Weight	Description
Win Detection	±1,000,000	Terminal state detection
Threat Analysis	5-10x	Consecutive tokens in all directions
Position Value	80x	Center positions score higher

Position Value Matrix:

[ 3,  4,  5,  7,  5,  4,  3]
[ 4,  6,  8, 10,  8,  6,  4]
[ 5,  8, 11, 13, 11,  8,  5]
[ 5,  8, 11, 13, 11,  8,  5]
[ 4,  6,  8, 10,  8,  6,  4]
[ 3,  4,  5,  7,  5,  4,  3]

Characteristics:

• Search depth: 4 moves ahead
• Opening optimization: Always plays center
• Very challenging to beat

---

Installation

Prerequisites

• Python 3.7+
• NumPy
• Pygame

Setup

# Clone the repository
git clone https://github.com/mathisdelsart/Connect4Zero.git
cd Connect4Zero

# Install dependencies
pip install pygame numpy

# Run the game
python main.py

---

Usage

Playing the Game

1. Launch with python main.py
2. Select difficulty: 1 for Easy, 2 for Hard
3. Click columns to drop your red tokens
4. Connect 4 horizontally, vertically, or diagonally to win

Monte Carlo Simulation

Compare AI performance with statistical analysis:

python montecarlo.py

Runs tournaments between Hard AI and Easy AI, generating win rate statistics and visualization graphs.

---

Project Structure

Connect4_AI/
├── main.py                 # Entry point
├── montecarlo.py           # AI performance benchmarking
├── src/
│   ├── configs.py          # Game settings, colors, fonts, sounds
│   ├── functions.py        # Core logic: board ops, win detection, minimax
│   ├── game.py             # Game loop, rendering, event handling
│   └── players/
│       ├── player.py       # Human player (console mode)
│       ├── easy_ai.py      # Easy AI - random with basic strategy
│       └── hard_ai.py      # Hard AI - minimax algorithm
└── assets/
    ├── sounds/             # Audio files (drop, win, lose, select)
    └── *.png               # Game images

---

Game Rules

1. Players alternate turns dropping tokens into columns
2. Tokens fall to the lowest available row
3. First to connect 4 tokens in a row wins
4. Connections: horizontal, vertical, or diagonal
5. Draw if all 42 cells are filled

---

Author

Mathis DELSART