js/genetics.js

import { NeuralNetwork } from "./neuralNetwork.js";
import { randomRange, randomGaussian, clamp } from "./utils.js";

/** Neural network topology shared by all entities */
export const NET_TOPOLOGY = [14, 12, 8];

/** Entity archetype definitions */
export const ENTITY_TYPES = {
  gatherer: {
    name: "Gatherer",
    baseTraits: {
      speed: 2.5,
      size: 8,
      vision: 120,
      attack: 0.2,
      defense: 0.5,
      metabolism: 0.8,
    },
    color: "#00ff88",
    fitnessWeights: { survival: 1.0, energy: 2.0, offspring: 1.5, kills: 0.0 },
    description: "Herbivores that seek food resources",
  },
  predator: {
    name: "Predator",
    baseTraits: {
      speed: 3.5,
      size: 10,
      vision: 140,
      attack: 1.5,
      defense: 0.3,
      metabolism: 1.2,
    },
    color: "#ff3366",
    fitnessWeights: { survival: 0.5, energy: 1.0, offspring: 1.0, kills: 2.5 },
    description: "Carnivores that hunt other entities",
  },
  builder: {
    name: "Builder",
    baseTraits: {
      speed: 1.8,
      size: 9,
      vision: 100,
      attack: 0.1,
      defense: 1.5,
      metabolism: 0.6,
    },
    color: "#3399ff",
    fitnessWeights: { survival: 2.0, energy: 1.5, offspring: 2.0, kills: 0.0 },
    description: "Defensive entities that build and fortify",
  },
  explorer: {
    name: "Explorer",
    baseTraits: {
      speed: 4.0,
      size: 6,
      vision: 180,
      attack: 0.3,
      defense: 0.3,
      metabolism: 1.0,
    },
    color: "#ffaa00",
    fitnessWeights: { survival: 1.5, energy: 1.0, offspring: 1.0, kills: 0.5 },
    description: "Fast scouts with wide vision range",
  },
  hybrid: {
    name: "Hybrid",
    baseTraits: {
      speed: 3.0,
      size: 8,
      vision: 130,
      attack: 0.8,
      defense: 0.8,
      metabolism: 1.0,
    },
    color: "#cc66ff",
    fitnessWeights: { survival: 1.0, energy: 1.0, offspring: 1.5, kills: 1.0 },
    description: "Balanced generalists that adapt to conditions",
  },
};

/** Trait constraints to prevent runaway evolution */
const TRAIT_BOUNDS = {
  speed: { min: 0.5, max: 8.0 },
  size: { min: 2.0, max: 14.0 },
  vision: { min: 30, max: 250 },
  attack: { min: 0.0, max: 3.0 },
  defense: { min: 0.0, max: 3.0 },
  metabolism: { min: 0.2, max: 2.5 },
};

/**
 * Genome encapsulates an entity's heritable traits and neural weights.
 */
export class Genome {
  /**
   * @param {string} type - Entity archetype key
   * @param {object} [traits] - Physical traits (or auto-generated)
   * @param {number[]} [neuralWeights] - Flat weight array (or auto-generated)
   */
  constructor(type, traits = null, neuralWeights = null) {
    this.type = type;
    const base = ENTITY_TYPES[type].baseTraits;

    if (traits) {
      this.traits = { ...traits };
    } else {
      this.traits = {};
      for (const key of Object.keys(base)) {
        const variance = base[key] * 0.15;
        this.traits[key] = clamp(
          base[key] + randomGaussian(0, variance),
          TRAIT_BOUNDS[key].min,
          TRAIT_BOUNDS[key].max,
        );
      }
    }

    if (neuralWeights) {
      this.neuralWeights = neuralWeights.slice();
    } else {
      const tempNet = new NeuralNetwork(NET_TOPOLOGY);
      this.neuralWeights = tempNet.serialize();
    }

    /** Communication symbol preference (0-7), evolvable */
    this.signalPreference = traits
      ? traits._signal || Math.floor(Math.random() * 8)
      : Math.floor(Math.random() * 8);
  }

  /** Create a brain (NeuralNetwork) from this genome's stored weights */
  buildBrain() {
    const nn = new NeuralNetwork(NET_TOPOLOGY);
    nn.deserialize(this.neuralWeights);
    return nn;
  }

  /** Deep copy */
  clone() {
    const g = new Genome(
      this.type,
      { ...this.traits },
      this.neuralWeights.slice(),
    );
    g.signalPreference = this.signalPreference;
    return g;
  }

  /**
   * Single-point crossover of two genomes.
   * Child inherits type from the fitter parent.
   * @param {Genome} other
   * @param {string} [childType] - Override child type
   * @returns {Genome}
   */
  crossover(other, childType = null) {
    const type = childType || this.type;

    const childTraits = {};
    for (const key of Object.keys(this.traits)) {
      if (key === "_signal") continue;
      const t = Math.random();
      const raw = this.traits[key] * t + other.traits[key] * (1 - t);
      childTraits[key] = clamp(
        raw,
        TRAIT_BOUNDS[key].min,
        TRAIT_BOUNDS[key].max,
      );
    }

    const w1 = this.neuralWeights;
    const w2 = other.neuralWeights;
    const len = Math.min(w1.length, w2.length);
    const childWeights = new Array(len);
    const crossPoint = Math.floor(Math.random() * len);
    for (let i = 0; i < len; i++) {
      childWeights[i] = i < crossPoint ? w1[i] : w2[i];
    }

    const child = new Genome(type, childTraits, childWeights);

    child.signalPreference =
      Math.random() < 0.5 ? this.signalPreference : other.signalPreference;

    return child;
  }

  /**
   * Apply mutations to traits and weights.
   * @param {number} traitRate - Probability per trait (e.g. 0.3)
   * @param {number} weightRate - Probability per weight (e.g. 0.05)
   * @param {number} magnitude - Gaussian std for weight mutation
   */
  mutate(traitRate = 0.3, weightRate = 0.05, magnitude = 0.3) {
    const base = ENTITY_TYPES[this.type].baseTraits;

    for (const key of Object.keys(this.traits)) {
      if (key === "_signal") continue;
      if (Math.random() < traitRate) {
        const scale = base[key] * 0.1;
        this.traits[key] = clamp(
          this.traits[key] + randomGaussian(0, scale),
          TRAIT_BOUNDS[key].min,
          TRAIT_BOUNDS[key].max,
        );
      }
    }

    for (let i = 0; i < this.neuralWeights.length; i++) {
      if (Math.random() < weightRate) {
        this.neuralWeights[i] += randomGaussian(0, magnitude);
      }
    }

    if (Math.random() < 0.05) {
      this.signalPreference = Math.floor(Math.random() * 8);
    }
  }

  /** Serialize to a plain object (for LocalStorage) */
  toJSON() {
    return {
      type: this.type,
      traits: { ...this.traits },
      neuralWeights: this.neuralWeights,
      signalPreference: this.signalPreference,
    };
  }

  /** Deserialize from a plain object */
  static fromJSON(data) {
    const g = new Genome(data.type, data.traits, data.neuralWeights);
    g.signalPreference = data.signalPreference || 0;
    return g;
  }
}

/**
 * Create a new random genome of a given type.
 * @param {string} type - Archetype key
 * @returns {Genome}
 */
export function createGenome(type) {
  return new Genome(type);
}

/**
 * Tournament selection: pick the fittest from a random subset.
 * @param {Array<{genome: Genome, fitness: number}>} population
 * @param {number} tournamentSize
 * @returns {Genome}
 */
export function tournamentSelect(population, tournamentSize = 3) {
  let best = null;
  let bestFitness = -Infinity;
  for (let i = 0; i < tournamentSize; i++) {
    const idx = Math.floor(Math.random() * population.length);
    const candidate = population[idx];
    if (candidate.fitness > bestFitness) {
      best = candidate;
      bestFitness = candidate.fitness;
    }
  }
  return best.genome;
}

/**
 * Calculate fitness score for an entity's lifetime stats.
 * @param {string} type - Entity type
 * @param {object} stats - { survivalTime, energyGathered, offspring, kills }
 * @returns {number}
 */
export function calculateFitness(type, stats) {
  const w = ENTITY_TYPES[type].fitnessWeights;
  return (
    w.survival * stats.survivalTime +
    w.energy * stats.energyGathered +
    w.offspring * stats.offspring * 10 +
    w.kills * stats.kills * 5
  );
}

/**
 * Measure genetic diversity as average pairwise trait distance.
 * @param {Genome[]} genomes
 * @returns {number} 0-1 normalized diversity score
 */
export function measureDiversity(genomes) {
  if (genomes.length < 2) return 0;

  const traitKeys = Object.keys(TRAIT_BOUNDS);
  let totalDist = 0;
  let pairs = 0;

  const sampleSize = Math.min(genomes.length, 30);
  for (let i = 0; i < sampleSize; i++) {
    for (let j = i + 1; j < sampleSize; j++) {
      let dist = 0;
      for (const key of traitKeys) {
        const range = TRAIT_BOUNDS[key].max - TRAIT_BOUNDS[key].min;
        const diff =
          Math.abs(genomes[i].traits[key] - genomes[j].traits[key]) / range;
        dist += diff * diff;
      }
      totalDist += Math.sqrt(dist / traitKeys.length);
      pairs++;
    }
  }

  return pairs > 0 ? totalDist / pairs : 0;
}

/**
 * Produce a child genome from the population using selection + crossover + mutation.
 * @param {Array<{genome: Genome, fitness: number}>} population
 * @returns {Genome}
 */
export function reproduce(population) {
  const parent1 = tournamentSelect(population, 3);
  const parent2 = tournamentSelect(population, 3);
  const child = parent1.crossover(parent2);
  child.mutate(0.25, 0.05, 0.2);
  return child;
}