Crate simple_mcts

A Rust library providing a highly configurable and efficient Monte Carlo Tree Search (MCTS) implementation.

This library supports various game types and integrates with external game evaluators for flexible AI development. It includes both single-instance and batch-processing capabilities for MCTS.

Modules

• tree: Implements the core tree data structure used by MCTS.
• game: Defines traits for game logic and state evaluation.
• mcts: Provides the single-instance MCTS algorithm.
• mcts_batch: Offers an MCTS implementation capable of processing multiple game instances in parallel for increased throughput.
• utils: Contains general utility functions.
• test_utils: Provides helper implementations for testing the MCTS algorithms.

Examples

use simple_mcts::{Mcts, test_utils::{GameTest, GameEvaluatorTest2}, MctsError};

fn main() -> Result<(), MctsError> {
    let mut mcts: Mcts<GameTest, 4> = Mcts::<GameTest, 4>::new();
    let evaluator = GameEvaluatorTest2::new();

    // Perform 100 MCTS iterations
    for _ in 0..100 {
        mcts.iterate(&evaluator)?;
    }

    // Get the best action based on visit counts
    let (score, policy) = mcts.get_result();
    println!("Best action score: {}, Policy: {:?}", score, policy);

    // Play the best action and update the MCTS tree
    let best_action_index = policy.iter()
                                  .enumerate()
                                  .max_by(|(_, &a), (_, &b)| a.partial_cmp(&b).unwrap())
                                  .map(|(index, _)| index)
                                  .unwrap_or(0); // Default to first action if policy is empty
    mcts.play(best_action_index)?;

    // Continue with the next game state
    Ok(())
}

This library aims to be a robust foundation for AI development in board games, particularly those benefiting from tree search algorithms like AlphaZero.

Modules

utils

Structs

Mcts

The Monte Carlo Tree Search algorithm implementation.

MctsBatch

Manages a collection of MCTS simulations that can be processed in parallel.

MctsBatchConfig

Configuration parameters for an MctsBatch manager.

MctsConfig

Configuration parameters for a single Monte Carlo Tree Search (MCTS) instance.

MctsState

Represents the current state of an MCTS (Monte Carlo Tree Search) instance, controlling the flow of operations and preventing invalid sequential calls.

Enums

MctsError

Represents possible errors that can occur during MCTS (Monte Carlo Tree Search) operations.

Traits

Game

Trait defining the interface for a game that can be used with MCTS.

GameEvaluator

Trait for evaluating game states and providing policy suggestions.

Functions

default_selection_score

A selection function that combines value, visit count, and initial policy.

ucb1

The standard Upper Confidence Bound 1 (UCB1) selection function.

Type Aliases

SelectionFunction

Type alias for a function pointer used to determine a node’s selection score during MCTS.