| 1 | 46 | ▶ | use strict; #-*-CPerl-*- |
|
Subroutine
BEGIN: called CC=1 SCAR=0.0
| |||
| 2 | 46 | ▶ | use warnings; |
|
Subroutine
BEGIN: called CC=1 SCAR=0.0
| |||
| 3 | |||
| 4 | 46 | ▶ | use lib qw( ../../../lib ); |
|
Subroutine
BEGIN: called CC=1 SCAR=0.0
| |||
| 5 | |||
| 6 | =head1 NAME | ||
| 7 | |||
| 8 | Algorithm::Evolutionary::Utils - Container module with a hodgepodge of functions | ||
| 9 | |||
| 10 | =head1 SYNOPSIS | ||
| 11 | |||
| 12 | use Algorithm::Evolutionary::Utils qw(entropy genotypic_entropy hamming consensus average random_bitstring random_number_array decode_string vector_compare ); | ||
| 13 | |||
| 14 | my $this_entropy = entropy( $population ); | ||
| 15 | |||
| 16 | #Computes consensus sequence (for binary chromosomes | ||
| 17 | my $this_consensus = consensus( $population); | ||
| 18 | |||
| 19 | =head1 DESCRIPTION | ||
| 20 | |||
| 21 | Miscellaneous class that contains functions that might be useful | ||
| 22 | somewhere else, especially when computing EA statistics. | ||
| 23 | |||
| 24 | =cut | ||
| 25 | |||
| 26 | |||
| 27 | =head1 METHODS | ||
| 28 | |||
| 29 | =cut | ||
| 30 | |||
| 31 | package Algorithm::Evolutionary::Utils; | ||
| 32 | |||
| 33 | 46 | ▶ | use Exporter; |
|
Subroutine
BEGIN: called CC=1 SCAR=0.0
| |||
| 34 | our @ISA = qw(Exporter); | ||
| 35 | |||
| 36 | our $VERSION = sprintf "3.4"; | ||
| 37 | |||
| 38 | our @EXPORT_OK = qw( entropy genotypic_entropy consensus hamming | ||
| 39 | random_bitstring random_number_array average | ||
| 40 | parse_xml decode_string vector_compare); | ||
| 41 | |||
| 42 | 46 | ▶ | use Carp; |
|
Subroutine
BEGIN: called CC=1 SCAR=0.0
| |||
| 43 | 46 | ▶ | use String::Random; |
|
Subroutine
BEGIN: called CC=1 SCAR=0.0
| |||
| 44 | 0 | ▶ | use XML::Parser; |
|
Subroutine
BEGIN: called CC=1 SCAR=2.6
| |||
| 45 | use Statistics::Basic qw(mean); | ||
| 46 | |||
| 47 | =head2 entropy( $population) | ||
| 48 | |||
| 49 | Computes the entropy using the well known Shannon's formula: L<http://en.wikipedia.org/wiki/Information_entropy> | ||
| 50 | 'to avoid botching highlighting | ||
| 51 | |||
| 52 | =cut | ||
| 53 | |||
| 54 | sub entropy { | ||
| 55 | my $population = shift; | ||
| 56 | my %frequencies; | ||
| 57 | map( (defined $_->Fitness())?$frequencies{$_->Fitness()}++:1, @$population ); | ||
| 58 | my $entropy = 0; | ||
| 59 | my $gente = scalar(@$population); # Population size | ||
| 60 | for my $f ( keys %frequencies ) { | ||
| 61 | my $this_freq = $frequencies{$f}/$gente; | ||
| 62 | $entropy -= $this_freq*log( $this_freq ); | ||
| 63 | } | ||
| 64 | return $entropy; | ||
| 65 | } | ||
| 66 | |||
| 67 | =head2 genotypic_entropy( $population) | ||
| 68 | |||
| 69 | Computes the entropy using the well known Shannon's formula: | ||
| 70 | L<http://en.wikipedia.org/wiki/Information_entropy> 'to avoid botching | ||
| 71 | highlighting; in this case we use chromosome frequencies instead of | ||
| 72 | fitness. | ||
| 73 | |||
| 74 | =cut | ||
| 75 | |||
| 76 | sub genotypic_entropy { | ||
| 77 | my $population = shift; | ||
| 78 | my %frequencies; | ||
| 79 | map( $frequencies{$_->{'_str'}}++, @$population ); | ||
| 80 | my $entropy = 0; | ||
| 81 | my $gente = scalar(@$population); # Population size | ||
| 82 | for my $f ( keys %frequencies ) { | ||
| 83 | my $this_freq = $frequencies{$f}/$gente; | ||
| 84 | $entropy -= $this_freq*log( $this_freq ); | ||
| 85 | } | ||
| 86 | return $entropy; | ||
| 87 | } | ||
| 88 | |||
| 89 | =head2 hamming( $string_a, $string_b ) | ||
| 90 | |||
| 91 | Computes the number of positions that are different among two strings, the well known Hamming distance. | ||
| 92 | |||
| 93 | =cut | ||
| 94 | |||
| 95 | sub hamming { | ||
| 96 | my ($string_a, $string_b) = @_; | ||
| 97 | return ( ( $string_a ^ $string_b ) =~ tr/\1//); | ||
| 98 | } | ||
| 99 | |||
| 100 | =head2 consensus( $population, $rough = 0 ) | ||
| 101 | |||
| 102 | Consensus sequence representing the majoritary value for each bit; | ||
| 103 | returns the consensus binary string. If "rough", then the bit is set only if the | ||
| 104 | difference is bigger than 0.4 (60/40 proportion). | ||
| 105 | |||
| 106 | =cut | ||
| 107 | |||
| 108 | sub consensus { | ||
| 109 | my $population = shift; | ||
| 110 | my $rough = shift; | ||
| 111 | my @frequencies; | ||
| 112 | for ( @$population ) { | ||
| 113 | for ( my $i = 0; $i < $_->size(); $i ++ ) { | ||
| 114 | if ( !$frequencies[$i] ) { | ||
| 115 | $frequencies[$i]={ 0 => 0, | ||
| 116 | 1 => 0}; | ||
| 117 | } | ||
| 118 | $frequencies[$i]->{substr($_->{'_str'}, $i, 1)}++; | ||
| 119 | } | ||
| 120 | } | ||
| 121 | my $consensus; | ||
| 122 | for my $f ( @frequencies ) { | ||
| 123 | if ( !$rough ) { | ||
| 124 | if ( $f->{'0'} > $f->{'1'} ) { | ||
| 125 | $consensus.='0'; | ||
| 126 | } else { | ||
| 127 | $consensus.='1'; | ||
| 128 | } | ||
| 129 | } else { | ||
| 130 | my $difference = abs( $f->{'0'} - $f->{'1'} ); | ||
| 131 | if ( $difference < 0.4 ) { | ||
| 132 | $consensus .= '-'; | ||
| 133 | } else { | ||
| 134 | if ( $f->{'0'} > $f->{'1'} ) { | ||
| 135 | $consensus.='0'; | ||
| 136 | } else { | ||
| 137 | $consensus.='1'; | ||
| 138 | } | ||
| 139 | } | ||
| 140 | } | ||
| 141 | } | ||
| 142 | return $consensus; | ||
| 143 | } | ||
| 144 | |||
| 145 | =head2 average( $population ) | ||
| 146 | |||
| 147 | Computes an average of population fitness | ||
| 148 | |||
| 149 | =cut | ||
| 150 | |||
| 151 | sub average { | ||
| 152 | my $population = shift; | ||
| 153 | my @frequencies; | ||
| 154 | my @fitnesses = map( $_->Fitness(), @$population ); | ||
| 155 | return mean( @fitnesses ); | ||
| 156 | |||
| 157 | } | ||
| 158 | |||
| 159 | =head2 random_bitstring( $bits ) | ||
| 160 | |||
| 161 | Returns a random bitstring with the stated number of bits. Useful for testing,mainly | ||
| 162 | |||
| 163 | =cut | ||
| 164 | |||
| 165 | sub random_bitstring { | ||
| 166 | my $bits = shift || croak "No bits!"; | ||
| 167 | my $generator = new String::Random; | ||
| 168 | my $regex = "\[01\]{$bits}"; | ||
| 169 | return $generator->randregex($regex); | ||
| 170 | } | ||
| 171 | |||
| 172 | =head2 random_number_array( $dimensions [, $min = -1] [, $range = 2] ) | ||
| 173 | |||
| 174 | Returns a random number array with the stated length. Useful for testing, mainly. | ||
| 175 | |||
| 176 | =cut | ||
| 177 | |||
| 178 | sub random_number_array { | ||
| 179 | my $dimensions = shift || croak "No bits!"; | ||
| 180 | my $min = shift || -1; | ||
| 181 | my $range = shift || 2; | ||
| 182 | |||
| 183 | my @array; | ||
| 184 | for ( my $i = 0; $i < $dimensions; $i ++ ) { | ||
| 185 | push @array, $min + rand($range); | ||
| 186 | } | ||
| 187 | return @array; | ||
| 188 | } | ||
| 189 | |||
| 190 | =head2 parse_xml( $string ) | ||
| 191 | |||
| 192 | Parses the string and returns an XML tree | ||
| 193 | |||
| 194 | =cut | ||
| 195 | |||
| 196 | sub parse_xml { | ||
| 197 | my $string = shift || croak "No string to parse!\n"; | ||
| 198 | my $p=new XML::Parser(Style=>'EasyTree'); | ||
| 199 | $XML::Parser::EasyTree::Noempty=1; | ||
| 200 | my $xml_dom = $p->parse($string) || croak "Problems parsing $string: $!\n"; | ||
| 201 | return $xml_dom; | ||
| 202 | } | ||
| 203 | |||
| 204 | =head2 decode_string( $chromosome, $gene_size, $min, $range ) | ||
| 205 | |||
| 206 | Decodes to a vector, each one of whose components ranges between $min | ||
| 207 | and $max. Returns that vector. | ||
| 208 | |||
| 209 | It does not work for $gene_size too big. Certainly not for 64, maybe for 32. | ||
| 210 | |||
| 211 | =cut | ||
| 212 | |||
| 213 | sub decode_string { | ||
| 214 | my ( $chromosome, $gene_size, $min, $range ) = @_; | ||
| 215 | |||
| 216 | my @output_vector; | ||
| 217 | my $max_range = eval "0b"."1"x$gene_size; | ||
| 218 | for (my $i = 0; $i < length($chromosome)/$gene_size; $i ++ ) { | ||
| 219 | my $substr = substr( $chromosome, $i*$gene_size, $gene_size ); | ||
| 220 | push @output_vector, (($range - $min) * eval("0b$substr") / $max_range) + $min; | ||
| 221 | } | ||
| 222 | return @output_vector; | ||
| 223 | } | ||
| 224 | |||
| 225 | =head2 vector_compare( $vector_1, $vector_2 ) | ||
| 226 | |||
| 227 | Compares vectors, returns 1 if 1 dominates 2, -1 if it's the other way | ||
| 228 | round, and 0 if neither dominates the other. Both vectors are supposed | ||
| 229 | to be numeric. Returns C<undef> if neither is bigger, and they are not | ||
| 230 | equal. Fails if the length is not the same. | ||
| 231 | |||
| 232 | =cut | ||
| 233 | |||
| 234 | sub vector_compare { | ||
| 235 | my ( $vector_1, $vector_2 ) = @_; | ||
| 236 | |||
| 237 | if ( scalar @$vector_1 != scalar @$vector_2 ) { | ||
| 238 | croak "Different lengths, can't compare\n"; | ||
| 239 | } | ||
| 240 | |||
| 241 | my $length = scalar @$vector_1; | ||
| 242 | my @results = map( $vector_1->[$_] <=> $vector_2->[$_], 0..($length-1)); | ||
| 243 | my %comparisons; | ||
| 244 | map( $comparisons{$_}++, @results ); | ||
| 245 | if ( $comparisons{1} && !$comparisons{-1} ) { | ||
| 246 | return 1; | ||
| 247 | } | ||
| 248 | if ( !$comparisons{1} && $comparisons{-1} ) { | ||
| 249 | return -1; | ||
| 250 | } | ||
| 251 | if ( defined $comparisons{0} && $comparisons{0} == $length ) { | ||
| 252 | return 0; | ||
| 253 | } | ||
| 254 | } | ||
| 255 | |||
| 256 | =head1 Copyright | ||
| 257 | |||
| 258 | This file is released under the GPL. See the LICENSE file included in this distribution, | ||
| 259 | or go to http://www.fsf.org/licenses/gpl.txt | ||
| 260 | |||
| 261 | =cut | ||
| 262 | |||
| 263 | "Still there?"; | ||