Metaheuristics

A ruby library of metaheuristics for search/optimisation.

gem install metaheuristics

Usage

For a given problem, the following must be implemented:

  • a genome class (details below)
  • a lambda taking no argument and producing a different randomly generated genome each time it is called
  • an evaluator lambda taking as argument a genome object, and returning a fitness object which much implement the Comparable interface (usually float is used)

The genome class must implements the following methods:

  • clone() return a deep copy of the genome
  • mutate!(mutation_rate) mutates each gene of the genome with probability mutation_rate, return self

  • crossover(genome) return a new genome which is a combination of this genome and the genome parameter

    require 'metaheuristics/metaheuristic_init_factory'

    search_definition = { :name => :tournament_selection, :mutation_rate => 0.1 # depends on the number of genes in a genome }

    search_init = MetaheuristicInitFactory.from_definition(search_definition)

    search = search_init.call( :genome_init => genome_init, # the lambda generating random genomes :evaluator => evaluator_minimiser) # the evaluator lambda

    # run the search until the fitness is over 9000! begin search.run_once end while search.results.best[:fitness] <= 9000

A rule of thumb for the mutation rate is to aim for one mutation per genome, the mutation rate should then be the inverse of the number of genes in a genome.

The file examples/example_minimiser.rb demonstrates usage for a simple minimisation problem.

Implementing new algorithms

New algorithms should implement the MetaheuristicInterface, and be given an entry in MetaheuristicInitFactory::from_definition(). Then if you are happy with the licensing, please send me a pull request!

Algorithms Details

At the moment metaheuristics contains only Genetic Algorithms (GAs), but has vocation to not only contain genome based methods, but also eventually array-of-values based methods.

  • The traditional tournament selection.
  • An implementation of the ALPS paradigm using tournament selection in the layers.
  • FgrnGa, a GA in which fittest individual are kept in the population across several generations.

If you don't know which one to use, ALPS is recommended.

For more on metaheuristics, check out [Sean Luke's great book][http://cs.gmu.edu/~sean/book/metaheuristics/], available for download for free, or in printed paper form for money.

Tournament selection

A simple, basic GA, where each parent is selected by taking a random sample of the population, and choosing the fittest individual in that sample.

Age-Layered Population Structure (ALPS)

An implementation of the ALPS paradigm which aims to prevent premature convergence, combined here with tournament selection in the layers. [Greg Hornby's ALPS paper][http://idesign.ucsc.edu/papers/hornby_gecco06.pdf]

FgrnGa

An implementation of a GA keeping fit parents across multiple generations. It has notably been used to evolve FGRN genomes, which explains the naming. A description is available in Peter Bentley's thesis.

License

This software is provided under an open source MIT-like license:

Copyright (c) 2012 Jean-Baptiste Krohn http://susano.org

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.