stepga.m

matlab7 gads工具箱

function [nextScore,nextPopulation,state] = stepGA(thisScore,thisPopulation,options,state,GenomeLength,FitnessFcn)
%STEPGA Moves the genetic algorithm forward by one generation
%   This function is private to GA.

%   Copyright 2004 The MathWorks, Inc.
%   $Revision: 1.8 $  $Date: 2004/01/16 16:50:05 $

% how many crossover offspring will there be from each source?
nEliteKids = options.EliteCount;
nXoverKids = round(options.CrossoverFraction * (size(thisPopulation,1) - nEliteKids));
nMutateKids = size(thisPopulation,1) - nEliteKids - nXoverKids;
% how many parents will we need to complete the population?
nParents = 2 * nXoverKids + nMutateKids;

% decide who will contribute to the next generation

% fitness scaling
state.Expectation = feval(options.FitnessScalingFcn,thisScore,nParents,options.FitnessScalingFcnArgs{:});

% selection. parents are indicies into thispopulation
parents = feval(options.SelectionFcn,state.Expectation,nParents,options,options.SelectionFcnArgs{:});

% shuffle to prevent locality effects. It is not the responsibility
% if the selection function to return parents in a "good" order so
% we make sure there is a random order here.
parents = parents(randperm(length(parents)));

[unused,k] = sort(thisScore);

% Everyones parents are stored here for genealogy display
state.Selection = [k(1:options.EliteCount);parents'];

% here we make all of the members of the next generation
eliteKids  = thisPopulation(k(1:options.EliteCount),:);
xoverKids  = feval(options.CrossoverFcn, parents(1:(2 * nXoverKids)),options,GenomeLength,FitnessFcn,thisScore,thisPopulation,options.CrossoverFcnArgs{:});
mutateKids = feval(options.MutationFcn,  parents((1 + 2 * nXoverKids):end), options,GenomeLength,FitnessFcn,state,thisScore,thisPopulation,options.MutationFcnArgs{:});

% group them into the next generation
nextPopulation = [ eliteKids ; xoverKids ; mutateKids ];


% score the population
%We want to add the vectorizer if fitness function is NOT vectorized
if strcmpi(options.Vectorized, 'off') 
    nextScore = feval(@fcnvectorizer,nextPopulation,FitnessFcn,options.FitnessFcnArgs{:});
else
    nextScore = feval(FitnessFcn,nextPopulation,options.FitnessFcnArgs{:});
end