• Stars
    star
    123
  • Rank 290,145 (Top 6 %)
  • Language
    Go
  • Created over 12 years ago
  • Updated almost 12 years ago

Reviews

There are no reviews yet. Be the first to send feedback to the community and the maintainers!

Repository Details

Genetic problem solver written in Go

GeneticGo - genetic problem solver written in Go

This genetic solver adaptively adjusts the strategies being used to those that are most successful at finding solutions to your problem.

Usage

GeneticGo is compatible with Go 1. Add it to your package repository:

go get "github.com/handcraftsman/Random"
go get "github.com/handcraftsman/GeneticGo"

then use it in your program:

import "github.com/handcraftsman/GeneticGo"

solver := new(genetic.Solver)
solver.MaxSecondsToRunWithoutImprovement = 20 // you decide
solver.LowerFitnessesAreBetter = true // you decide

// Create a fitness function:
//   Return a negative value if the sequence is invalid, otherwise
//   return a value that approaches 0 or MaxInt32 as it gets better
//   depending on what's best for your problem.
//   When hillclimbing, fitness values that are closer to the optimal
//   value are assumed to be better, even if higher than the optimal.
getFitness := func(candidate string) int {
	return ?? // evaluate the candidate and return a fitness value
}

// create a display function
display := func(genes string) {
	println(??) // provide some output to the user if desired
}

// each gene is a single character
geneSet := "abc123..." // you decide the set of valid genes
numberOfGenesInAChromosome := 1 // you decide

solver.NumberOfConcurrentEvolvers = 4 // you decide, defaults to 1
solver.MaxProcs = 4 // you decide, defaults to 1	

if your problem can be solved with a fixed number of genes:

numberOfChromosomes := 10 // you decide
var result = solver.GetBest(getFitness, display, geneSet, numberOfChromosomes, numberOfGenesInAChromosome)

alternatively, if you want the gene sequence to grow as necessary:

solver.MaxRoundsWithoutImprovement = 10 // you decide
bestPossibleFitness := 0 // you decide
maxNumberOfChromosomes := 50 // you decide

var result = solver.GetBestUsingHillClimbing(getFitness, display, geneSet, maxNumberOfChromosomes, numberOfGenesInAChromosome, bestPossibleFitness)

Sample programs (in order of genetic complexity)

  • string_duplication.go - duplicates a string, see related blog post

    go run samples/string_duplication/string_duplication.go

  • 8queens.go - solves the 8 Queens Puzzle, see related blog post

    go run samples/8queens/8queens.go

  • tsp.go - travelling salesperson problem solver. See related blog post

    go run samples/tsp/tsp.go samples/tsp/data/eil51.tsp

    prerequisite: go get "github.com/handcraftsman/File"

  • regex.go - genetically builds a regular expression. See related blog post

    go run samples/regex/builder.go

  • unbounded knapsack problem solver. See related blog post

    go run samples/ukp/rosetta/rosetta.go

    go run samples/ukp/standard/standard.go samples/ukp/data/exnsd16.ukp

    prerequisite: go get "github.com/handcraftsman/File"

  • lawnmower problem solver. See related blog post

    go run samples/lawnmower/*.go

    prerequisite: go get "github.com/handcraftsman/Interpreter"

License

MIT License