Granted that others have already posted on this, but here's my take on it. :)

First, I'd do the split as "split(/\s+/,$list_of_num)". \s is 'whitespace' which includes ' ', \t (tab), \f (form feed), \r (carrage return) and, \n (newline).

If you don't need the original inner lists in the form they were read in, why not make them a hash first? this would save you a traversal of the list to make a hash of it for the uniqe-ing part.

use strict;
use Data::Dumper;

my @input = ("1 2 3",
             "5 5 6",
             "1 4 6"
            );

# outer map: take each input line
# inner map: split it on white space and return it as a hash ref
my @array = map { 
                 { map { $_ => 1 } split(/\s+/,$_) }
                } @input;

print Dumper @array;

my %unique;
for (my $i=0; $i<=$#array; $i++) {
        foreach (keys %{$array[$i]}) {
                $unique{$_}->{'pos'}->{$i} = 1;
                $unique{$_}->{'count'}++;
        }
}

my @duplicates =  grep { $unique{$_}->{'count'} > 1 } keys %unique;

foreach (@duplicates) {
        print "DUP: $_ In lists: ";

        print join(",",keys %{$unique{$_}->{'pos'}});

        print "\n";
}
[download]

This is essentially the same idea as others have posted, execpt this keeps track of which sets each duplicate was found in.

/\/\averick