improvements, password test and wip salesman

src/demos/findPassword.ts

@@ -0,0 +1,54 @@
+import { Selection } from "../genetic/methods";
+import { Candidate, Population } from "../genetic/model";
+
+export function findPassword(password: string) {
+
+	const fitness = (pass: string) => (genes: string[]) => {
+		const passwordArray = pass.split("");
+		if (genes.length !== pass.length) {
+			throw Error("passwords length not matching");
+		}
+		let score = 0;
+		for (let i = 0; i < genes.length; i++) {
+			if (genes[i] === passwordArray[i]) {
+				score += 1;
+			}
+		}
+		return score * 100 / pass.length;
+	};
+
+	const mutate = (pass: string) => (genes: string[]) => {
+		const genesCopy = genes.slice();
+		const mutatedIndex = Math.floor(Math.random() * pass.length);
+		genesCopy[mutatedIndex] = getRandomLetter();
+		return genesCopy;
+	};
+
+	const generateRandomGenes = (pass: string) => () => {
+		const genes = [];
+		for (let i = 0; i < pass.length; i++) {
+			genes.push(getRandomLetter());
+		}
+		return genes;
+	};
+
+	let pop = Population.generatePopulation(
+		100,
+		generateRandomGenes(password),
+		{
+			fitness: fitness(password),
+			mutate: mutate(password),
+			mutationProbability: 0.4
+		}
+	);
+
+	for (let i = 0; i < 100; i++) {
+		document.getElementById("main")!.innerHTML
+			+= ("<br/><br />" + pop.candidates[0].genes + ", fitness: " + pop.candidates[0].fitness(pop.candidates[0].genes));
+		pop = pop.createNextGeneration();
+	}
+
+	function getRandomLetter() {
+		return String.fromCharCode(97 + Math.floor(Math.random() * 26));
+	}
+}

src/demos/salesman.ts

@@ -0,0 +1,70 @@
+import { Selection } from "../genetic/methods";
+import { Candidate, Population } from "../genetic/model";
+
+export interface Position {
+	x: number;
+	y: number;
+}
+
+export function salesman(towns: Position[]) {
+
+	const fitness = (genes: number[]) => {
+		const totalDistance = genes.reduce((acc, gene, index) => {
+			if (index === 0) {
+				return 0;
+			}
+			return acc + distance(towns[gene], towns[genes[index - 1]]);
+		});
+		return 1 / totalDistance;
+	};
+
+	const mutate = (genes: number[]) => {
+		const genesCopy = genes.slice();
+		const mutatedIndex1 = Math.floor(Math.random() * genes.length);
+		const mutatedIndex2 = Math.floor(Math.random() * genes.length);
+		[genesCopy[mutatedIndex1], genesCopy[mutatedIndex2]] = [genesCopy[mutatedIndex2], genesCopy[mutatedIndex1]];
+		return genesCopy;
+	};
+
+	const cross = (motherGenes: number[], fatherGenes: number[]) => {
+		const startPos = 0;
+	}
+
+	let pop = Population.generatePopulation(
+		20,
+		createRandomGenotype,
+		{
+			fitness,
+			mutate,
+			cross,
+			mutationProbability: 0.4
+		}
+	);
+
+	for (let i = 0; i < 40; i++) {
+		document.getElementById("main")!.innerHTML
+			+= ("<br/><br />" + pop.candidates[0].genes + ", fitness: " + pop.candidates[0].fitness(pop.candidates[0].genes));
+		pop = pop.createNextGeneration();
+	}
+
+	function createRandomGenotype() {
+		const genotype = Array.from({length: towns.length}, (v, k) => k);
+		shuffleArray(genotype);
+		return genotype;
+	}
+}
+
+function distance(pos1: Position, pos2: Position) {
+	return Math.sqrt(Math.pow(pos2.x - pos1.x, 2) + Math.pow(pos2.y - pos1.y, 2));
+}
+
+/**
+ * Randomize array element order in-place.
+ * Using Durstenfeld shuffle algorithm.
+ */
+function shuffleArray(array: any[]) {
+	for (let i = array.length - 1; i > 0; i--) {
+		const j = Math.floor(Math.random() * (i + 1));
+		[array[i], array[j]] = [array[j], array[i]];
+	}
+}

src/genetic/methods/crossover.ts

@@ -0,0 +1,19 @@
+export class Crossover {
+	static SINGLE_POINT = (crossProbability: number = 0.8) => <T>(motherGenes: T[], fatherGenes: T[]) => {
+		const firstChildGenes: T[] = [];
+		const secondChildGenes: T[] = [];
+		/**
+		 * Cross the two parents genes according to the cross probability
+		 */
+		for (let i = 0; i < motherGenes.length ; i++) {
+			firstChildGenes.push(Math.random() < crossProbability ? motherGenes[i] : fatherGenes[i]);
+			secondChildGenes.push(Math.random() < crossProbability ? fatherGenes[i] : motherGenes[i]);
+		}
+
+		return [
+			firstChildGenes,
+			secondChildGenes
+		];
+
+	}
+}

src/genetic/model/candidate.ts

@@ -2,59 +2,20 @@ import { Gene } from "./gene";
 
 interface CandidateOptions<T> {
 	fitness: (genes: T[]) => number;
-	mutate: (genes: T[]) => T[];
-	crossProbability?: number;
-	mutationProbability?: number;
 }
 
 export class Candidate<T extends Gene> {
 	genes: T[];
 	fitness: (genes: T[]) => number;
-	mutate: (genes: T[]) => T[];
-	crossProbability: number;
-	mutationProbability: number;
 
 	constructor(genes: T[], options: CandidateOptions<T>) {
 		this.genes = genes;
 		this.fitness = options.fitness;
-		this.mutate = options.mutate;
-		this.crossProbability = options.crossProbability || 0.8;
-		this.mutationProbability = options.mutationProbability || 0.2;
-	}
-
-	cross = (other: Candidate<T>) => {
-		let firstChildGenes: T[] = [];
-		let secondChildGenes: T[] = [];
-		/**
-		 * Cross the two parents genes according to the cross probability
-		 */
-		for (let i = 0; i < this.genes.length ; i++) {
-			firstChildGenes.push(Math.random() < this.crossProbability ? this.genes[i] : other.genes[i]);
-			secondChildGenes.push(Math.random() < this.crossProbability ? other.genes[i] : this.genes[i]);
-		}
-
-		/**
-		 * Each child may mutate according to the mutation probability
-		 */
-		if (Math.random() < this.mutationProbability) {
-			firstChildGenes = this.mutate(firstChildGenes);
-		}
-		if (Math.random() < this.mutationProbability) {
-			secondChildGenes = this.mutate(secondChildGenes);
-		}
-
-		return [
-			new Candidate(firstChildGenes, this.getOptions()),
-			new Candidate(secondChildGenes, this.getOptions())
-		];
 	}
 
 	getOptions = (): CandidateOptions<T> => {
 		return {
-			fitness: this.fitness,
-			mutate: this.mutate,
-			crossProbability: this.crossProbability,
-			mutationProbability: this.mutationProbability
+			fitness: this.fitness
 		};
 	}
 }

src/genetic/model/population.ts

@@ -1,13 +1,14 @@
 import { Selection } from "../methods";
+import { Crossover } from "../methods/crossover";
 import { Candidate } from "./candidate";
 import { Gene } from "./gene";
 
 interface PopulationOptions<T> {
 	select?: (candidates: Candidate<T>[]) => Candidate<T>[];
 	fitness?: (genes: T[]) => number;
 	mutate?: (genes: T[]) => T[];
+	cross?: (motherGenes: T[], fatherGenes: T[]) => T[][];
 	mutationProbability?: number;
-	crossProbability?: number;
 	ellitism?: boolean;
 }
 
@@ -16,8 +17,8 @@ export class Population<T extends Gene> {
 	select: (candidates: Candidate<T>[]) => Candidate<T>[];
 	fitness: (genes: T[]) => number;
 	mutate: (genes: T[]) => T[];
-	mutationProbability?: number;
-	crossProbability?: number;
+	cross: (motherGenes: T[], fatherGenes: T[]) => T[][];
+	mutationProbability: number;
 	ellitism: boolean;
 
 	/**
@@ -28,13 +29,10 @@ export class Population<T extends Gene> {
 	 */
 	static generatePopulation<T>(size: number, geneGenerator: () => T[], options: PopulationOptions<T>) {
 		const candidates = [];
-		const { fitness, mutate, mutationProbability, crossProbability } = options;
+		const { fitness, mutate, mutationProbability } = options;
 		for (let i = 0; i < size; i++) {
 			candidates.push(new Candidate(geneGenerator(), {
-				fitness: fitness || (() => 0),
-				mutate: mutate || (g => g),
-				mutationProbability,
-				crossProbability
+				fitness: fitness || (() => 0)
 			}));
 		}
 		return new Population(candidates, options);
@@ -46,6 +44,8 @@ export class Population<T extends Gene> {
 		this.fitness = options.fitness || (() => 0);
 		this.mutate = options.mutate || (g => g);
 		this.ellitism = options.ellitism || false;
+		this.cross = options.cross || Crossover.SINGLE_POINT();
+		this.mutationProbability = options.mutationProbability || 0.2;
 	}
 
 	createNextGeneration = (): Population<T> => {
@@ -57,7 +57,7 @@ export class Population<T extends Gene> {
 		/**
 		 * Cross stronger candidates and mutates their children
 		 */
-		const newCandidates = this.cross(selectedCandidates);
+		const newCandidates = this.crossCandidates(selectedCandidates);
 		/**
 		 * Add the children to the population
 		 */
@@ -80,14 +80,22 @@ export class Population<T extends Gene> {
 		return new Population(cleanedNewGeneration, this.getOptions());
 	}
 
-	cross = (candidates: Candidate<T>[]): Candidate<T>[] => {
+	crossCandidates = (candidates: Candidate<T>[]): Candidate<T>[] => {
 		if (candidates.length % 2 !== 0) {
 			throw Error("Selection size should be a multiple of 2");
 		}
-		let children: Candidate<T>[] = [];
+		const children: Candidate<T>[] = [];
+		const options = candidates[0].getOptions();
 		for (let i = 0; i < candidates.length; i += 2) {
-			const tChildren = candidates[i].cross(candidates[i + 1]);
-			children = children.concat(tChildren);
+			let [firstChildGenes, secondChildGenes] = this.cross(candidates[i].genes, candidates[i + 1].genes);
+			if (Math.random() < this.mutationProbability) {
+				firstChildGenes = this.mutate(firstChildGenes);
+			}
+			if (Math.random() < this.mutationProbability) {
+				secondChildGenes = this.mutate(secondChildGenes);
+			}
+			children.push(new Candidate(firstChildGenes, options));
+			children.push(new Candidate(secondChildGenes, options));
 		}
 		return children;
 	}
@@ -97,8 +105,7 @@ export class Population<T extends Gene> {
 			fitness: this.fitness,
 			mutate: this.mutate,
 			select: this.select,
-			mutationProbability: this.mutationProbability,
-			crossProbability: this.crossProbability
+			mutationProbability: this.mutationProbability
 		};
 	}
 }

src/index.ts

@@ -1,62 +1,13 @@
-import { Selection } from "./genetic/methods";
-import { Candidate, Population } from "./genetic/model";
-
-const PASSWORD = "hardcorepasswordmasterrace";
-const fitness = (password: string) => (genes: string[]) => {
-	const passwordArray = password.split("");
-	if (genes.length !== password.length) {
-		throw Error("passwords length not matching");
-	}
-	let score = 0;
-	for (let i = 0; i < genes.length; i++) {
-		if (genes[i] === passwordArray[i]) {
-			score += 1;
-		}
-	}
-	return score * 100 / password.length;
-};
-
-const mutate = (password: string) => (genes: string[]) => {
-	const genesCopy = genes.slice();
-	const mutatedIndex = Math.floor(Math.random() * PASSWORD.length);
-	genesCopy[mutatedIndex] = getRandomLetter();
-	return genesCopy;
-};
-
-// Create the networks
-const candidates = [];
-
-for (let i = 0; i < 100; i ++) {
-	const genes = generateRandomGenes();
-	const candidate = new Candidate<string>(
-		generateRandomGenes(),
-		{
-			fitness: fitness(PASSWORD),
-			mutate: mutate(PASSWORD),
-			mutationProbability: 0.4
-		}
-	);
-	candidates.push(candidate);
-}
-
-const initialPop = new Population(candidates);
-
-let pop = initialPop;
-
-for (let i = 0; i < 100; i ++) {
-	document.getElementById("main")!.innerHTML
-		+= ("<br/><br />" + pop.candidates[0].genes + ", fitness: " + pop.candidates[0].fitness(pop.candidates[0].genes));
-	pop = pop.createNextGeneration();
-}
-
-function generateRandomGenes() {
-	const genes = [];
-	for (let i = 0; i < PASSWORD.length; i++) {
-		genes.push(getRandomLetter());
-	}
-	return genes;
-}
-
-function getRandomLetter() {
-	return String.fromCharCode(97 + Math.floor(Math.random() * 26));
-}
+import { findPassword } from "./demos/findPassword";
+import { salesman } from "./demos/salesman";
+
+// findPassword("passwordhardcoretropduratrouver");
+salesman([
+	{x: 0, y: 0},
+	{x: 1, y: 1},
+	{x: 2, y: 2},
+	{x: 3, y: 3},
+	{x: 4, y: 4},
+	{x: 4, y: 0},
+	{x: 0, y: 4}
+]);