lib/Algorithm/Evolutionary/Op/Breeder_Diverser.pm

filter: Statementstmt 100% 15 / 15 Branchbran n/a Conditioncond n/a MC/DCmcdc n/a Subroutinesub 100% 5 / 5 Podpod n/a totaltotal 100% 20 / 20 CC 1 SCAR 0.0
CC
1
Cov
100%
BEGIN
1.0
BEGIN
1.0
BEGIN
1.0
CRAP 1.0
« lib/Algorithm/Evolutionary/Op/Breeder.pm ↑ Summary lib/Algorithm/Evolutionary/Op/CanonicalGA.pm »
1 1 ▶use strict;
Subroutine BEGIN: called CC=1 SCAR=0.0
2 1 ▶use warnings;
Subroutine BEGIN: called CC=1 SCAR=0.0
3  
4 =head1 NAME
5 Algorithm::Evolutionary::Op::Breeder_Diverser - Like Breeder, only it tries to cross only individuals that are different
6  
7 =head1 SYNOPSIS
8  
9 use Algorithm::Evolutionary qw( Individual::BitString
10 Op::Mutation Op::Crossover
11 Op::RouletteWheel
12 Op::Breeder_Diverser);
13  
14 use Algorithm::Evolutionary::Utils qw(average);
15  
16 my @pop;
17 my $number_of_bits = 20;
18 my $population_size = 20;
19 my $replacement_rate = 0.5;
20 for ( 1..$population_size ) {
21 my $indi = new Algorithm::Evolutionary::Individual::BitString $number_of_bits ; #Creates random individual
22 $indi->evaluate( $onemax );
23 push( @pop, $indi );
24 }
25  
26 my $m = new Algorithm::Evolutionary::Op::Mutation 0.5;
27 my $c = new Algorithm::Evolutionary::Op::Crossover; #Classical 2-point crossover
28  
29 my $selector = new Algorithm::Evolutionary::Op::RouletteWheel $population_size; #One of the possible selectors
30  
31 my $generation =
32 new Algorithm::Evolutionary::Op::Breeder_Diverser( $selector, [$m, $c] );
33  
34 my @sortPop = sort { $b->Fitness() <=> $a->Fitness() } @pop;
35 my $bestIndi = $sortPop[0];
36 my $previous_average = average( \@sortPop );
37 $generation->apply( \@sortPop );
38  
39 =head1 Base Class
40  
41 L<Algorithm::Evolutionary::Op::Base>
42  
43 =head1 DESCRIPTION
44  
45 Breeder part of the evolutionary algorithm; takes a population and
46 returns another created from the first. Different from
47 L<Algorithm::Evolutionary::Op::Breeder>: tries to avoid crossover
48 among the same individuals and also re-creating an individual already
49 in the pool. In that sense it "diverses", tries to diversify the
50 population. In general, it works better in environments where high
51 diversity is needed (like, for instance, in L<Algorithm::MasterMind>.
52  
53 =head1 METHODS
54  
55 =cut
56  
57 package Algorithm::Evolutionary::Op::Breeder_Diverser;
58  
59 1 ▶use lib qw(../../..);
Subroutine BEGIN: called CC=1 SCAR=0.0
60  
61 our ($VERSION) = ( '$Revision: 1.7 $ ' =~ / (\d+\.\d+)/ ) ;
62  
63 1 ▶use Carp;
Subroutine BEGIN: called CC=1 SCAR=0.0
64  
65 1 ▶use base 'Algorithm::Evolutionary::Op::Base';
Subroutine BEGIN: called CC=1 SCAR=0.0
66  
67 use Algorithm::Evolutionary qw(Wheel
68 Op::Tournament_Selection);
69  
70 # Class-wide constants
71 our $APPLIESTO = 'ARRAY';
72 our $ARITY = 1;
73  
74 =head2 new( $ref_to_operator_array[, $selector = new Algorithm::Evolutionary::Op::Tournament_Selection 2 ] )
75  
76 Creates a breeder, with a selector and array of operators
77  
78 =cut
79  
80 sub new {
81 my $class = shift;
82 my $self = {};
83 $self->{_ops} = shift || croak "No operators found";
84 $self->{_selector} = shift
85 || new Algorithm::Evolutionary::Op::Tournament_Selection 2;
86 bless $self, $class;
87 return $self;
88 }
89  
90 =head2 apply( $population[, $how_many || $population_size] )
91  
92 Applies the algorithm to the population, which should have
93 been evaluated first; checks that it receives a
94 ref-to-array as input, croaks if it does not. Returns a sorted,
95 culled, evaluated population for next generation.
96  
97 It is valid only for string-denominated chromosomes. Checks that the
98 offspring is different from parents before inserting it.
99  
100 =cut
101  
102 sub apply ($) {
103 my $self = shift;
104 my $pop = shift || croak "No population here";
105 my $output_size = shift || @$pop; # Defaults to pop size
106 my @ops = @{$self->{_ops}};
107  
108 #Select for breeding
109 my $selector = $self->{_selector};
110 my @genitors = $selector->apply( $pop );
111  
112 #Reproduce
113 my $totRate = 0;
114 my @rates;
115 for ( @ops ) {
116 push( @rates, $_->{rate});
117 }
118 my $op_wheel = new Algorithm::Evolutionary::Wheel @rates;
119  
120 my @new_population;
121 my $i = 0;
122 while ( @new_population < $output_size ) {
123 my @offspring;
124 my $selected_op = $ops[ $op_wheel->spin()];
125 my $chosen = $genitors[ $i++ % @genitors]; #Chosen in turn
126 push( @offspring, $chosen->clone() );
127 if( $selected_op->{'_arity'} == 2 ) {
128 my $another_one;
129 do {
130 $another_one = $genitors[ rand( @genitors )];
131 } until ( $another_one->{'_str'} ne $chosen->{'_str'} );
132 push( @offspring, $another_one );
133 } elsif ( $selected_op->{'_arity'} > 2 ) {
134 for ( my $j = 1; $j < $selected_op->arity(); $j ++ ) {
135 my $chosen = $genitors[ rand( @genitors )];
136 push( @offspring, $chosen->clone() );
137 }
138 }
139 my $mutant = $selected_op->apply( @offspring );
140 my $equal;
141 for my $o (@offspring) {
142 $equal += $o->{'_str'} eq $mutant->{'_str'};
143 }
144 if ( !$equal ) {
145 push( @new_population, $mutant );
146 }
147  
148 }
149 return \@new_population;
150 }
151  
152 =head1 SEE ALSO
153  
154 More or less in the same ballpark, alternatives to this one
155  
156 =over 4
157  
158 =item *
159  
160 L<Algorithm::Evolutionary::Op::GeneralGeneration>
161  
162 =item *
163  
164 L<Algorithm::Evolutionary::Op::Breeder>
165  
166 =back
167  
168 =head1 Copyright
169
170 This file is released under the GPL. See the LICENSE file included in this distribution,
171 or go to http://www.fsf.org/licenses/gpl.txt
172  
173 CVS Info: $Date: 2013/01/07 13:54:20 $
174 $Header: /media/Backup/Repos/opeal/opeal/Algorithm-Evolutionary/lib/Algorithm/Evolutionary/Op/Breeder_Diverser.pm,v 1.7 2013/01/07 13:54:20 jmerelo Exp $
175 $Author: jmerelo $
176 $Revision: 1.7 $
177  
178 =cut
179  
180 "The truth is out there";
« lib/Algorithm/Evolutionary/Op/Breeder.pm ↑ Summary lib/Algorithm/Evolutionary/Op/CanonicalGA.pm »