A Distributable Steady State Genetic Algorithm. There are three functions:
(make-population num-indivs genes-per-indiv) ;--> new random population
(get-individual pop) ;--> new untested individual
(put-individual indiv pop) ;--> latest populationIndividuals are maps {:fitness 0 :chromosome [0.3224 -0.9324 0.43214 0.001 -0.43224 ...]}
Genes are random between -1.0 and 1.0.
Populations are lists of individuals.
get-individual breeds a new individual from two adults using roulette wheel selection, or any untested individuals (with fitness 0)
put-individual takes an individual with fitness > 0 and returns a population with the new indiv added and one other dropped out. The killed individual is based on inverted roulette wheel selection, where lower fitness has higher chance of being selected.
Just add the dssga dependency to your project.clj file:
(defproject my-project "1.0.0"
:dependencies [[org.clojure/clojure "1.5.1"]
[dssga "0.1.1"]])Evolve a chromosome with max positive genes
(use 'dssga.core)
(defn max-ones [pop individuals_tested]
(loop [pop pop individuals_tested individuals_tested]
(if (= 0 individuals_tested)
(reduce (fn [a b] (if (> (:fitness a) (:fitness b)) a b)) {:fitness 0} pop)
(letfn [(calc-fitness [indiv]
(max 0.1 (* 1.0 (reduce (fn [a b] (+ a b)) 0 (:chromosome indiv)))))]
(let [indiv (get-individual pop)
fitness (calc-fitness indiv)]
(recur (put-individual (assoc indiv :fitness fitness) pop) (dec individuals_tested)))))))
;evolves population of 100 individuals, 10 genes each, for 10000 individuals (100 generations)
(max-ones (make-population 100 10) 10000)We're going to add dssga-server and dssga-client functions that use TCP/IP so that the evolution can be distributed across a network.
Copyright © 2014 Matt Parker
Distributed under the Eclipse Public License either version 1.0 or (at your option) any later version.