Recipe 2.13 Multiplying Matrices

2.13.1 Problem

You want to multiply a pair of two-dimensional arrays. Mathematicians and engineers often need this.

2.13.2 Solution

Use the PDL modules, available from CPAN. PDL is the Perl Data Language modules that give fast access to compact matrix and mathematical functions:

use PDL; # $a and $b are both pdl objects $c = $a x $b;

Alternatively, apply the matrix multiplication algorithm to your two-dimensional array:

sub mmult {     my ($m1,$m2) = @_;     my ($m1rows,$m1cols) = matdim($m1);     my ($m2rows,$m2cols) = matdim($m2);     unless ($m1cols =  = $m2rows) {  # raise exception         die "IndexError: matrices don't match: $m1cols != $m2rows";     }     my $result = [  ];     my ($i, $j, $k);     for $i (range($m1rows)) {         for $j (range($m2cols)) {             for $k (range($m1cols)) {                 $result->[$i][$j] += $m1->[$i][$k] * $m2->[$k][$j];             }         }     }     return $result; } sub range { 0 .. ($_[0] - 1) } sub veclen {     my $ary_ref = $_[0];     my $type = ref $ary_ref;     if ($type ne "ARRAY") { die "$type is bad array ref for $ary_ref" }     return scalar(@$ary_ref); } sub matdim {     my $matrix = $_[0];     my $rows = veclen($matrix);     my $cols = veclen($matrix->[0]);     return ($rows, $cols); }

2.13.3 Discussion

If you have the PDL library installed, you can use its lightning-fast manipulation of numbers. This requires far less memory and CPU than Perl's array manipulation. When using PDL objects, many numeric operators (such as + and *) are overloaded and work on an element-by-element basis (e.g., * is the so-called scalar multiplication operator). To get true matrix multiplication, use the overloaded x operator.

use PDL; $a = pdl [     [ 3, 2, 3 ],     [ 5, 9, 8 ], ]; $b = pdl [     [ 4, 7 ],     [ 9, 3 ],     [ 8, 1 ], ]; $c = $a x $b;  # x overload

If you don't have the PDL library, or don't feel like pulling it in for a small problem, you can always do the work yourself the good old-fashioned way.

# mmult( ) and other subroutines as shown earlier $x = [        [ 3, 2, 3 ],        [ 5, 9, 8 ], ]; $y = [        [ 4, 7 ],        [ 9, 3 ],        [ 8, 1 ], ]; $z = mmult($x, $y);

2.13.4 See Also

The documentation with the CPAN module PDL



Perl Cookbook
Perl Cookbook, Second Edition
ISBN: 0596003137
EAN: 2147483647
Year: 2003
Pages: 501

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net