agentjs-core/dist/utils/ParameterTuner.d.ts

A comprehensive agent-based modeling framework with built-in p5.js visualization

import { EventEmitter } from 'eventemitter3';
import type { ConfigurationManager } from './ConfigurationManager';
import type { DataCollector } from '../analysis/DataCollector';
import type { StatisticsEngine } from '../analysis/StatisticsEngine';
export interface TuningTarget {
    metric: string;
    target: 'maximize' | 'minimize' | 'target';
    targetValue?: number;
    weight: number;
}
export interface TuningParameter {
    key: string;
    min: number;
    max: number;
    step: number;
    current: number;
}
export interface TuningExperiment {
    id: string;
    parameters: Record<string, number>;
    results: Record<string, number>;
    score: number;
    timestamp: number;
}
export interface TuningConfig {
    algorithm: 'grid' | 'random' | 'genetic' | 'bayesian';
    maxIterations: number;
    maxRuntime: number;
    convergenceThreshold: number;
    parallelRuns: number;
    stepsPerExperiment: number;
}
export declare class ParameterTuner extends EventEmitter {
    private configManager;
    private dataCollector;
    private statisticsEngine;
    private config;
    private tuningParameters;
    private tuningTargets;
    private experiments;
    private bestExperiment;
    private isRunning;
    private currentIteration;
    private startTime;
    constructor(configManager: ConfigurationManager, dataCollector: DataCollector, statisticsEngine: StatisticsEngine, config?: Partial<TuningConfig>);
    addTuningParameter(key: string, min: number, max: number, step?: number): void;
    removeTuningParameter(key: string): void;
    addTuningTarget(target: TuningTarget): void;
    removeTuningTarget(metric: string): void;
    startTuning(): Promise<TuningExperiment | null>;
    stopTuning(): void;
    private runGridSearch;
    private runRandomSearch;
    private runGeneticAlgorithm;
    private runBayesianOptimization;
    private runExperiment;
    private simulateSteps;
    private collectExperimentResults;
    private calculateScore;
    private generateRandomParameters;
    private generateGridCombinations;
    private generateAcquisitionParameters;
    private tournamentSelection;
    private crossover;
    private mutate;
    private shouldStop;
    private hasConverged;
    getTuningStatus(): {
        isRunning: boolean;
        currentIteration: number;
        maxIterations: number;
        runtime: number;
        bestScore: number | undefined;
        experimentsCount: number;
    };
    getExperiments(): TuningExperiment[];
    getBestExperiment(): TuningExperiment | null;
    applyBestParameters(): boolean;
    exportResults(): {
        config: TuningConfig;
        parameters: Array<{
            key: string;
        } & TuningParameter>;
        targets: TuningTarget[];
        experiments: TuningExperiment[];
        bestExperiment: TuningExperiment | null;
        summary: {
            totalExperiments: number;
            bestScore: number | null;
            convergenceReached: boolean;
        };
    };
}
//# sourceMappingURL=ParameterTuner.d.ts.map