# -*- Mode: Perl -*- # Matrix.pm -- # ITIID : $ITI$ $Header $__Header$ # Author : Ulrich Pfeifer # Created On : Tue Oct 24 18:34:08 1995 # Last Modified By: Ulrich Pfeifer # Last Modified On: Sun Nov 16 10:52:30 2003 # Language : Perl # Update Count : 202 # Status : Unknown, Use with caution! # # Copyright (C) 2002, Bill Denney , all rights reserved. # Copyright (C) 2001, Brian J. Watson , all rights reserved. # Copyright (C) 2001, Ulrich Pfeifer , all rights reserved. # Copyright (C) 1995, Universität Dortmund, all rights reserved. # Copyright (C) 2001, Matthew Brett # # Permission to use this software is granted under the same # restrictions as for Perl itself. # # Revision 0.5 2002/06/02 15:47:40 # Bill Denney added pinvert function # # Revision 0.3 2001/04/17 11:10:15 # Extensions from Brian Watson # # Revision 0.2 1996/07/10 17:48:14 pfeifer # Fixes from Mike Beachy # # Revision 0.1 1995/10/25 09:48:39 pfeifer # Initial revision # =head1 NAME Math::Matrix - Multiply and invert Matrices =head1 SYNOPSIS use Math::Matrix; =head1 DESCRIPTION The following methods are available: =head2 new Constructor arguments are a list of references to arrays of the same length. The arrays are copied. The method returns B in case of error. $a = new Math::Matrix ([rand,rand,rand], [rand,rand,rand], [rand,rand,rand]); If you call C as method, a zero filled matrix with identical deminsions is returned. =head2 clone You can clone a matrix by calling: $b = $a->clone; =head2 size You can determine the dimensions of a matrix by calling: ($m, $n) = $a->size; =head2 concat Concatenates two matrices of same row count. The result is a new matrix or B in case of error. $b = new Math::Matrix ([rand],[rand],[rand]); $c = $a->concat($b); =head2 transpose Returns the transposed matrix. This is the matrix where colums and rows of the argument matrix are swaped. =head2 multiply Multiplies two matrices where the length of the rows in the first matrix is the same as the length of the columns in the second matrix. Returns the product or B in case of error. =head2 solve Solves a equation system given by the matrix. The number of colums must be greater than the number of rows. If variables are dependent from each other, the second and all further of the dependent coefficients are 0. This means the method can handle such systems. The method returns a matrix containing the solutions in its columns or B in case of error. =head2 invert Invert a Matrix using C. =head2 multiply_scalar Multiplies a matrix and a scalar resulting in a matrix of the same dimensions with each element scaled with the scalar. $a->multiply_scalar(2); scale matrix by factor 2 =head2 add Add two matrices of the same dimensions. =head2 substract Shorthand for Cnegative)> =head2 equal Decide if two matrices are equal. The criterion is, that each pair of elements differs less than $Math::Matrix::eps. =head2 slice Extract columns: a->slice(1,3,5); =head2 determinant Compute the determinant of a matrix. =head2 dot_product Compute the dot product of two vectors. =head2 absolute Compute the absolute value of a vector. =head2 normalizing Normalize a vector. =head2 cross_product Compute the cross-product of vectors. =head2 print Prints the matrix on STDOUT. If the method has additional parameters, these are printed before the matrix is printed. =head2 pinvert Compute the pseudo-inverse of the matrix: ((A'A)^-1)A' =head1 EXAMPLE use Math::Matrix; srand(time); $a = new Math::Matrix ([rand,rand,rand], [rand,rand,rand], [rand,rand,rand]); $x = new Math::Matrix ([rand,rand,rand]); $a->print("A\n"); $E = $a->concat($x->transpose); $E->print("Equation system\n"); $s = $E->solve; $s->print("Solutions s\n"); $a->multiply($s)->print("A*s\n"); =head1 AUTHOR Ulrich Pfeifer EFE Brian J. Watson EFE Matthew Brett Ematthew.brett@mrc-cbu.cam.ac.ukE =cut package Math::Matrix; use vars qw($VERSION $eps); use strict; $VERSION = 0.5; use overload '~' => 'transpose', '+' => 'add', '-' => 'subtract', '*' => 'multiply', '""' => 'as_string'; sub version { return "Math::Matrix $VERSION"; } # Implement - array copy, inheritance # class call - new matrix as input # object call - creates matrix with same dimensions matrix sub new { my $that = shift; my $class = ref($that) || $that; my $self = []; if (ref($that) && (@_ == 0)) { # object call no args -> copy matrix for (@$that) { push(@{$self}, [map {0} @{$_}]); } } else { # class call / object call -> matrix as input my $len = scalar(@{$_[0]}); for (@_) { return undef if scalar(@{$_}) != $len; push(@{$self}, [@{$_}]); } } bless $self, $class; } sub clone { my $that = shift; my $self = []; for (@$that) { push(@{$self}, [@{$_}]); } bless $self, ref($that)||$that; } sub size { my $self = shift; my $m = @{$self}; my $n = @{$self->[0]}; ($m, $n); } sub concat { my $self = shift; my $other = shift; my $result = $self->clone(); return undef if scalar(@{$self}) != scalar(@{$other}); for my $i (0 .. $#{$self}) { push @{$result->[$i]}, @{$other->[$i]}; } $result; } sub transpose { my $self = shift; my $class = ref($self); my @result; my $m; for my $col (@{$self->[0]}) { push @result, []; } for my $row (@{$self}) { $m=0; for my $col (@{$row}) { push(@{$result[$m++]}, $col); } } $class->new(@result); } sub vekpro { my($a, $b) = @_; my $result=0; for my $i (0 .. $#{$a}) { $result += $a->[$i] * $b->[$i]; } $result; } sub multiply { my $self = shift; my $class = ref($self); my $other = shift->transpose; my @result; my $m; return undef if $#{$self->[0]} != $#{$other->[0]}; for my $row (@{$self}) { my $rescol = []; for my $col (@{$other}) { push(@{$rescol}, vekpro($row,$col)); } push(@result, $rescol); } $class->new(@result); } $eps = 0.00001; sub solve { my $self = shift; my $class = ref($self); my $m = $self->clone(); my $mr = $#{$m}; my $mc = $#{$m->[0]}; my $f; my $try; return undef if $mc <= $mr; ROW: for(my $i = 0; $i <= $mr; $i++) { $try=$i; # make diagonal element nonzero if possible while (abs($m->[$i]->[$i]) < $eps) { last ROW if $try++ > $mr; my $row = splice(@{$m},$i,1); push(@{$m}, $row); } # normalize row $f = $m->[$i]->[$i]; for(my $k = 0; $k <= $mc; $k++) { $m->[$i]->[$k] /= $f; } # subtract multiple of designated row from other rows for(my $j = 0; $j <= $mr; $j++) { next if $i == $j; $f = $m->[$j]->[$i]; for(my $k = 0; $k <= $mc; $k++) { $m->[$j]->[$k] -= $m->[$i]->[$k] * $f; } } } # Answer is in augmented column transpose $class->new(@{$m->transpose}[$mr+1 .. $mc]); } sub pinvert { my $self = shift; my $class = ref($self); my $m = $self->clone(); $m->transpose->multiply($m)->invert->multiply($m->transpose); } sub print { my $self = shift; print @_ if scalar(@_); print $self->as_string; } sub as_string { my $self = shift; my $out = ""; for my $row (@{$self}) { for my $col (@{$row}) { $out = $out . sprintf "%10.5f ", $col; } $out = $out . sprintf "\n"; } $out; } sub new_identity { my $type = shift; my $class = ref($type) || $type; my $self = []; my $size = shift; for my $i (1..$size) { my $row = []; for my $j (1..$size) { push @$row, $i==$j ? 1 : 0; } push @$self, $row; } bless $self, $class; } sub eye { &new_identity(@_); } sub multiply_scalar { my $self = shift; my $factor = shift; my $result = $self->new(); my $last = $#{$self->[0]}; for my $i (0 .. $#{$self}) { for my $j (0 .. $last) { $result->[$i][$j] = $factor * $self->[$i][$j]; } } $result; } sub negative { shift->multiply_scalar(-1); } sub subtract { my $self = shift; my $other = shift; $self->add($other->negative); } sub equal { my $A = shift; my $B = shift; my $ok = 1; my $last = $#{$A->[0]}; for my $i (0 .. $#{$A}) { for my $j (0 .. $last) { abs($A->[$i][$j]-$B->[$i][$j])<$eps or $ok=0; } } $ok; } sub add { my $self = shift; my $other = shift; my $result = $self->new(); return undef if $#{$self} != $#{$other}; my $last= $#{$self->[0]}; return undef if $last != $#{$other->[0]}; for my $i (0 .. $#{$self}) { for my $j (0 .. $last) { $result->[$i][$j] = $self->[$i][$j] + $other->[$i][$j]; } } $result; } sub slice { my $self = shift; my $class = ref($self); my $result = $class->new([]); foreach my $j (@_) { for my $i (0..$#{$self}) { push @{$result->[$i]}, $self->[$i][$j]; } } $result; } sub determinant { my $self = shift; my $class = ref($self); my $last= $#{$self->[0]}; return undef unless $last == $#{$self}; if ($last == 0) { return $self->[0][0]; } else { my $result = 0; foreach my $col (0..$last) { my $matrix = $self->slice(0..$col-1,$col+1..$last); $matrix = $class->new(@$matrix[1..$last]); my $term += $matrix->determinant(); $term *= $self->[0][$col]; $term *= $col % 2 ? -1 : 1; $result += $term; } return $result; } } # # For vectors only # sub dot_product { my $vector1 = shift; my $vector2 = shift; $vector1 = $vector1->transpose() unless @$vector1 == 1; return undef unless @$vector1 == 1; $vector2 = $vector2->transpose() unless @{$vector2->[0]} == 1; return undef unless @{$vector2->[0]} == 1; return $vector1->multiply($vector2)->[0][0]; } sub absolute { my $vector = shift; sqrt $vector->dot_product($vector); } sub normalize { my $vector = shift; my $length = $vector->absolute(); return undef unless $length; $vector->multiply_scalar(1 / $length); } sub cross_product { my $vectors = shift; my $class = ref($vectors); my $dimensions = @{$vectors->[0]}; return undef unless $dimensions == @$vectors + 1; my @axis; foreach my $column (0..$dimensions-1) { my $tmp = $vectors->slice(0..$column-1, $column+1..$dimensions-1); my $scalar = $tmp->determinant; $scalar *= ($column % 2) ? -1 : 1; push @axis, $scalar; } my $axis = $class->new(\@axis); $axis = $axis->multiply_scalar(($dimensions % 2) ? 1 : -1); } sub invert { my $M = shift; my ($m, $n) = $M->size; my (@I); die "Matrix dimensions are $m X $n. -- Matrix not invertible.\n" if $m != $n; my $I = $M->new_identity($n); ($M->concat($I))->solve; } 1;