Algorithm::Graphs::TransitiveClosure - Calculate the transitive closure.

    use Algorithm::Graphs::TransitiveClosure qw /floyd_warshall/;

    my $graph = [[1, 0, 0, 0], [0, 1, 1, 1], [0, 1, 1, 0], [1, 0, 1, 1]];
    floyd_warshall $graph;
    print "There is a path from 2 to 0.\n" if $graph -> [2] -> [0];

    my $graph2 = {one   => {one => 1},
                  two   => {two => 1, three => 1, four => 1},
                  three => {two => 1, three => 1},
                  four  => {one => 1, four  => 1}};
    floyd_warshall $graph2;
    print "There is a path from three to one.\n" if
        $graph2 -> {three} -> {one};

This is an implementation of the well known Floyd-Warshall algorithm. [1,2]

The subroutine floyd_warshall takes a directed graph, and calculates its transitive closure, which will be returned. The given graph is actually modified, so be sure to pass a copy of the graph to the routine if you need to keep the original graph.

The subroutine takes graphs in one of the two following formats:

floyd_warshall ARRAYREF

The graph G = (V, E) is described with a list of lists, $graph, representing V x V. If there is an edge between vertices $i and $j (or if $i == $j), then $graph -> [$i] -> [$j] == 1. For all other pairs ($k, $l) from V x V, $graph -> [$k] -> [$l] == 0.

The resulting $graph will have $graph -> [$i] -> [$j] == 1 iff $i == $j or there is a path in G from $i to $j, and $graph -> [$i] -> [$j] == 0 otherwise.

floyd_warshall HASHREF

The graph G = (V, E), with labeled vertices, is described with a hash of hashes, $graph, representing V x V. If there is an edge between vertices $label1 and $label2 (or if $label1 eq $label2), then $graph -> {$label1} -> {$label2} == 1. For all other pairs ($label3, $label4) from V x V, $graph -> {$label3} -> {$label4} does not exist.

The resulting $graph will have $graph -> {$label1} -> {$label2} == 1 iff $label1 eq $label2 or there is a path in G from $label1 to $label2, and $graph -> {$label1} -> {$label2} does not exist otherwise.

    my $graph = [[1, 0, 0, 0],
                 [0, 1, 1, 1],
                 [0, 1, 1, 0],
                 [1, 0, 1, 1]];
    floyd_warshall $graph;
    foreach my $row (@$graph) {print "@$row\n"}

    1 0 0 0
    1 1 1 1
    1 1 1 1
    1 1 1 1

    my $graph = {one   => {one => 1},
                 two   => {two => 1, three => 1, four => 1},
                 three => {two => 1, three => 1},
                 four  => {one => 1, three => 1, four => 1}};
    floyd_warshall $graph;
    foreach my $l1 (qw /one two three four/) {
        print "$l1: ";
        foreach my $l2 (qw /one two three four/) {
            next if $l1 eq $l2;
            print "$l2 " if $graph -> {$l1} -> {$l2};
        }
        print "\n";
    }

    one: 
    two: one three four 
    three: one two four 
    four: one two three 

The running time of the algorithm is cubed in the number of vertices of the graph. The author of this package is not aware of any faster algorithms, nor of a proof if this is optimal. Note than in specific cases, when the graph can be embedded on surfaces of bounded genus, or in the case of sparse connection matrices, faster algorithms than cubed in the number of vertices exist.

The space used by this algorithm is at most quadratic in the number of vertices, which is optimal as the resulting transitive closure can have a quadratic number of edges. In case when the graph is represented as a list of lists, the quadratic bound will always be achieved, as the list of lists already has that size. The hash of hashes however will use space linear to the size of the resulting transitive closure.

The Floyd-Warshall algorithm is due to Floyd [2], and based on a theorem of Warshall [3]. The implemation of this package is based on an implementation of Floyd-Warshall found in Cormen, Leiserson and Rivest [1].

[1]

Thomas H. Cormen, Charles E. Leiserson and Ronald L. Rivest: Introduction to Algorithms. Cambridge: MIT Press, 1990. ISBN 0-262-03141-8.

[2]

Robert W. Floyd: "Algorithm 97 (SHORTEST PATH)". Communications of the ACM, 5(6):345, 1962.

[3]

Stephan Warshall: "A theorem on boolean matrices." Journal of the ACM, 9(1):11-12, 1962.

The current sources of this module are found on github, git://github.com/Abigail/test--regexp.git.

Abigail mailto:algorithm-graphs-transitiveclosure@abigail.be.

Copyright (C) 1998, 2009 by Abigail

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.