#!/usr/bin/perl -w
use strict;
use Data::Dumper;
use Math::Pari qw/binomial pari2num/;

my $val = hypergeometric(5,3,53,5); 
print "Hyper = $val\n";

sub hypergeometric {

    my ($M,$p,$F,$n) = @_;
    return unless $n>0 && $n == int($n) && $p > 0 && $p == int($p) &&
    $M > 0 && $M <= $n+$p;
    return 0 unless $p <= $M && $p == int($p);

    return pari2num((binomial($M,$p) * binomial($F-$M, $n-$p)
                      / binomial($F,$n))); 

    }
[download]

Hyper = 0.00393074501208321
[download]

Erm, more context might be helpful. Do you perhaps mean something to do with an hypergeometric distribution? If so nothing springs readily to mind from CPAN, but if you get anything working the author of Statistics::Distributions might be interested in a patch.

Are you looking for the Fisher's Exact Test perchance? If so, below is a home-grown implementation for the 2x2 one-tailed case. Requires Math::Pari, but you can eliminate this dependency by writing your own factorial function, and changing $max_arr to something like 170 (or whatever your architecture can handle without overflow).

The code is not very well documented. Sorry. You will have to read some source code to understand what it's doing. Feedback/comments welcome.

Update 3: This code is OBSOLETE. See the new improved version here.

Stupid question, but did you try Searching CPAN?

I found a few references to hypergeometric.