AI-Genetic

 view release on metacpan or  search on metacpan

Genetic.pm  view on Meta::CPAN

23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
 
                 tournamentSinglePoint => \&AI::Genetic::Defaults::tournamentSinglePoint,
                 tournamentTwoPoint    => \&AI::Genetic::Defaults::tournamentTwoPoint,
                 tournamentUniform     => \&AI::Genetic::Defaults::tournamentUniform,
 
                 randomSinglePoint => \&AI::Genetic::Defaults::randomSinglePoint,
                 randomTwoPoint    => \&AI::Genetic::Defaults::randomTwoPoint,
                 randomUniform     => \&AI::Genetic::Defaults::randomUniform,
                );
 
# this hash maps the genome types to the
# classes they're defined in.
 
my %_genome2class = (
                     bitvector   => 'AI::Genetic::IndBitVector',
                     rangevector => 'AI::Genetic::IndRangeVector',
                     listvector  => 'AI::Genetic::IndListVector',
                    );
 
##################

Genetic/Defaults.pm  view on Meta::CPAN

98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
    no strict 'refs';
    $ssub = \&{"AI::Genetic::OpSelection::$selOp"};
    $xsub = \&{"AI::Genetic::OpCrossover::$Xop"};
    $msub = \&{"AI::Genetic::OpMutation::$mutOp"};
  }
 
  for my $i (1 .. $size/2) {
    my @parents = $ssub->(@$selArgs);
    @parents < 2 and push @parents => $ssub->(@$selArgs);
 
    my @cgenes  = $xsub->($crossProb, map scalar $_->genes, @parents);
 
    # check if two didn't mate.
    unless (ref $cgenes[0]) {
      @cgenes = map scalar $_->genes, @parents;
    }
 
    # mutate them.
    $_ = $msub->(@mutArgs, $_) for @cgenes;
 
    # push them into pop.
    push @newPop => map $pop->[0]->new($_), @cgenes;
  }
 
  # assign the fitness function. This is UGLY.
  my $fit = $pop->[0]->fitness;
  $_->fitness($fit) for @newPop;
 
  # now chop in half and reassign the population.
  $ga->people(AI::Genetic::OpSelection::topN([@$pop, @newPop], $size));
}

Genetic/OpSelection.pm  view on Meta::CPAN

15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
my $pop = shift;
 
my $tot = 0;
$tot += $_->score for @$pop;
 
# if all population has zero score, then none
# deserves to be selected.
$tot = 1 unless $tot;    # to avoid div by zero
 
# normalize
my @norms = map {$_->score / $tot} @$pop;
 
@wheel = ();
 
my $cur = 0;
for my $i (@norms) {
  push @wheel => [$cur, $cur + $i];
  $cur += $i;
}
 
$wheelPop = $pop;

Genetic/OpSelection.pm  view on Meta::CPAN

69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
    my $rand = rand;
 
    for my $i (0 .. $#wheel) {
      if ($wheel[$i][0] <= $rand && $rand < $wheel[$i][1]) {
        $selected{$i} = 1;
        last;
      }
    }
  }
 
  return map $wheelPop->[$_], keys %selected;
}
 
# sub tournament():
# arguments are anon list of population, and number
# of individuals in tournament (def = 2).
# return 1 individual.
 
sub tournament {
  my ($pop, $num) = @_;
 
  $num ||= 2;
 
  my %s;
  while ($num > keys %s) {
    my $i = int rand @$pop;
    $s{$i} = 1;
  }
 
  return (sort {$b->score <=> $a->score}
          map {$_->score; $_}  # This avoids a bug in Perl. See Genetic.pm.
          map $pop->[$_], keys %s)[0];
}
 
# sub random():
# pure random choice of individuals.
# arguments are anon list of population, and number
# of individuals to select (def = 1).
# returns selected individual(s).
 
sub random {
  my ($pop, $num) = @_;
 
  $num ||= 1;
 
  my %s;
  while ($num > keys %s) {
    my $i = int rand @$pop;
    $s{$i} = 1;
  }
 
  return map $pop->[$_], keys %s;
}
 
# sub topN():
# fittest N individuals.
# arguments are anon list of pop, and N (def = 1).
# return anon list of top N individuals.
 
sub topN {
  my ($pop, $N) = @_;
 
  $N ||= 1;
 
  # hmm .. are inputs already sorted?
  return [(sort {$b->score <=> $a->score}
           map {$_->score; $_}  # This avoids a bug in Perl. See Genetic.pm.
           @$pop)[0 .. $N-1]];
}
 
1;
 
__END__
 
=head1 NAME
 
AI::Genetic::OpSelection - A class that implements various selection operators.