reversi42 7.0.6

Reversi42: Fast AI Reversi

Reversi42

Reversi42: Fast AI Reversi

A tournament-grade Reversi implementation featuring high-performance bitboard-based AI engine, comprehensive opening book system, and modern web-based interface. Designed for both casual play and competitive AI research.

Copyright (C) 2011-2025 Luca Amore
Website: https://www.lucaamore.com

---

Overview

Reversi42 is a Reversi implementation that combines classical game AI techniques with modern software engineering practices. The engine utilizes bitboard representation for optimal performance, achieving 50-100x speedup over standard implementations through efficient 64-bit integer operations.

The system includes 12 pre-configured AI players with ELO ratings ranging from 1250 to 1880, a comprehensive opening book with 644 professional sequences, and a fully-featured tournament system for AI benchmarking and competition.

---

Key Features

Core Engine

• Bitboard-Based Architecture: 64-bit integer operations for O(1) board operations
• Advanced Search Algorithms: Alpha-beta pruning with transposition tables, null-move pruning, futility pruning, late move reduction (LMR), and multi-cut pruning
• Multiple Search Strategies: Fixed depth, iterative deepening, and adaptive depth search
• Parallel Processing: Multi-core support for parallel search execution
• Evaluation Functions: Four specialized evaluators (Mobility, Positional, Stability, Parity) with configurable weights

AI Players

• 12 Pre-configured Players: Ranging from beginner (ELO 1250) to expert level (ELO 1880)
• Configurable Difficulty: Each player features unique strategies and evaluation profiles
• YAML-Based Configuration: No-code AI player creation through declarative configuration files
• Custom Avatars: Support for custom player avatars (PNG/JPEG, 512x512 recommended)

Opening Book System

• 644 Professional Sequences: Comprehensive opening book derived from tournament play
• Trie-Based Lookup: O(m) complexity for efficient sequence matching
• Multiple Modes: Instant play and evaluated modes for flexible gameplay

User Interface

• Modern Web Interface: Browser-based UI built with FastAPI and WebSocket for real-time updates
• Game Management: Save and load functionality using XOT (eXtended Othello Transcript) format
• Real-time Analysis: Live move evaluation and game state visualization

Tournament System

• AI Competitions: Automated tournament mode for benchmarking and competition
• Statistical Analysis: Comprehensive statistics and performance metrics
• Custom Configurations: Flexible tournament setup through JSON configuration files

---

Installation

Requirements

• Python 3.9 or higher
• pip package manager

Install from PyPI

pip install reversi42

Install from Source

# Clone the repository
git clone https://github.com/lucaamore/reversi42.git
cd reversi42

# Install dependencies
pip install -r requirements.txt

# Or install in development mode
pip install -e .

---

Quick Start

Web Interface (Recommended)

Launch the web-based interface:

reversi42

The interface will be available at http://localhost:8000 in your web browser.

Python API

Use Reversi42 as a Python library:

from Reversi.BitboardGame import BitboardGame
from Players.PlayerFactory import PlayerFactory

# Initialize game
game = BitboardGame()

# Create AI player from YAML configuration
player = PlayerFactory.create_from_yaml("config/players/enabled/divzero.yaml")

# Play game programmatically
# ... game logic ...

Tournament Mode

Run AI competitions:

# Quick tournament match
python tournament/quick_tournament.py

# Custom tournament configuration
python tournament/tournament.py ring/config.json

---

AI Players

Reversi42 includes 12 pre-configured AI players with varying skill levels and strategies:

Player	ELO Rating	Strategy	Search Depth	Description
DIVZERO.EXE	1880	Adaptive	8/12/16	Maximum strength player with adaptive depth and parallel processing
The Oracle	1850	Endgame Focus	7/9/14	Specialized in endgame positions with parity evaluation
Apocalyptron	1850	Balanced	Adaptive	Standard strong AI with balanced evaluation
Fortress Eternal	1800	Defensive	10	Defensive specialist with stability focus
The Executioner	1770	Aggressive	9	Aggressive player with mobility emphasis
The Strangler	1750	Mobility Control	11	Focuses on restricting opponent mobility
Corner Reaper	1720	Positional	8	Positional player with corner control emphasis
Glitch Lord	1500±200	Chaotic	Variable	Unpredictable player with randomized behavior
Lightning Strike	1400	Speed	4	Fast-playing player for rapid games
Blitz Demon	1350	Rapid Fire	5	Ultra-fast player optimized for speed
Zen Master	1250	Balanced	3	Beginner-friendly balanced player

For detailed player profiles and configurations, see EPIC_GLADIATORS.md.

---

Custom AI Player Configuration

Create custom AI players using YAML configuration files without programming:

Quick Start

# Copy template
cp config/players/00_AI_CONFIG_TEMPLATE.yaml config/players/enabled/my_ai.yaml

# Edit configuration
vim config/players/enabled/my_ai.yaml

# Player is automatically discovered on startup
reversi42

Configuration Options

• Search Depth (4-16): Controls AI strength vs. speed tradeoff
• Search Strategy: fixed, iterative, or adaptive
• Evaluation Presets: balanced, aggressive, defensive, endgame_specialist
• Parallel Search: Enable multi-core processing
• Pruning Techniques: Configure null-move, futility, LMR, and multi-cut pruning
• Opening Book: Enable/disable and configure opening book usage
• Custom Avatars: Specify custom avatar images (PNG/JPEG)

Example Configurations

Speed-Optimized Player (ELO ~1400):

• Depth: 4-5
• Minimal pruning
• Average move time: <100ms

Tactical Player (ELO ~1750):

• Depth: 9
• Mobility weight: ×2.5
• Average move time: ~5s

Defensive Specialist (ELO ~1800):

• Depth: 10
• Stability weight: ×2.5
• Average move time: ~10s

Endgame Specialist (ELO ~1850):

• Adaptive depth: 7/9/14
• Parity weight: ×2.0
• Variable move time

For comprehensive configuration documentation, see CREATE_CUSTOM_PLAYER.md and AI_CONFIGURATION_SYSTEM.md.

---

Technical Architecture

Core Technologies

• Bitboard Representation: 64-bit integers for efficient board state manipulation
• Classical AI Algorithms: Minimax with alpha-beta pruning (no neural networks)
• Advanced Pruning: Null-move, futility, LMR, and multi-cut techniques
• Transposition Tables: Efficient position caching for improved performance
• Move Ordering: PV-move, killer moves, and history heuristic optimization

Performance Characteristics

• Speed: 50-100x faster than standard array-based implementations
• Search Depth: Configurable from 4 to 16 ply
• Parallel Processing: Multi-core support for parallel search
• Memory Efficiency: Optimized data structures for minimal memory footprint

Project Statistics

• Codebase: ~12,000 lines of Python code
• Test Coverage: 220+ unit and integration tests
• Documentation: 40+ documentation files
• AI Players: 12 pre-configured players
• Opening Sequences: 644 professional opening sequences

For detailed technical documentation, see:

• Apocalyptron Engine Architecture
• Bitboard Implementation
• System Architecture

---

Project Structure

Reversi42/
├── src/                      # Source code
│   ├── Reversi/              # Core game engine (BitboardGame, Game)
│   ├── AI/Apocalyptron/      # AI engine (search, evaluation, pruning, caching)
│   ├── Players/              # Player system and configurations
│   ├── webgui/               # FastAPI WebSocket server
│   ├── domain/               # Opening book system
│   └── ui/                   # UI implementations
├── config/                   # YAML configurations
│   └── players/enabled/      # AI player configurations
├── docs/                     # Comprehensive documentation
├── tests/                    # Test suite (220+ tests)
├── tournament/               # Tournament system
└── saves/                    # Saved games (XOT format)

---

Documentation

Comprehensive documentation is available in the docs/ directory:

• User Guide: Getting started, game rules, strategies, and FAQ
• API Reference: Complete API documentation for BitboardGame, Player, and View classes
• Architecture Documentation: System design, engine architecture, and technical details
• Tutorials: Step-by-step guides for creating custom AI players
• Tournament System: Tournament configuration and usage
• Development Guide: Setup, testing, and contribution guidelines

---

Testing

Run the comprehensive test suite:

# Full test suite
./scripts/run_tests.sh

# Quick unit tests
pytest tests/apocalyptron/unit/ -v

# With coverage
pytest --cov=src tests/

The project includes 220+ tests covering unit, integration, and characterization scenarios. See test documentation for details.

---

Game File Format

Games are saved in XOT (eXtended Othello Transcript) format, a human-readable format that stores complete move history, game metadata, and player information. Saved games are stored in the saves/ directory.

---

License

This project is licensed under the GNU General Public License v3.0 or later. See the LICENSE file for details.

---

Development

Reversi42 was developed using modern software engineering practices and AI-assisted development tools. The codebase demonstrates:

• SOLID principles and clean architecture
• Comprehensive test coverage
• Extensive documentation
• Type hints and static analysis
• CI/CD integration

For development setup and contribution guidelines, see Development Guide.

---

Acknowledgments

• Donato Barnaba and Federazione Italiana Gioco Othello (FNGO) - Reversi expertise and guidance
• PointyStone3 Project - Opening book data contributions
• Cursor - AI-powered development environment

---

Author

Luca Amore
Email: [e-mail]
Website: https://www.lucaamore.com

---

Links

• Homepage: https://www.lucaamore.com
• Repository: https://github.com/lucaamore/reversi42
• Documentation: https://github.com/lucaamore/reversi42/tree/main/docs
• Issue Tracker: https://github.com/lucaamore/reversi42/issues

---

Reversi42 - Professional-grade Reversi implementation for players and researchers alike.