FIDE-Google-Efficient-Chess-AI-Challenge

An efficient chess AI agent for the FIDE & Google Kaggle competition.

https://www.kaggle.com/competitions/fide-google-efficiency-chess-ai-challenge

Features

• Resource-efficient chess engine (< 5 MiB RAM, single 2.20GHz CPU core)
• Minimax search with alpha-beta pruning
• Iterative deepening with time management
• Basic evaluation function including:
  • Material balance
  • Piece-square tables
  • Mobility evaluation

Requirements

• Python 3.7+
• python-chess
• numpy

Installation

1. Clone this repository
2. Install dependencies:

pip install -r requirements.txt

Usage

The agent can be used directly with the Kaggle competition framework. The main entry point is the agent() function in main.py.

Implementation Details

The chess engine implements several key features:

1. Search Algorithm: Minimax with alpha-beta pruning and iterative deepening
2. Evaluation Function:
   • Material counting
   • Basic positional evaluation using piece-square tables
   • Mobility evaluation
3. Time Management: Conservative time usage with early stopping

Competition Constraints

This implementation adheres to the competition constraints:

• 5 MiB RAM limit
• Single 2.20GHz CPU core
• 64KiB compressed submission size limit
• 10s with 0.1s Simple Delay time control