github.com/tomjcleveland/genetic

This package simplifies the creation of genetic algorithms. To use this package, you must provide an implementation of genetic.Individual, which has the following interface:

// Individual is one member of a population
type Individual interface {
	Crossover(Individual) (Individual, error)
	Mutate(float64) (Individual, error)
	Fitness() (float64, error)
}

In a genetic algorithm, an individual is a candidate solution. The Crossover() and Mutate() methods produce offspring that are minor variations of the parents. In this way, the genetic algorithm traverses the search space, keeping inividuals with high fitness (as determined by Fitness()) and discarding those with low fitness.

Quickstart

package main

import (
    "log"

    "github.com/tomjcleveland/genetic"
)

func main() {
    ctrl, _ := genetic.NewController(genetic.Params{
        Elitism:          3,
        Mutation:         0.5,
        Crossover:        0.7,
        TargetFitness:    30,
        Parallelism:      10,
        SelectionMethod:  genetic.Tournament(10),
        InitPop:          testPopulation(50),
        AdaptiveMutation: true,
    })
    ctrl.Run()
    fittest, _ := ctrl.Fittest()
    log.Printf("Best solution: %v", fittest)
}

func testPopulation(n int) []genetic.Individual {
    // This is the tricky part: implementing genetic.Individual
    // See examples for inspiration
}

API Documentation (GoDoc)

Examples

String Distance

In this simple example, the genetic algorithm starts with randomly generated strings, and searches for strings that are similar to the target string.

Traveling Salesman

This example searches for solutions to the traveling salesman problem, where the goal is to find the shortest possible between a set of points in 2D space.