ai-othello

An Othello rules and search library with bitboards and complete pass handling.

You can play against it compiled to Wasm here: https://thegustafson.com/games/othello

Self-play

The selfplay binary runs deterministic paired-color matches. Each opening is played twice with the engines swapping Black and White.

cargo run --release --bin selfplay -- \
  --a maximum --b expert --games 12 --opening-plies 8 --seed 1

The six built-in profiles are beginner, easy, medium, hard, expert, and maximum. They progressively increase ordinary search depth and introduce an empty-square threshold where the engine solves the rest of the game exactly. Use the --a-* and --b-* options shown by selfplay --help to override them. Library users can read the same settings with search_preset or SEARCH_PRESETS.

Use it

Add the crate to your project:

[dependencies]
ai-othello = "0.1"

use ai_othello::{EvaluationProfile, Position, SearchConfig, search};

let position = Position::default();
let config = SearchConfig::fixed_depth(5, EvaluationProfile::Phase);
let result = search(position, config);

assert!(result.best_move.is_some());

The crate includes reversible moves, perft, phase-aware evaluation, alpha-beta search, exact endgame search, and a Cassio protocol adapter.

cargo test
cargo run
cargo run -- --cassio

Licensed under the MIT License.