MonteCarloTreeSearchPolicy

public protocol MonteCarloTreeSearchPolicy: TreeSearchPolicy, ScoringHeuristic

Policy for more direct control over a strategy’s execution

  • Whether the strategy should abort a given simulation.

    Declaration

    Swift

    func hasReachedMaxSimulationDepth(depth: Int) -> Bool

  • Whether the strategy should execute another simulation.

    Declaration

    Swift

    func shouldContinueSimulations(game: Game, simulationCount: Int) -> Bool

  • Whether the strategy should collapse a given tree into a single leaf node.

    Declaration

    Swift

    func shouldCollapseTree(stats: TreeStats, subtrees: Int, depth: Int) -> Bool

  • Heuristic used for scoring a given move.

    Declaration

    Swift

    func scoreMove(stats: TreeStats, parentPlays: Int) -> Score

  • Calculate reward from an evaluation.

    Example:

    func reward(evaluation: Evaluation<Game.Score>) -> Int {
    switch evaluation {
    case .Victory: return 1
    case .Defeat: return 0
    default: return 0
    }
}

    Declaration

    Swift

    func reward(evaluation: Evaluation<Game.Score>) -> Int

  • Heuristic used for choosing game state subtree to further explore.

    Declaration

    Swift

    func explorationMove<M: GeneratorType where M.Element == (Game.Move, TreeStats)>(availableMoves: M, explorationDepth: Int, plays: Int, randomSource: RandomSource) -> Game.Move?

  • Heuristic used for choosing game state subtree to further explore.

    Declaration

    Swift

    func simulationMove<M: GeneratorType where M.Element == Game.Move>(availableMoves: M, simulationDepth: Int, randomSource: RandomSource) -> Game.Move?