qin-andy / shark-chess

An chess tourney simulator for evaluating move generation algorithms' relative strengths

shark-chess

An fullstack automatic chess tournament simulator and visualizer written in Python and TypeScript/React.

• Implement bots with custom move generation heuristics (compatible with existing UCIs such as Stockfish, Komodo, etc)
• Arrange round robin tourneys between bots with customizable match lengths, max moves, engine thinking speeds, etc
• Export tourney settings game PGNs, aggregate match results, bot settings, and bot performance statistics to CSV, JSON, or MongoDB
• Continue existing tourney results with new bots, appending to existing results and statistics
• Visualize matchups, performance and inspect individual games with a React frontend

Inspired by an incredible series of papers written by Tom Murphy for SIGBOVIK..

Check out some excample games on this Lichess study: https://lichess.org/study/PadCquFg

Setting up Local Development

If developing on Windows, run all commands through WSL

Setting up MongoDB

1. In terminal, start MongoDB with mongod
2. In a different terminal, Start MongoDB shell with mongosh. Run use shark-chess

Setting up Flask server

Uses Python3

1. From root, run cd backend
2. Set up venv with python3 -m venv env and source venv/bin/activate
3. Install dependencies with pip install -r requirements.txt
4. Run flask app with python -m flask --app src/app.py run. API runs on localhost:5000

Setting up Frontend

Check node version with node -v. Developed on v18.14.2

1. From root, cd ui
2. Install dependencies with npm install
3. Run npm start to run live reload on localhost:3000

Running a Tourney

WIP. Gist is to change function calls in main.py to adjust tourney settings, run python src/main.py, Check DB in MongoSh, change API call in MainPage.tsx.