evjs v0.2.6

EvJs

Advanced genetic and evolutionary algorithm library written in Javascript by notVitaliy.

Install

yarn add evjs

How To

import { EvJs } from 'evjs'

const seed = () => {
  // Seed code here
}
const fitness = () => {
  // calculate fitness score
}
const mutate = () => {
  // mutate an individuals param(s)
}
const mate = () => {
  // breed 2 individuals
}

const evjsConfig = {
  notification: 0.5 // emit 50% of the logs
}

const generationConfig = {
  size: 10,
  crossover: 0.7,
  mutation: 0.4,
  keepFittest: true,
  select: 'random',
  pair: 'tournament2',
  optimizeKey: 'Max'
}

const individualConfig = {
  fitness,
  mutate,
  mate
}

const config = Object.assign({}, evjsConfig, generationConfig, individualConfig)

const evjs = new EvJs(config)

evjs.populate(seed)
evjs.run()

Generation Configuration Parameters

interface GenerationConfig {
  size?: number
  crossover?: number
  mutation?: number
  keepFittest?: boolean
  optimizeKey?: 'Max' | 'Min'
  select: string
  selectN?: number
  pair?: string
}

SelectType

Optimizer

The optimizer specifies how to rank individuals against each other based on an arbitrary fitness score. For example, minimizing the sum of squared error for a regression curve Min would be used, as a smaller fitness score is indicative of better fit.

Selection

An algorithm can be either genetic or evolutionary depending on which selection operations are used. An algorithm is evolutionary if it only uses a Single SelectType. If both Single and Pair-wise operations are used (and if crossover is implemented) it is genetic.

Individual Configuration Parameters

interface IndividualConfig {
  fitness: (entity: any): number
  mutate: (entity: any): any
  mate: (mother: any, father: any): [any, any]
}

Building

To clone, build, and test Genetic.js issue the following command:

git clone git@github.com:notvitaliy/evjs.git

Contributing

Feel free to open issues and send pull-requests.