I know it's a tad late, but here is my $0.01. To my opinion PDL is neglected around here, and I know I use it too little myself.

With the help of PDL I only use 43 chars for the sub, and if I looked right, it's a record in this thread. You may add 8 chars if you want to account for the extra 'use PDL;' statement:

use PDL:
sub p{$t=pdl+pop;$t=$t*(pdl$_)for@_;[list$t]}
[download]

This doesn't work with arrays of unequal length, but that wasn't in the spec. The output:

print join " ", @{p([1..5],[1..5])}; 
1 4 9 16 25
[download]

BTW, tilly, why did you name it 'polynomials'? Isn't it just a breed of recursive vector multiplication? Do I miss something?

And, 'strictness' requires only 3 extra chars with a total of 46/54 chars (54 still is the record):

sub p{my $t=pdl+pop;$t=$t*(pdl$_)for@_;[list$t]}
[download]

Jeroen
"We are not alone"(FZ) Update Thx to tye I now know why tilly talked about polynomials. Sorry. If we still want a matrix-kind solution, it would be like:

A = p1'*p2;
product = crossdiags(A);
[download]

But that's difficult to code in PDL, unfortunately. Here is my 2nd attempt, works only if poly's are equal length (pathetic 164 chars):

sub p{$t=pdl+pop;for(@_){
  my $a=outer$t,pdl+pop;
  transpose($a);
  my $n=-1+nelem$t;
  $t=zeroes($n*2+1);
  $t->slice("$_:".($_+$n))+=$a->slice("$_,:")for 0..$n;
}
[list$t]}
[download]

Will ponder it a bit more and get back....