File: minimcl

package info (click to toggle)
mcl 1%3A14-137-1
  • links: PTS, VCS
  • area: main
  • in suites: jessie, jessie-kfreebsd, stretch
  • size: 10,704 kB
  • ctags: 5,125
  • sloc: ansic: 53,217; sh: 4,448; perl: 3,967; makefile: 422
file content (266 lines) | stat: -rwxr-xr-x 6,632 bytes parent folder | download | duplicates (5)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
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
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
#!/usr/local/bin/perl -w

#  (C) Copyright 2006, 2007, 2008, 2009 Stijn van Dongen
 #
#  This file is part of MCL.  You can redistribute and/or modify MCL under the
#  terms of the GNU General Public License; either version 3 of the License or
#  (at your option) any later version.  You should have received a copy of the
#  GPL along with MCL, in the file COPYING.


sub explain {
print <<EOH;
purpose:
   A small mcl implementation for educational purposes.  It is written
   in moderately terse perl.

implementation:
   It is hash based, which implies that we get sparse matrices easily but at
   the cost of using hashes.  The hash-based matrices only store non-zero
   entries.

   The code is pretty straightforward. The interpretation routine implements
   the mapping as described in the publications referenced in the (maxi) mcl
   manual.

bonus:
   Since the implementation is hash based you can use any type of labels, not
   necessarily numbers.

Usage:
   minimcl [--I=<num>] [--verbose] LABEL-INPUT

   This means --I=<num> is optional (with 2.0 the default) and so is --verbose.
   LABEL-INPUT should be a file name or stream (STDIN) where each line is of
   the form
            LABEL1 LABEL2 NUMBER
   or
            LABEL1 LABEL2
EOH
}

use strict;
use Getopt::Long;

$::verbose  =  0;
my $I       =  2.0;
my $help    =  0;

if (!@ARGV) {
   print STDERR "issue 'minimcl --help' for help\n";
   print STDERR "expecting STDIN now\n";
}

if
(! GetOptions
   (  "verbose"         =>   \$::verbose
   ,  "I=f"             =>   \$I
   ,  "help"            =>   \$help
   )
)
   {  print STDERR "option processing failed\n";
      exit(1);
   }

&explain && exit(0) if $help;

my $mx = {};

   ##    This reads the labels into a graph structure.
   ##
while (<>) {
   next if /^\s*#/;
   if (/(\S+)\s+(\S+)\s+(\S+)/) {
      my ($x, $y, $val) = ($1, $2, $3);
      $val = 1.0 if $val !~ /^[0-9]/;
      $mx->{$x}{$y} = $val+0;
      $mx->{$y}{$x} = $val+0;
   }
   elsif (/(\S+)\s+(\S+)/) {
      $mx->{$1}{$2} = 1.0;
      $mx->{$2}{$1} = 1.0;
   }
}

matrix_add_loops($mx);
matrix_make_stochastic($mx);
matrix_dump($mx, 3, "start") if $::verbose;

my ($cl, $limit) = mcl($mx, $I);

matrix_dump($limit, 1, "limit") if $::verbose;
matrix_dump($cl, 0, "clustering");


sub mcl {
   my ($mx, $I) = @_;
   my $chaos = 1;
   my $ite = 1;
   while ($chaos > 0.001) {
      my $sq = matrix_square($mx);
      my $progress = sprintf "chaos %.5f ite %d", $chaos, $ite;
matrix_dump($sq, 3, "X $progress") if $::verbose;
      $chaos = matrix_inflate($sq, $I);
matrix_dump($sq, 3, sprintf "I $progress") if $::verbose;
print STDERR "$progress\n" if !$::verbose;
      $mx = $sq;
      $ite++;
   }
   my $cl = matrix_interpret($mx);
   return ($cl, $mx);
}

                        # dangersign:
                        # can this yield a < b < c < a ?

sub cmpany { local $^W = 0; $a <=> $b || $a cmp $b }

sub matrix_dump {
   my ($mx, $modes, $msg) = @_;
   print "($msg\n";
   for my $n (sort cmpany keys %$mx) {
      my @nb =    $modes & 2
               ?  map { sprintf "%s:%.3f", $_, $mx->{$n}{$_}; } sort cmpany keys %{$mx->{$n}}
               :  map { sprintf "%s", $_; } sort cmpany keys %{$mx->{$n}};
      local $" = "\t";
      if ($modes & 1) {
         printf "%-20s%s\n", $n, "@nb";
      }
      else {
         print "@nb\n";
      }
   }
   print ")\n";
}


sub matrix_square {
   my ($mx) = @_;
   my $sq = {};
   my @nodes = keys %$mx;
   for my $n (@nodes) {
      $sq->{$n} = matrix_multiply_vector($mx, $mx->{$n});
   }
   return $sq;
}


sub matrix_multiply_vector {
   my ($mx, $v) = @_;
   my $w = {};
   for my $e (keys %$v) {
      my $val = $v->{$e};
      for my $f (keys %{$mx->{$e}}) {
         $w->{$f} += $val * $mx->{$e}{$f};
      }
   }
   return $w;
}


sub matrix_make_stochastic {
   my ($mx) = @_;
   matrix_inflate($mx, 1); # return value chaos is meaningless for
                           # non stochastic input.
}

sub matrix_add_loops {
   my ($mx) = @_;
   for my $n (keys %$mx) {
      $mx->{$n}{$n} = 0 if defined($mx->{$n}{$n});
      my $max        =  vector_max($mx->{$n});
      $mx->{$n}{$n}  =  $max ? $max : 1;
   }
}

sub vector_max {
   my ($v) = (@_);
   my $max = 0;
   for my $n (keys %$v) {
      $max = $v->{$n} if $v->{$n} > $max;
   }
   return $max;
}

sub vector_sum {
   my ($v, $p) = (@_);
   my $sum = 0;
   for my $n (keys %$v) {
      $sum += $v->{$n} ** $p;
   }
   return $sum;
}


sub matrix_inflate {          # prunes small elements as well.
   my ($mx, $I) = @_;
   my @nodes = keys %$mx;
   my $chaos = 0;
   for my $n (@nodes) {
      my $sum = 0;
      my $sumsq = 0;
      my $max = 0;
      for my $nb (keys %{$mx->{$n}}) {
         if ($mx->{$n}{$nb} < 0.00001) {
            delete($mx->{$n}{$nb});
            next;
         }
         $mx->{$n}{$nb} **= $I;
         $sum += $mx->{$n}{$nb};
      }
      if ($sum) {
         for my $nb (keys %{$mx->{$n}}) {
            $mx->{$n}{$nb} /= $sum;
            $sumsq += $mx->{$n}{$nb} ** 2;  # sum x_i^2 over stochastic vector x
            $max = $mx->{$n}{$nb} if $max < $mx->{$n}{$nb};
         }
      }
      $chaos = $max - $sumsq if $max - $sumsq > $chaos;
   }
   return $chaos;       # only meaningful if input is stochastic
}


                        # assumes but does not check doubly idempotent matrix.
                        # can handle attractor systems of size < 10.
sub matrix_interpret {  # recognizes/preserves overlap.

   my ($limit) =  @_;
   my $clusters=  {};   # hash of arrayrefs.
   my $attrid  =  {};
   my $clid    =  0;

   for my $n (keys %$limit) {           # crude removal of small elements.
      for my $nb (keys %{$limit->{$n}}) {
         delete $limit->{$n}{$nb} if $limit->{$n}{$nb} < 0.1;
      }
   }
   my $attr    =  { map { ($_, 1) } grep { $limit->{$_}{$_} } keys %$limit };
                        # _ contract 'connected attractors', assign cluster id.
   for my $a (keys %$attr) {
      next if defined($attrid->{$a});
      my @aa = ($a);
      while (@aa) {
         my @bb = ();
         for my $aa (@aa) {
            $attrid->{$aa} = $clid;
            push @bb, grep { defined($attr->{$_}) } keys %{$limit->{$aa}};
         }
         @aa = grep { !defined($attrid->{$_}) } @bb;
      }
      $clid++;
   }
   for my $n (keys %$limit) {
      if (!defined($attr->{$n})) {     # look at attractors
         for my $a (grep { defined($attr->{$_}) } keys %{$limit->{$n}}) {
            $clusters->{$attrid->{$a}}{$n}++;
         }
      }
      else {
         $clusters->{$attrid->{$n}}{$n}++; 
      }
   }
   return $clusters;
}