Math::MatrixReal(3) User Contributed Perl Documentation Math::MatrixReal(3)NAMEMath::MatrixReal - Matrix of Reals
Implements the data type "matrix of real numbers" (and consequently
also "vector of real numbers").
SYNOPSIS
my $a = Math::MatrixReal->new_random(5, 5);
my $b = $a->new_random(10, 30, { symmetric=>1, bounded_by=>[-1,1] });
my $c = $b * $a ** 3;
my $d = $b->new_from_rows( [ [ 5, 3 ,4], [3, 4, 5], [ 2, 4, 1 ] ] );
print $a;
my $row = ($a * $b)->row(3);
my $col = (5*$c)->col(2);
my $transpose = ~$c;
my $transpose = $c->transpose;
my $inverse = $a->inverse;
my $inverse = 1/$a;
my $inverse = $a ** -1;
my $determinant= $a->det;
· $matrix->display_precision($integer)
Sets the default precision when matrices are printed or
stringified. $matrix->display_precision(0) will only show the
integer part of all the entries of $matrix and
$matrix->display_precision() will return to the default scientific
display notation. This method does not effect the precision of the
calculations.
FUNCTIONS
Constructor Methods And Such
· use Math::MatrixReal;
Makes the methods and overloaded operators of this module available
to your program.
· $new_matrix = new Math::MatrixReal($rows,$columns);
The matrix object constructor method. A new matrix of size $rows by
$columns will be created, with the value 0.0 for all elements.
Note that this method is implicitly called by many of the other
methods in this module.
· $new_matrix = $some_matrix->new($rows,$columns);
Another way of calling the matrix object constructor method.
Matrix $some_matrix is not changed by this in any way.
· $new_matrix = $matrix->new_from_cols( [
$column_vector|$array_ref|$string, ... ] )
Creates a new matrix given a reference to an array of any of the
following:
· column vectors ( n by 1 Math::MatrixReal matrices )
· references to arrays
· strings properly formatted to create a column with
Math::MatrixReal's new_from_string command
You may mix and match these as you wish. However, all must be of
the same dimension--no padding happens automatically. Example:
my $matrix = Math::MatrixReal->new_from_cols( [ [1,2], [3,4] ] );
print $matrix;
will print
[ 1.000000000000E+00 3.000000000000E+00 ]
[ 2.000000000000E+00 4.000000000000E+00 ]
· new_from_rows( [ $row_vector|$array_ref|$string, ... ] )
Creates a new matrix given a reference to an array of any of the
following:
· row vectors ( 1 by n Math::MatrixReal matrices )
· references to arrays
· strings properly formatted to create a row with
Math::MatrixReal's new_from_string command
You may mix and match these as you wish. However, all must be of
the same dimension--no padding happens automatically. Example:
my $matrix = Math::MatrixReal->new_from_rows( [ [1,2], [3,4] ] );
print $matrix;
will print
[ 1.000000000000E+00 2.000000000000E+00 ]
[ 3.000000000000E+00 4.000000000000E+00 ]
· $new_matrix = Math::MatrixReal->new_random($rows, $cols, %options
);
This method allows you to create a random matrix with various
properties controlled by the %options matrix, which is optional.
The default values of the %options matrix are { integer => 0,
symmetric => 0, tridiagonal => 0, diagonal => 0, bounded_by =>
[0,10] } .
Example:
$matrix = Math::MatrixReal->new_random(4, { diagonal => 1, integer => 1 } );
print $matrix;
will print a 4x4 random diagonal matrix with integer entries
between zero and ten, something like
[ 5.000000000000E+00 0.000000000000E+00 0.000000000000E+00 0.000000000000E+00 ]
[ 0.000000000000E+00 2.000000000000E+00 0.000000000000E+00 0.000000000000E+00 ]
[ 0.000000000000E+00 0.000000000000E+00 1.000000000000E+00 0.000000000000E+00 ]
[ 0.000000000000E+00 0.000000000000E+00 0.000000000000E+00 8.000000000000E+00 ]
· $new_matrix = Math::MatrixReal->new_diag( $array_ref );
This method allows you to create a diagonal matrix by only
specifying the diagonal elements. Example:
$matrix = Math::MatrixReal->new_diag( [ 1,2,3,4 ] );
print $matrix;
will print
[ 1.000000000000E+00 0.000000000000E+00 0.000000000000E+00 0.000000000000E+00 ]
[ 0.000000000000E+00 2.000000000000E+00 0.000000000000E+00 0.000000000000E+00 ]
[ 0.000000000000E+00 0.000000000000E+00 3.000000000000E+00 0.000000000000E+00 ]
[ 0.000000000000E+00 0.000000000000E+00 0.000000000000E+00 4.000000000000E+00 ]
· $new_matrix = Math::MatrixReal->new_tridiag( $lower, $diag, $upper
);
This method allows you to create a tridiagonal matrix by only
specifying the lower diagonal, diagonal and upper diagonal,
respectively.
$matrix = Math::MatrixReal->new_tridiag( [ 6, 4, 2 ], [1,2,3,4], [1, 8, 9] );
print $matrix;
will print
[ 1.000000000000E+00 1.000000000000E+00 0.000000000000E+00 0.000000000000E+00 ]
[ 6.000000000000E+00 2.000000000000E+00 8.000000000000E+00 0.000000000000E+00 ]
[ 0.000000000000E+00 4.000000000000E+00 3.000000000000E+00 9.000000000000E+00 ]
[ 0.000000000000E+00 0.000000000000E+00 2.000000000000E+00 4.000000000000E+00 ]
· $new_matrix = Math::MatrixReal->new_from_string($string);
This method allows you to read in a matrix from a string (for
instance, from the keyboard, from a file or from your code).
The syntax is simple: each row must start with ""[ "" and end with
"" ]\n"" (""\n"" being the newline character and "" "" a space or
tab) and contain one or more numbers, all separated from each other
by spaces or tabs.
Additional spaces or tabs can be added at will, but no comments.
Examples:
$string = "[ 1 2 3 ]\n[ 2 2 -1 ]\n[ 1 1 1 ]\n";
$matrix = Math::MatrixReal->new_from_string($string);
print "$matrix";
By the way, this prints
[ 1.000000000000E+00 2.000000000000E+00 3.000000000000E+00 ]
[ 2.000000000000E+00 2.000000000000E+00 -1.000000000000E+00 ]
[ 1.000000000000E+00 1.000000000000E+00 1.000000000000E+00 ]
But you can also do this in a much more comfortable way using the
shell-like "here-document" syntax:
$matrix = Math::MatrixReal->new_from_string(<<'MATRIX');
[ 1 0 0 0 0 0 1 ]
[ 0 1 0 0 0 0 0 ]
[ 0 0 1 0 0 0 0 ]
[ 0 0 0 1 0 0 0 ]
[ 0 0 0 0 1 0 0 ]
[ 0 0 0 0 0 1 0 ]
[ 1 0 0 0 0 0 -1 ]
MATRIX
You can even use variables in the matrix:
$c1 = 2 / 3;
$c2 = -2 / 5;
$c3 = 26 / 9;
$matrix = Math::MatrixReal->new_from_string(<<"MATRIX");
[ 3 2 0 ]
[ 0 3 2 ]
[ $c1 $c2 $c3 ]
MATRIX
(Remember that you may use spaces and tabs to format the matrix to
your taste)
Note that this method uses exactly the same representation for a
matrix as the "stringify" operator "": this means that you can
convert any matrix into a string with "$string = "$matrix";" and
read it back in later (for instance from a file!).
Note however that you may suffer a precision loss in this process
because only 13 digits are supported in the mantissa when printed!!
If the string you supply (or someone else supplies) does not obey
the syntax mentioned above, an exception is raised, which can be
caught by "eval" as follows:
print "Please enter your matrix (in one line): ";
$string = <STDIN>;
$string =~ s/\\n/\n/g;
eval { $matrix = Math::MatrixReal->new_from_string($string); };
if ($@)
{
print "$@";
# ...
# (error handling)
}
else
{
# continue...
}
or as follows:
eval { $matrix = Math::MatrixReal->new_from_string(<<"MATRIX"); };
[ 3 2 0 ]
[ 0 3 2 ]
[ $c1 $c2 $c3 ]
MATRIX
if ($@)
# ...
Actually, the method shown above for reading a matrix from the
keyboard is a little awkward, since you have to enter a lot of
"\n"'s for the newlines.
A better way is shown in this piece of code:
while (1)
{
print "\nPlease enter your matrix ";
print "(multiple lines, <ctrl-D> = done):\n";
eval { $new_matrix =
Math::MatrixReal->new_from_string(join('',<STDIN>)); };
if ($@)
{
$@ =~ s/\s+at\b.*?$//;
print "${@}Please try again.\n";
}
else { last; }
}
Possible error messages of the "new_from_string()" method are:
Math::MatrixReal::new_from_string(): syntax error in input string
Math::MatrixReal::new_from_string(): empty input string
If the input string has rows with varying numbers of columns, the
following warning will be printed to STDERR:
Math::MatrixReal::new_from_string(): missing elements will be set to zero!
If everything is okay, the method returns an object reference to
the (newly allocated) matrix containing the elements you specified.
· $new_matrix = $some_matrix->shadow();
Returns an object reference to a NEW but EMPTY matrix (filled with
zero's) of the SAME SIZE as matrix "$some_matrix".
Matrix "$some_matrix" is not changed by this in any way.
· $matrix1->copy($matrix2);
Copies the contents of matrix "$matrix2" to an ALREADY EXISTING
matrix "$matrix1" (which must have the same size as matrix
"$matrix2"!).
Matrix "$matrix2" is not changed by this in any way.
· $twin_matrix = $some_matrix->clone();
Returns an object reference to a NEW matrix of the SAME SIZE as
matrix "$some_matrix". The contents of matrix "$some_matrix" have
ALREADY BEEN COPIED to the new matrix "$twin_matrix". This is the
method that the operator "=" is overloaded to when you type "$a =
$b", when $a and $b are matrices.
Matrix "$some_matrix" is not changed by this in any way.
Matrix Row, Column and Element operations
· $row_vector = $matrix->row($row);
This is a projection method which returns an object reference to a
NEW matrix (which in fact is a (row) vector since it has only one
row) to which row number "$row" of matrix "$matrix" has already
been copied.
Matrix "$matrix" is not changed by this in any way.
· $column_vector = $matrix->column($column);
This is a projection method which returns an object reference to a
NEW matrix (which in fact is a (column) vector since it has only
one column) to which column number "$column" of matrix "$matrix"
has already been copied.
Matrix "$matrix" is not changed by this in any way.
· $matrix->assign($row,$column,$value);
Explicitly assigns a value "$value" to a single element of the
matrix "$matrix", located in row "$row" and column "$column",
thereby replacing the value previously stored there.
· $value = $matrix->element($row,$column);
Returns the value of a specific element of the matrix "$matrix",
located in row "$row" and column "$column".
· $new_matrix = $matrix->each( \&function );
Creates a new matrix by evaluating a code reference on each element
of the given matrix. The function is passed the element, the row
index and the column index, in that order. The value the function
returns ( or the value of the last executed statement ) is the
value given to the corresponding element in $new_matrix.
Example:
# add 1 to every element in the matrix
$matrix = $matrix->each ( sub { (shift) + 1 } );
Example:
my $cofactor = $matrix->each( sub { my(undef,$i,$j) = @_;
($i+$j) % 2 == 0 ? $matrix->minor($i,$j)->det()
: -1*$matrix->minor($i,$j)->det();
} );
This code needs some explanation. For each element of $matrix, it
throws away the actual value and stores the row and column indexes
in $i and $j. Then it sets element [$i,$j] in $cofactor to the
determinant of "$matrix->minor($i,$j)" if it is an "even" element,
or "-1*$matrix->minor($i,$j)" if it is an "odd" element.
· $new_matrix = $matrix->each_diag( \&function );
Creates a new matrix by evaluating a code reference on each
diagonal element of the given matrix. The function is passed the
element, the row index and the column index, in that order. The
value the function returns ( or the value of the last executed
statement ) is the value given to the corresponding element in
$new_matrix.
· $matrix->swap_col( $col1, $col2 );
This method takes two one-based column numbers and swaps the values
of each element in each column. "$matrix->swap_col(2,3)" would
replace column 2 in $matrix with column 3, and replace column 3
with column 2.
· $matrix->swap_row( $row1, $row2 );
This method takes two one-based row numbers and swaps the values of
each element in each row. "$matrix->swap_row(2,3)" would replace
row 2 in $matrix with row 3, and replace row 3 with row 2.
· $matrix->assign_row( $row_number , $new_row_vector );
This method takes a one-based row number and assigns row
$row_number of $matrix with $new_row_vector and returns the
resulting matrix. "$matrix->assign_row(5, $x)" would replace row 2
in $matrix with the row vector $x.
Matrix Operations
· $det = $matrix->det();
Returns the determinant of the matrix, without going through the
rigamarole of computing a LR decomposition. This method should be
much faster than LR decomposition if the matrix is diagonal or
triangular. Otherwise, it is just a wrapper for
"$matrix->decompose_LR->det_LR". If the determinant is zero, there
is no inverse and vice-versa. Only quadratic matrices have
determinants.
· "$inverse = $matrix->inverse();"
Returns the inverse of a matrix, without going through the
rigamarole of computing a LR decomposition. If no inverse exists,
undef is returned and an error is printed via "carp()". This is
nothing but a wrapper for "$matrix->decompose_LR->invert_LR".
· "($rows,$columns) = $matrix->dim();"
Returns a list of two items, representing the number of rows and
columns the given matrix "$matrix" contains.
· "$norm_one = $matrix->norm_one();"
Returns the "one"-norm of the given matrix "$matrix".
The "one"-norm is defined as follows:
For each column, the sum of the absolute values of the elements in
the different rows of that column is calculated. Finally, the
maximum of these sums is returned.
Note that the "one"-norm and the "maximum"-norm are mathematically
equivalent, although for the same matrix they usually yield a
different value.
Therefore, you should only compare values that have been calculated
using the same norm!
Throughout this package, the "one"-norm is (arbitrarily) used for
all comparisons, for the sake of uniformity and comparability,
except for the iterative methods "solve_GSM()", "solve_SSM()" and
"solve_RM()" which use either norm depending on the matrix itself.
· "$norm_max = $matrix->norm_max();"
Returns the "maximum"-norm of the given matrix $matrix.
The "maximum"-norm is defined as follows:
For each row, the sum of the absolute values of the elements in the
different columns of that row is calculated. Finally, the maximum
of these sums is returned.
Note that the "maximum"-norm and the "one"-norm are mathematically
equivalent, although for the same matrix they usually yield a
different value.
Therefore, you should only compare values that have been calculated
using the same norm!
Throughout this package, the "one"-norm is (arbitrarily) used for
all comparisons, for the sake of uniformity and comparability,
except for the iterative methods "solve_GSM()", "solve_SSM()" and
"solve_RM()" which use either norm depending on the matrix itself.
· "$norm_sum = $matrix->norm_sum();"
This is a very simple norm which is defined as the sum of the
absolute values of every element.
· $p_norm = $matrix->norm_p($n);>
This function returns the "p-norm" of a vector. The argument $n
must be a number greater than or equal to 1 or the string "Inf".
The p-norm is defined as (sum(x_i^p))^(1/p). In words, it raised
each element to the p-th power, adds them up, and then takes the
p-th root of that number. If the string "Inf" is passed, the
"infinity-norm" is computed, which is really the limit of the
p-norm as p goes to infinity. It is defined as the maximum element
of the vector. Also, note that the familiar Euclidean distance
between two vectors is just a special case of a p-norm, when p is
equal to 2.
Example:
$a = Math::MatrixReal->new_from_cols([[1,2,3]]);
$p1 = $a->norm_p(1);
$p2 = $a->norm_p(2);
$p3 = $a->norm_p(3);
$pinf = $a->norm_p("Inf");
print "(1,2,3,Inf) norm:\n$p1\n$p2\n$p3\n$pinf\n";
$i1 = $a->new_from_rows([[1,0]]);
$i2 = $a->new_from_rows([[0,1]]);
# this should be sqrt(2) since it is the same as the
# hypotenuse of a 1 by 1 right triangle
$dist = ($i1-$i2)->norm_p(2);
print "Distance is $dist, which should be " . sqrt(2) . "\n";
Output:
(1,2,3,Inf) norm:
6
3.74165738677394139
3.30192724889462668
3
Distance is 1.41421356237309505, which should be 1.41421356237309505
· $frob_norm = "$matrix->norm_frobenius();"
This norm is similar to that of a p-norm where p is 2, except it
acts on a matrix, not a vector. Each element of the matrix is
squared, this is added up, and then a square root is taken.
· "$matrix->spectral_radius();"
Returns the maximum value of the absolute value of all eigenvalues.
Currently this computes all eigenvalues, then sifts through them to
find the largest in absolute value. Needless to say, this is very
inefficient, and in the future an algorithm that computes only the
largest eigenvalue may be implemented.
· "$matrix1->transpose($matrix2);"
Calculates the transposed matrix of matrix $matrix2 and stores the
result in matrix "$matrix1" (which must already exist and have the
same size as matrix "$matrix2"!).
This operation can also be carried out "in-place", i.e., input and
output matrix may be identical.
Transposition is a symmetry operation: imagine you rotate the
matrix along the axis of its main diagonal (going through elements
(1,1), (2,2), (3,3) and so on) by 180 degrees.
Another way of looking at it is to say that rows and columns are
swapped. In fact the contents of element "(i,j)" are swapped with
those of element "(j,i)".
Note that (especially for vectors) it makes a big difference if you
have a row vector, like this:
[ -1 0 1 ]
or a column vector, like this:
[ -1 ]
[ 0 ]
[ 1 ]
the one vector being the transposed of the other!
This is especially true for the matrix product of two vectors:
[ -1 ]
[ -1 0 1 ] * [ 0 ] = [ 2 ] , whereas
[ 1 ]
* [ -1 0 1 ]
[ -1 ] [ 1 0 -1 ]
[ 0 ] * [ -1 0 1 ] = [ -1 ] [ 1 0 -1 ] = [ 0 0 0 ]
[ 1 ] [ 0 ] [ 0 0 0 ] [ -1 0 1 ]
[ 1 ] [ -1 0 1 ]
So be careful about what you really mean!
Hint: throughout this module, whenever a vector is explicitly
required for input, a COLUMN vector is expected!
· "$trace = $matrix->trace();"
This returns the trace of the matrix, which is defined as the sum
of the diagonal elements. The matrix must be quadratic.
· "$minor = $matrix->minor($row,$col);"
Returns the minor matrix corresponding to $row and $col. $matrix
must be quadratic. If $matrix is n rows by n cols, the minor of
$row and $col will be an (n-1) by (n-1) matrix. The minor is
defined as crossing out the row and the col specified and returning
the remaining rows and columns as a matrix. This method is used by
"cofactor()".
· "$cofactor = $matrix->cofactor();"
The cofactor matrix is constructed as follows:
For each element, cross out the row and column that it sits in.
Now, take the determinant of the matrix that is left in the other
rows and columns. Multiply the determinant by (-1)^(i+j), where i
is the row index, and j is the column index. Replace the given
element with this value.
The cofactor matrix can be used to find the inverse of the matrix.
One formula for the inverse of a matrix is the cofactor matrix
transposed divided by the original determinant of the matrix.
The following two inverses should be exactly the same:
my $inverse1 = $matrix->inverse;
my $inverse2 = ~($matrix->cofactor)->each( sub { (shift)/$matrix->det() } );
Caveat: Although the cofactor matrix is simple algorithm to compute
the inverse of a matrix, and can be used with pencil and paper for
small matrices, it is comically slower than the native "inverse()"
function. Here is a small benchmark:
# $matrix1 is 15x15
$det = $matrix1->det;
timethese( 10,
{'inverse' => sub { $matrix1->inverse(); },
'cofactor' => sub { (~$matrix1->cofactor)->each ( sub { (shift)/$det; } ) }
} );
Benchmark: timing 10 iterations of LR, cofactor, inverse...
inverse: 1 wallclock secs ( 0.56 usr + 0.00 sys = 0.56 CPU) @ 17.86/s (n=10)
cofactor: 36 wallclock secs (36.62 usr + 0.01 sys = 36.63 CPU) @ 0.27/s (n=10)
· "$adjoint = $matrix->adjoint();"
The adjoint is just the transpose of the cofactor matrix. This
method is just an alias for " ~($matrix->cofactor)".
Arithmetic Operations
· "$matrix1->add($matrix2,$matrix3);"
Calculates the sum of matrix "$matrix2" and matrix "$matrix3" and
stores the result in matrix "$matrix1" (which must already exist
and have the same size as matrix "$matrix2" and matrix
"$matrix3"!).
This operation can also be carried out "in-place", i.e., the output
and one (or both) of the input matrices may be identical.
· "$matrix1->subtract($matrix2,$matrix3);"
Calculates the difference of matrix "$matrix2" minus matrix
"$matrix3" and stores the result in matrix "$matrix1" (which must
already exist and have the same size as matrix "$matrix2" and
matrix "$matrix3"!).
This operation can also be carried out "in-place", i.e., the output
and one (or both) of the input matrices may be identical.
Note that this operation is the same as
"$matrix1->add($matrix2,-$matrix3);", although the latter is a
little less efficient.
· "$matrix1->multiply_scalar($matrix2,$scalar);"
Calculates the product of matrix "$matrix2" and the number
"$scalar" (i.e., multiplies each element of matrix "$matrix2" with
the factor "$scalar") and stores the result in matrix "$matrix1"
(which must already exist and have the same size as matrix
"$matrix2"!).
This operation can also be carried out "in-place", i.e., input and
output matrix may be identical.
· "$product_matrix = $matrix1->multiply($matrix2);"
Calculates the product of matrix "$matrix1" and matrix "$matrix2"
and returns an object reference to a new matrix "$product_matrix"
in which the result of this operation has been stored.
Note that the dimensions of the two matrices "$matrix1" and
"$matrix2" (i.e., their numbers of rows and columns) must harmonize
in the following way (example):
[ 2 2 ]
[ 2 2 ]
[ 2 2 ]
[ 1 1 1 ] [ * * ]
[ 1 1 1 ] [ * * ]
[ 1 1 1 ] [ * * ]
[ 1 1 1 ] [ * * ]
I.e., the number of columns of matrix "$matrix1" has to be the same
as the number of rows of matrix "$matrix2".
The number of rows and columns of the resulting matrix
"$product_matrix" is determined by the number of rows of matrix
"$matrix1" and the number of columns of matrix "$matrix2",
respectively.
· "$matrix1->negate($matrix2);"
Calculates the negative of matrix "$matrix2" (i.e., multiplies all
elements with "-1") and stores the result in matrix "$matrix1"
(which must already exist and have the same size as matrix
"$matrix2"!).
This operation can also be carried out "in-place", i.e., input and
output matrix may be identical.
· "$matrix_to_power = $matrix1->exponent($integer);"
Raises the matrix to the $integer power. Obviously, $integer must
be an integer. If it is zero, the identity matrix is returned. If a
negative integer is given, the inverse will be computed (if it
exists) and then raised the the absolute value of $integer. The
matrix must be quadratic.
Boolean Matrix Operations
· "$matrix->is_quadratic();"
Returns a boolean value indicating if the given matrix is quadratic
(also know as "square" or "n by n"). A matrix is quadratic if it
has the same number of rows as it does columns.
· "$matrix->is_square();"
This is an alias for "is_quadratic()".
· "$matrix->is_symmetric();"
Returns a boolean value indicating if the given matrix is
symmetric. By definition, a matrix is symmetric if and only if
(M[i,j]=M[j,i]). This is equivalent to "($matrix == ~$matrix)" but
without memory allocation. Only quadratic matrices can be
symmetric.
Notes: A symmetric matrix always has real eigenvalues/eigenvectors.
A matrix plus its transpose is always symmetric.
· "$matrix->is_skew_symmetric();"
Returns a boolean value indicating if the given matrix is skew
symmetric. By definition, a matrix is symmetric if and only if
(M[i,j]=-M[j,i]). This is equivalent to "($matrix == -(~$matrix))"
but without memory allocation. Only quadratic matrices can be skew
symmetric.
· "$matrix->is_diagonal();"
Returns a boolean value indicating if the given matrix is diagonal,
i.e. all of the nonzero elements are on the main diagonal. Only
quadratic matrices can be diagonal.
· "$matrix->is_tridiagonal();"
Returns a boolean value indicating if the given matrix is
tridiagonal, i.e. all of the nonzero elements are on the main
diagonal or the diagonals above and below the main diagonal. Only
quadratic matrices can be tridiagonal.
· "$matrix->is_upper_triangular();"
Returns a boolean value indicating if the given matrix is upper
triangular, i.e. all of the nonzero elements not on the main
diagonal are above it. Only quadratic matrices can be upper
triangular. Note: diagonal matrices are both upper and lower
triangular.
· "$matrix->is_lower_triangular();"
Returns a boolean value indicating if the given matrix is lower
triangular, i.e. all of the nonzero elements not on the main
diagonal are below it. Only quadratic matrices can be lower
triangular. Note: diagonal matrices are both upper and lower
triangular.
· "$matrix->is_orthogonal();"
Returns a boolean value indicating if the given matrix is
orthogonal. An orthogonal matrix is has the property that the
transpose equals the inverse of the matrix. Instead of computing
each and comparing them, this method multiplies the matrix by it's
transpose, and returns true if this turns out to be the identity
matrix, false otherwise. Only quadratic matrices can orthogonal.
· "$matrix->is_binary();"
Returns a boolean value indicating if the given matrix is binary.
A matrix is binary if it contains only zeroes or ones.
· "$matrix->is_gramian();"
Returns a boolean value indicating if the give matrix is Gramian.
A matrix $A is Gramian if and only if there exists a square matrix
$B such that "$A = ~$B*$B". This is equivalent to checking if $A is
symmetric and has all nonnegative eigenvalues, which is what
Math::MatrixReal uses to check for this property.
· "$matrix->is_LR();"
Returns a boolean value indicating if the matrix is an LR
decomposition matrix.
· "$matrix->is_positive();"
Returns a boolean value indicating if the matrix contains only
positive entries. Note that a zero entry is not positive and will
cause "is_positive()" to return false.
· "$matrix->is_negative();"
Returns a boolean value indicating if the matrix contains only
negative entries. Note that a zero entry is not negative and will
cause "is_negative()" to return false.
· "$matrix->is_periodic($k);"
Returns a boolean value indicating if the matrix is periodic with
period $k. This is true if "$matrix ** ($k+1) == $matrix". When
"$k == 1", this reduces down to the "is_idempotent()" function.
· "$matrix->is_idempotent();"
Returns a boolean value indicating if the matrix is idempotent,
which is defined as the square of the matrix being equal to the
original matrix, i.e "$matrix ** 2 == $matrix".
· "$matrix->is_row_vector();"
Returns a boolean value indicating if the matrix is a row vector.
A row vector is a matrix which is 1xn. Note that the 1x1 matrix is
both a row and column vector.
· "$matrix->is_col_vector();"
Returns a boolean value indicating if the matrix is a col vector.
A col vector is a matrix which is nx1. Note that the 1x1 matrix is
both a row and column vector.
Eigensystems
· "($l, $V) = $matrix->sym_diagonalize();"
This method performs the diagonalization of the quadratic symmetric
matrix M stored in $matrix. On output, l is a column vector
containing all the eigenvalues of M and V is an orthogonal matrix
which columns are the corresponding normalized eigenvectors. The
primary property of an eigenvalue l and an eigenvector x is of course
that: M * x = l * x.
The method uses a Householder reduction to tridiagonal form followed
by a QL algoritm with implicit shifts on this tridiagonal. (The
tridiagonal matrix is kept internally in a compact form in this
routine to save memory.) In fact, this routine wraps the
householder() and tri_diagonalize() methods described below when
their intermediate results are not desired. The overall algorithmic
complexity of this technique is O(N^3). According to several books,
the coefficient hidden by the 'O' is one of the best possible for
general (symmetric) matrixes.
· "($T, $Q) = $matrix->householder();"
This method performs the Householder algorithm which reduces the n by
n real symmetric matrix M contained in $matrix to tridiagonal form.
On output, T is a symmetric tridiagonal matrix (only diagonal and
off-diagonal elements are non-zero) and Q is an orthogonal matrix
performing the tranformation between M and T ("$M == $Q * $T * ~$Q").
· "($l, $V) = $T->tri_diagonalize([$Q]);"
This method diagonalizes the symmetric tridiagonal matrix T. On
output, $l and $V are similar to the output values described for
sym_diagonalize().
The optional argument $Q corresponds to an orthogonal transformation
matrix Q that should be used additionally during V (eigenvectors)
computation. It should be supplied if the desired eigenvectors
correspond to a more general symmetric matrix M previously reduced by
the householder() method, not a mere tridiagonal. If T is really a
tridiagonal matrix, Q can be omitted (it will be internally created
in fact as an identity matrix). The method uses a QL algorithm (with
implicit shifts).
· "$l = $matrix->sym_eigenvalues();"
This method computes the eigenvalues of the quadratic symmetric
matrix M stored in $matrix. On output, l is a column vector
containing all the eigenvalues of M. Eigenvectors are not computed
(on the contrary of "sym_diagonalize()") and this method is more
efficient (even though it uses a similar algorithm with two phases).
However, understand that the algorithmic complexity of this technique
is still also O(N^3). But the coefficient hidden by the 'O' is better
by a factor of..., well, see your benchmark, it's wiser.
This routine wraps the householder_tridiagonal() and
tri_eigenvalues() methods described below when the intermediate
tridiagonal matrix is not needed.
· "$T = $matrix->householder_tridiagonal();"
This method performs the Householder algorithm which reduces the n by
n real symmetric matrix M contained in $matrix to tridiagonal form.
On output, T is the obtained symmetric tridiagonal matrix (only
diagonal and off-diagonal elements are non-zero). The operation is
similar to the householder() method, but potentially a little more
efficient as the transformation matrix is not computed.
· $l = $T->tri_eigenvalues();
This method computesthe eigenvalues of the symmetric tridiagonal
matrix T. On output, $l is a vector containing the eigenvalues
(similar to "sym_eigenvalues()"). This method is much more efficient
than tri_diagonalize() when eigenvectors are not needed.
Miscellaneous
· $matrix->zero();
Assigns a zero to every element of the matrix "$matrix", i.e.,
erases all values previously stored there, thereby effectively
transforming the matrix into a "zero"-matrix or "null"-matrix, the
neutral element of the addition operation in a Ring.
(For instance the (quadratic) matrices with "n" rows and columns
and matrix addition and multiplication form a Ring. Most prominent
characteristic of a Ring is that multiplication is not commutative,
i.e., in general, ""matrix1 * matrix2"" is not the same as
""matrix2 * matrix1""!)
· $matrix->one();
Assigns one's to the elements on the main diagonal (elements (1,1),
(2,2), (3,3) and so on) of matrix "$matrix" and zero's to all
others, thereby erasing all values previously stored there and
transforming the matrix into a "one"-matrix, the neutral element of
the multiplication operation in a Ring.
(If the matrix is quadratic (which this method doesn't require,
though), then multiplying this matrix with itself yields this same
matrix again, and multiplying it with some other matrix leaves that
other matrix unchanged!)
· "$latex_string = $matrix->as_latex( align=> "c", format => "%s",
name => "" );"
This function returns the matrix as a LaTeX string. It takes a hash
as an argument which is used to control the style of the output.
The hash element "align" may be "c","l" or "r", corresponding to
center, left and right, respectively. The "format" element is a
format string that is given to "sprintf" to control the style of
number format, such a floating point or scientific notation. The
"name" element can be used so that a LaTeX string of "$name = " is
prepended to the string.
Example:
my $a = Math::MatrixReal->new_from_cols([[ 1.234, 5.678, 9.1011],[1,2,3]] );
print $a->as_latex( ( format => "%.2f", align => "l",name => "A" ) );
Output:
$A = $ $
\left( \begin{array}{ll}
1.23&1.00 \\
5.68&2.00 \\
9.10&3.00
\end{array} \right)
$
· "$yacas_string = $matrix->as_yacas( format => "%s", name => "",
semi => 0 );"
This function returns the matrix as a string that can be read by
Yacas. It takes a hash as an an argument which controls the style
of the output. The "format" element is a format string that is
given to "sprintf" to control the style of number format, such a
floating point or scientific notation. The "name" element can be
used so that "$name = " is prepended to the string. The <semi>
element can be set to 1 to that a semicolon is appended (so Matlab
does not print out the matrix.)
Example:
$a = Math::MatrixReal->new_from_cols([[ 1.234, 5.678, 9.1011],[1,2,3]] );
print $a->as_yacas( ( format => "%.2f", align => "l",name => "A" ) );
Output:
A := {{1.23,1.00},{5.68,2.00},{9.10,3.00}}
· "$matlab_string = $matrix->as_matlab( format => "%s", name => "",
semi => 0 );"
This function returns the matrix as a string that can be read by
Matlab. It takes a hash as an an argument which controls the style
of the output. The "format" element is a format string that is
given to "sprintf" to control the style of number format, such a
floating point or scientific notation. The "name" element can be
used so that "$name = " is prepended to the string. The <semi>
element can be set to 1 to that a semicolon is appended (so Matlab
does not print out the matrix.)
Example:
my $a = Math::MatrixReal->new_from_rows([[ 1.234, 5.678, 9.1011],[1,2,3]] );
print $a->as_matlab( ( format => "%.3f", name => "A",semi => 1 ) );
Output:
A = [ 1.234 5.678 9.101;
1.000 2.000 3.000];
· "$scilab_string = $matrix->as_scilab( format => "%s", name => "",
semi => 0 );"
This function is just an alias for "as_matlab()", since both Scilab
and Matlab have the same matrix format.
· "$minimum = Math::MatrixReal::min($number1,$number2);" "$minimum =
Math::MatrixReal::min($matrix);" "<$minimum = $matrix-"min;>>
Returns the minimum of the two numbers ""number1"" and ""number2""
if called with two arguments, or returns the value of the smallest
element of a matrix if called with one argument or as an object
method.
· "$maximum = Math::MatrixReal::max($number1,$number2);" "$maximum =
Math::MatrixReal::max($number1,$number2);" "$maximum =
Math::MatrixReal::max($matrix);" "<$maximum = $matrix-"max;>>
Returns the maximum of the two numbers ""number1"" and ""number2""
if called with two arguments, or returns the value of the largest
element of a matrix if called with one arguemnt or as on object
method.
· "$minimal_cost_matrix = $cost_matrix->kleene();"
Copies the matrix "$cost_matrix" (which has to be quadratic!) to a
new matrix of the same size (i.e., "clones" the input matrix) and
applies Kleene's algorithm to it.
See Math::Kleene(3) for more details about this algorithm!
The method returns an object reference to the new matrix.
Matrix "$cost_matrix" is not changed by this method in any way.
· "($norm_matrix,$norm_vector) = $matrix->normalize($vector);"
This method is used to improve the numerical stability when solving
linear equation systems.
Suppose you have a matrix "A" and a vector "b" and you want to find
out a vector "x" so that "A * x = b", i.e., the vector "x" which
solves the equation system represented by the matrix "A" and the
vector "b".
Applying this method to the pair (A,b) yields a pair (A',b') where
each row has been divided by (the absolute value of) the greatest
coefficient appearing in that row. So this coefficient becomes
equal to "1" (or "-1") in the new pair (A',b') (all others become
smaller than one and greater than minus one).
Note that this operation does not change the equation system itself
because the same division is carried out on either side of the
equation sign!
The method requires a quadratic (!) matrix "$matrix" and a vector
"$vector" for input (the vector must be a column vector with the
same number of rows as the input matrix) and returns a list of two
items which are object references to a new matrix and a new vector,
in this order.
The output matrix and vector are clones of the input matrix and
vector to which the operation explained above has been applied.
The input matrix and vector are not changed by this in any way.
Example of how this method can affect the result of the methods to
solve equation systems (explained immediately below following this
method):
Consider the following little program:
#!perl -w
use Math::MatrixRealqw(new_from_string);
$A = Math::MatrixReal->new_from_string(<<"MATRIX");
[ 1 2 3 ]
[ 5 7 11 ]
[ 23 19 13 ]
MATRIX
$b = Math::MatrixReal->new_from_string(<<"MATRIX");
[ 0 ]
[ 1 ]
[ 29 ]
MATRIX
$LR = $A->decompose_LR();
if (($dim,$x,$B) = $LR->solve_LR($b))
{
$test = $A * $x;
print "x = \n$x";
print "A * x = \n$test";
}
($A_,$b_) = $A->normalize($b);
$LR = $A_->decompose_LR();
if (($dim,$x,$B) = $LR->solve_LR($b_))
{
$test = $A * $x;
print "x = \n$x";
print "A * x = \n$test";
}
This will print:
x =
[ 1.000000000000E+00 ]
[ 1.000000000000E+00 ]
[ -1.000000000000E+00 ]
A * x =
[ 4.440892098501E-16 ]
[ 1.000000000000E+00 ]
[ 2.900000000000E+01 ]
x =
[ 1.000000000000E+00 ]
[ 1.000000000000E+00 ]
[ -1.000000000000E+00 ]
A * x =
[ 0.000000000000E+00 ]
[ 1.000000000000E+00 ]
[ 2.900000000000E+01 ]
You can see that in the second example (where "normalize()" has
been used), the result is "better", i.e., more accurate!
· "$LR_matrix = $matrix->decompose_LR();"
This method is needed to solve linear equation systems.
Suppose you have a matrix "A" and a vector "b" and you want to find
out a vector "x" so that "A * x = b", i.e., the vector "x" which
solves the equation system represented by the matrix "A" and the
vector "b".
You might also have a matrix "A" and a whole bunch of different
vectors "b1".."bk" for which you need to find vectors "x1".."xk" so
that "A * xi = bi", for "i=1..k".
Using Gaussian transformations (multiplying a row or column with a
factor, swapping two rows or two columns and adding a multiple of
one row or column to another), it is possible to decompose any
matrix "A" into two triangular matrices, called "L" and "R" (for
"Left" and "Right").
"L" has one's on the main diagonal (the elements (1,1), (2,2),
(3,3) and so so), non-zero values to the left and below of the main
diagonal and all zero's in the upper right half of the matrix.
"R" has non-zero values on the main diagonal as well as to the
right and above of the main diagonal and all zero's in the lower
left half of the matrix, as follows:
[ 1 0 0 0 0 ] [ x x x x x ]
[ x 1 0 0 0 ] [ 0 x x x x ]
L = [ x x 1 0 0 ] R = [ 0 0 x x x ]
[ x x x 1 0 ] [ 0 0 0 x x ]
[ x x x x 1 ] [ 0 0 0 0 x ]
Note that ""L * R"" is equivalent to matrix "A" in the sense that
"L * R * x = b <==> A * x = b" for all vectors "x", leaving out
of account permutations of the rows and columns (these are taken
care of "magically" by this module!) and numerical errors.
Trick:
Because we know that "L" has one's on its main diagonal, we can
store both matrices together in the same array without information
loss! I.e.,
[ R R R R R ]
[ L R R R R ]
LR = [ L L R R R ]
[ L L L R R ]
[ L L L L R ]
Beware, though, that "LR" and ""L * R"" are not the same!!!
Note also that for the same reason, you cannot apply the method
"normalize()" to an "LR" decomposition matrix. Trying to do so will
yield meaningless rubbish!
(You need to apply "normalize()" to each pair (Ai,bi) BEFORE
decomposing the matrix "Ai'"!)
Now what does all this help us in solving linear equation systems?
It helps us because a triangular matrix is the next best thing that
can happen to us besides a diagonal matrix (a matrix that has non-
zero values only on its main diagonal - in which case the solution
is trivial, simply divide ""b[i]"" by ""A[i,i]"" to get ""x[i]""!).
To find the solution to our problem ""A * x = b"", we divide this
problem in parts: instead of solving "A * x = b" directly, we first
decompose "A" into "L" and "R" and then solve ""L * y = b"" and
finally ""R * x = y"" (motto: divide and rule!).
From the illustration above it is clear that solving ""L * y = b""
and ""R * x = y"" is straightforward: we immediately know that
"y[1] = b[1]". We then deduce swiftly that
y[2] = b[2] - L[2,1] * y[1]
(and we know ""y[1]"" by now!), that
y[3] = b[3] - L[3,1] * y[1] - L[3,2] * y[2]
and so on.
Having effortlessly calculated the vector "y", we now proceed to
calculate the vector "x" in a similar fashion: we see immediately
that "x[n] = y[n] / R[n,n]". It follows that
x[n-1] = ( y[n-1] - R[n-1,n] * x[n] ) / R[n-1,n-1]
and
x[n-2] = ( y[n-2] - R[n-2,n-1] * x[n-1] - R[n-2,n] * x[n] )
/ R[n-2,n-2]
and so on.
You can see that - especially when you have many vectors "b1".."bk"
for which you are searching solutions to "A * xi = bi" - this
scheme is much more efficient than a straightforward, "brute force"
approach.
This method requires a quadratic matrix as its input matrix.
If you don't have that many equations, fill up with zero's (i.e.,
do nothing to fill the superfluous rows if it's a "fresh" matrix,
i.e., a matrix that has been created with "new()" or "shadow()").
The method returns an object reference to a new matrix containing
the matrices "L" and "R".
The input matrix is not changed by this method in any way.
Note that you can "copy()" or "clone()" the result of this method
without losing its "magical" properties (for instance concerning
the hidden permutations of its rows and columns).
However, as soon as you are applying any method that alters the
contents of the matrix, its "magical" properties are stripped off,
and the matrix immediately reverts to an "ordinary" matrix (with
the values it just happens to contain at that moment, be they
meaningful as an ordinary matrix or not!).
· "($dimension,$x_vector,$base_matrix) =
$LR_matrix""->""solve_LR($b_vector);"
Use this method to actually solve an equation system.
Matrix "$LR_matrix" must be a (quadratic) matrix returned by the
method "decompose_LR()", the LR decomposition matrix of the matrix
"A" of your equation system "A * x = b".
The input vector "$b_vector" is the vector "b" in your equation
system "A * x = b", which must be a column vector and have the same
number of rows as the input matrix "$LR_matrix".
The method returns a list of three items if a solution exists or an
empty list otherwise (!).
Therefore, you should always use this method like this:
if ( ($dim,$x_vec,$base) = $LR->solve_LR($b_vec) )
{
# do something with the solution...
}
else
{
# do something with the fact that there is no solution...
}
The three items returned are: the dimension "$dimension" of the
solution space (which is zero if only one solution exists, one if
the solution is a straight line, two if the solution is a plane,
and so on), the solution vector "$x_vector" (which is the vector
"x" of your equation system "A * x = b") and a matrix
"$base_matrix" representing a base of the solution space (a set of
vectors which put up the solution space like the spokes of an
umbrella).
Only the first "$dimension" columns of this base matrix actually
contain entries, the remaining columns are all zero.
Now what is all this stuff with that "base" good for?
The output vector "x" is ALWAYS a solution of your equation system
"A * x = b".
But also any vector "$vector"
$vector = $x_vector->clone();
$machine_infinity = 1E+99; # or something like that
for ( $i = 1; $i <= $dimension; $i++ )
{
$vector += rand($machine_infinity) * $base_matrix->column($i);
}
is a solution to your problem "A * x = b", i.e., if "$A_matrix"
contains your matrix "A", then
print abs( $A_matrix * $vector - $b_vector ), "\n";
should print a number around 1E-16 or so!
By the way, note that you can actually calculate those vectors
"$vector" a little more efficient as follows:
$rand_vector = $x_vector->shadow();
$machine_infinity = 1E+99; # or something like that
for ( $i = 1; $i <= $dimension; $i++ )
{
$rand_vector->assign($i,1, rand($machine_infinity) );
}
$vector = $x_vector + ( $base_matrix * $rand_vector );
Note that the input matrix and vector are not changed by this
method in any way.
· "$inverse_matrix = $LR_matrix->invert_LR();"
Use this method to calculate the inverse of a given matrix
"$LR_matrix", which must be a (quadratic) matrix returned by the
method "decompose_LR()".
The method returns an object reference to a new matrix of the same
size as the input matrix containing the inverse of the matrix that
you initially fed into "decompose_LR()" IF THE INVERSE EXISTS, or
an empty list otherwise.
Therefore, you should always use this method in the following way:
if ( $inverse_matrix = $LR->invert_LR() )
{
# do something with the inverse matrix...
}
else
{
# do something with the fact that there is no inverse matrix...
}
Note that by definition (disregarding numerical errors), the
product of the initial matrix and its inverse (or vice-versa) is
always a matrix containing one's on the main diagonal (elements
(1,1), (2,2), (3,3) and so on) and zero's elsewhere.
The input matrix is not changed by this method in any way.
· "$condition = $matrix->condition($inverse_matrix);"
In fact this method is just a shortcut for
abs($matrix) * abs($inverse_matrix)
Both input matrices must be quadratic and have the same size, and
the result is meaningful only if one of them is the inverse of the
other (for instance, as returned by the method "invert_LR()").
The number returned is a measure of the "condition" of the given
matrix "$matrix", i.e., a measure of the numerical stability of the
matrix.
This number is always positive, and the smaller its value, the
better the condition of the matrix (the better the stability of all
subsequent computations carried out using this matrix).
Numerical stability means for example that if
abs( $vec_correct - $vec_with_error ) < $epsilon
holds, there must be a "$delta" which doesn't depend on the vector
"$vec_correct" (nor "$vec_with_error", by the way) so that
abs( $matrix * $vec_correct - $matrix * $vec_with_error ) < $delta
also holds.
· "$determinant = $LR_matrix->det_LR();"
Calculates the determinant of a matrix, whose LR decomposition
matrix "$LR_matrix" must be given (which must be a (quadratic)
matrix returned by the method "decompose_LR()").
In fact the determinant is a by-product of the LR decomposition: It
is (in principle, that is, except for the sign) simply the product
of the elements on the main diagonal (elements (1,1), (2,2), (3,3)
and so on) of the LR decomposition matrix.
(The sign is taken care of "magically" by this module)
· "$order = $LR_matrix->order_LR();"
Calculates the order (called "Rang" in German) of a matrix, whose
LR decomposition matrix "$LR_matrix" must be given (which must be a
(quadratic) matrix returned by the method "decompose_LR()").
This number is a measure of the number of linear independent row
and column vectors (= number of linear independent equations in the
case of a matrix representing an equation system) of the matrix
that was initially fed into "decompose_LR()".
If "n" is the number of rows and columns of the (quadratic!)
matrix, then "n - order" is the dimension of the solution space of
the associated equation system.
· "$rank = $LR_matrix->rank_LR();"
This is an alias for the "order_LR()" function. The "order" is
usually called the "rank" in the United States.
· "$scalar_product = $vector1->scalar_product($vector2);"
Returns the scalar product of vector "$vector1" and vector
"$vector2".
Both vectors must be column vectors (i.e., a matrix having several
rows but only one column).
This is a (more efficient!) shortcut for
$temp = ~$vector1 * $vector2;
$scalar_product = $temp->element(1,1);
or the sum "i=1..n" of the products "vector1[i] * vector2[i]".
Provided none of the two input vectors is the null vector, then the
two vectors are orthogonal, i.e., have an angle of 90 degrees
between them, exactly when their scalar product is zero, and vice-
versa.
· "$vector_product = $vector1->vector_product($vector2);"
Returns the vector product of vector "$vector1" and vector
"$vector2".
Both vectors must be column vectors (i.e., a matrix having several
rows but only one column).
Currently, the vector product is only defined for 3 dimensions
(i.e., vectors with 3 rows); all other vectors trigger an error
message.
In 3 dimensions, the vector product of two vectors "x" and "y" is
defined as
| x[1] y[1] e[1] |
determinant | x[2] y[2] e[2] |
| x[3] y[3] e[3] |
where the ""x[i]"" and ""y[i]"" are the components of the two
vectors "x" and "y", respectively, and the ""e[i]"" are unity
vectors (i.e., vectors with a length equal to one) with a one in
row "i" and zero's elsewhere (this means that you have numbers and
vectors as elements in this matrix!).
This determinant evaluates to the rather simple formula
z[1] = x[2] * y[3] - x[3] * y[2]
z[2] = x[3] * y[1] - x[1] * y[3]
z[3] = x[1] * y[2] - x[2] * y[1]
A characteristic property of the vector product is that the
resulting vector is orthogonal to both of the input vectors (if
neither of both is the null vector, otherwise this is trivial),
i.e., the scalar product of each of the input vectors with the
resulting vector is always zero.
· "$length = $vector->length();"
This is actually a shortcut for
$length = sqrt( $vector->scalar_product($vector) );
and returns the length of a given column or row vector "$vector".
Note that the "length" calculated by this method is in fact the
"two"-norm (also know as the Euclidean norm) of a vector "$vector"!
The general definition for norms of vectors is the following:
sub vector_norm
{
croak "Usage: \$norm = \$vector->vector_norm(\$n);"
if (@_ != 2);
my($vector,$n) = @_;
my($rows,$cols) = ($vector->[1],$vector->[2]);
my($k,$comp,$sum);
croak "Math::MatrixReal::vector_norm(): vector is not a column vector"
unless ($cols == 1);
croak "Math::MatrixReal::vector_norm(): norm index must be > 0"
unless ($n > 0);
croak "Math::MatrixReal::vector_norm(): norm index must be integer"
unless ($n == int($n));
$sum = 0;
for ( $k = 0; $k < $rows; $k++ )
{
$comp = abs( $vector->[0][$k][0] );
$sum += $comp ** $n;
}
return( $sum ** (1 / $n) );
}
Note that the case "n = 1" is the "one"-norm for matrices applied
to a vector, the case "n = 2" is the euclidian norm or length of a
vector, and if "n" goes to infinity, you have the "infinity"- or
"maximum"-norm for matrices applied to a vector!
· "$xn_vector = $matrix->""solve_GSM($x0_vector,$b_vector,$epsilon);"
· "$xn_vector = $matrix->""solve_SSM($x0_vector,$b_vector,$epsilon);"
· "$xn_vector =
$matrix->""solve_RM($x0_vector,$b_vector,$weight,$epsilon);"
In some cases it might not be practical or desirable to solve an
equation system ""A * x = b"" using an analytical algorithm like
the "decompose_LR()" and "solve_LR()" method pair.
In fact in some cases, due to the numerical properties (the
"condition") of the matrix "A", the numerical error of the obtained
result can be greater than by using an approximative (iterative)
algorithm like one of the three implemented here.
All three methods, GSM ("Global Step Method" or
"Gesamtschrittverfahren"), SSM ("Single Step Method" or
"Einzelschrittverfahren") and RM ("Relaxation Method" or
"Relaxationsverfahren"), are fix-point iterations, that is, can be
described by an iteration function ""x(t+1) = Phi( x(t) )"" which
has the property:
Phi(x) = x <==> A * x = b
We can define "Phi(x)" as follows:
Phi(x) := ( En - A ) * x + b
where "En" is a matrix of the same size as "A" ("n" rows and
columns) with one's on its main diagonal and zero's elsewhere.
This function has the required property.
Proof:
A * x = b
<==> -( A * x ) = -b
<==> -( A * x ) + x = -b + x
<==> -( A * x ) + x + b = x
<==> x - ( A * x ) + b = x
<==> ( En - A ) * x + b = x
This last step is true because
x[i] - ( a[i,1] x[1] + ... + a[i,i] x[i] + ... + a[i,n] x[n] ) + b[i]
is the same as
( -a[i,1] x[1] + ... + (1 - a[i,i]) x[i] + ... + -a[i,n] x[n] ) + b[i]
qed
Note that actually solving the equation system ""A * x = b"" means
to calculate
a[i,1] x[1] + ... + a[i,i] x[i] + ... + a[i,n] x[n] = b[i]
<==> a[i,i] x[i] =
b[i]
- ( a[i,1] x[1] + ... + a[i,i] x[i] + ... + a[i,n] x[n] )
+ a[i,i] x[i]
<==> x[i] =
( b[i]
- ( a[i,1] x[1] + ... + a[i,i] x[i] + ... + a[i,n] x[n] )
+ a[i,i] x[i]
) / a[i,i]
<==> x[i] =
( b[i] -
( a[i,1] x[1] + ... + a[i,i-1] x[i-1] +
a[i,i+1] x[i+1] + ... + a[i,n] x[n] )
) / a[i,i]
There is one major restriction, though: a fix-point iteration is
guaranteed to converge only if the first derivative of the
iteration function has an absolute value less than one in an area
around the point "x(*)" for which ""Phi( x(*) ) = x(*)"" is to be
true, and if the start vector "x(0)" lies within that area!
This is best verified graphically, which unfortunately is
impossible to do in this textual documentation!
See literature on Numerical Analysis for details!
In our case, this restriction translates to the following three
conditions:
There must exist a norm so that the norm of the matrix of the
iteration function, "( En - A )", has a value less than one, the
matrix "A" may not have any zero value on its main diagonal and the
initial vector "x(0)" must be "good enough", i.e., "close enough"
to the solution "x(*)".
(Remember school math: the first derivative of a straight line
given by ""y = a * x + b"" is "a"!)
The three methods expect a (quadratic!) matrix "$matrix" as their
first argument, a start vector "$x0_vector", a vector "$b_vector"
(which is the vector "b" in your equation system ""A * x = b""), in
the case of the "Relaxation Method" ("RM"), a real number "$weight"
best between zero and two, and finally an error limit (real number)
"$epsilon".
(Note that the weight "$weight" used by the "Relaxation Method"
("RM") is NOT checked to lie within any reasonable range!)
The three methods first test the first two conditions of the three
conditions listed above and return an empty list if these
conditions are not fulfilled.
Therefore, you should always test their return value using some
code like:
if ( $xn_vector = $A_matrix->solve_GSM($x0_vector,$b_vector,1E-12) )
{
# do something with the solution...
}
else
{
# do something with the fact that there is no solution...
}
Otherwise, they iterate until "abs( Phi(x) - x ) < epsilon".
(Beware that theoretically, infinite loops might result if the
starting vector is too far "off" the solution! In practice, this
shouldn't be a problem. Anyway, you can always press <ctrl-C> if
you think that the iteration takes too long!)
The difference between the three methods is the following:
In the "Global Step Method" ("GSM"), the new vector ""x(t+1)""
(called "y" here) is calculated from the vector "x(t)" (called "x"
here) according to the formula:
y[i] =
( b[i]
- ( a[i,1] x[1] + ... + a[i,i-1] x[i-1] +
a[i,i+1] x[i+1] + ... + a[i,n] x[n] )
) / a[i,i]
In the "Single Step Method" ("SSM"), the components of the vector
""x(t+1)"" which have already been calculated are used to calculate
the remaining components, i.e.
y[i] =
( b[i]
- ( a[i,1] y[1] + ... + a[i,i-1] y[i-1] + # note the "y[]"!
a[i,i+1] x[i+1] + ... + a[i,n] x[n] ) # note the "x[]"!
) / a[i,i]
In the "Relaxation method" ("RM"), the components of the vector
""x(t+1)"" are calculated by "mixing" old and new value (like cold
and hot water), and the weight "$weight" determines the "aperture"
of both the "hot water tap" as well as of the "cold water tap",
according to the formula:
y[i] =
( b[i]
- ( a[i,1] y[1] + ... + a[i,i-1] y[i-1] + # note the "y[]"!
a[i,i+1] x[i+1] + ... + a[i,n] x[n] ) # note the "x[]"!
) / a[i,i]
y[i] = weight * y[i] + (1 - weight) * x[i]
Note that the weight "$weight" should be greater than zero and less
than two (!).
The three methods are supposed to be of different efficiency.
Experiment!
Remember that in most cases, it is probably advantageous to first
"normalize()" your equation system prior to solving it!
OVERLOADED OPERATORS
SYNOPSIS
· Unary operators:
""-"", ""~"", ""abs"", "test", ""!"", '""'
· Binary operators:
"".""
Binary (arithmetic) operators:
""+"", ""-"", ""*"", ""**"", ""+="", ""-="", ""*="", ""/="",""**=""
· Binary (relational) operators:
""=="", ""!="", ""<"", ""<="", "">"", "">=""
""eq"", ""ne"", ""lt"", ""le"", ""gt"", ""ge""
Note that the latter (""eq"", ""ne"", ... ) are just synonyms of the
former (""=="", ""!="", ... ), defined for convenience only.
DESCRIPTION
'.' Concatenation
Returns the two matrices concatenated side by side.
Example: $c = $a . $b;
For example, if
$a=[ 1 2 ] $b=[ 5 6 ]
[ 3 4 ] [ 7 8 ]
then
$c=[ 1 2 5 6 ]
[ 3 4 7 8 ]
Note that only matrices with the same number of rows may be
concatenated.
'-' Unary minus
Returns the negative of the given matrix, i.e., the matrix with
all elements multiplied with the factor "-1".
Example:
$matrix = -$matrix;
'~' Transposition
Returns the transposed of the given matrix.
Examples:
$temp = ~$vector * $vector;
$length = sqrt( $temp->element(1,1) );
if (~$matrix == $matrix) { # matrix is symmetric ... }
abs Norm
Returns the "one"-Norm of the given matrix.
Example:
$error = abs( $A * $x - $b );
test Boolean test
Tests wether there is at least one non-zero element in the matrix.
Example:
if ($xn_vector) { # result of iteration is not zero ... }
'!' Negated boolean test
Tests wether the matrix contains only zero's.
Examples:
if (! $b_vector) { # heterogenous equation system ... }
else { # homogenous equation system ... }
unless ($x_vector) { # not the null-vector! }
'""""'
"Stringify" operator
Converts the given matrix into a string.
Uses scientific representation to keep precision loss to a minimum
in case you want to read this string back in again later with
"new_from_string()".
By default a 13-digit mantissa and a 20-character field for each
element is used so that lines will wrap nicely on an 80-column
screen.
Examples:
$matrix = Math::MatrixReal->new_from_string(<<"MATRIX");
[ 1 0 ]
[ 0 -1 ]
MATRIX
print "$matrix";
[ 1.000000000000E+00 0.000000000000E+00 ]
[ 0.000000000000E+00 -1.000000000000E+00 ]
$string = "$matrix";
$test = Math::MatrixReal->new_from_string($string);
if ($test == $matrix) { print ":-)\n"; } else { print ":-(\n"; }
'+' Addition
Returns the sum of the two given matrices.
Examples:
$matrix_S = $matrix_A + $matrix_B;
$matrix_A += $matrix_B;
'-' Subtraction
Returns the difference of the two given matrices.
Examples:
$matrix_D = $matrix_A - $matrix_B;
$matrix_A -= $matrix_B;
Note that this is the same as:
$matrix_S = $matrix_A + -$matrix_B;
$matrix_A += -$matrix_B;
(The latter are less efficient, though)
'*' Multiplication
Returns the matrix product of the two given matrices or the
product of the given matrix and scalar factor.
Examples:
$matrix_P = $matrix_A * $matrix_B;
$matrix_A *= $matrix_B;
$vector_b = $matrix_A * $vector_x;
$matrix_B = -1 * $matrix_A;
$matrix_B = $matrix_A * -1;
$matrix_A *= -1;
'/' Division
Currently a shortcut for doing $a * $b ** -1 is $a / $b, which
works for square matrices. One can also use 1/$a .
'**' Exponentiation
Returns the matrix raised to an integer power. If 0 is passed, the
identity matrix is returned. If a negative integer is passed, it
computes the inverse (if it exists) and then raised the inverse to
the absolute value of the integer. The matrix must be quadratic.
Examples:
$matrix2 = $matrix ** 2;
$matrix **= 2;
$inv2 = $matrix ** -2;
$ident = $matrix ** 0;
'==' Equality
Tests two matrices for equality.
Example:
if ( $A * $x == $b ) { print "EUREKA!\n"; }
Note that in most cases, due to numerical errors (due to the
finite precision of computer arithmetics), it is a bad idea to
compare two matrices or vectors this way.
Better use the norm of the difference of the two matrices you want
to compare and compare that norm with a small number, like this:
if ( abs( $A * $x - $b ) < 1E-12 ) { print "BINGO!\n"; }
'!=' Inequality
Tests two matrices for inequality.
Example:
while ($x0_vector != $xn_vector) { # proceed with iteration ... }
(Stops when the iteration becomes stationary)
Note that (just like with the '==' operator), it is usually a bad
idea to compare matrices or vectors this way. Compare the norm of
the difference of the two matrices with a small number instead.
'<' Less than
Examples:
if ( $matrix1 < $matrix2 ) { # ... }
if ( $vector < $epsilon ) { # ... }
if ( 1E-12 < $vector ) { # ... }
if ( $A * $x - $b < 1E-12 ) { # ... }
These are just shortcuts for saying:
if ( abs($matrix1) < abs($matrix2) ) { # ... }
if ( abs($vector) < abs($epsilon) ) { # ... }
if ( abs(1E-12) < abs($vector) ) { # ... }
if ( abs( $A * $x - $b ) < abs(1E-12) ) { # ... }
Uses the "one"-norm for matrices and Perl's built-in "abs()" for
scalars.
'<=' Less than or equal
As with the '<' operator, this is just a shortcut for the same
expression with "abs()" around all arguments.
Example:
if ( $A * $x - $b <= 1E-12 ) { # ... }
which in fact is the same as:
if ( abs( $A * $x - $b ) <= abs(1E-12) ) { # ... }
Uses the "one"-norm for matrices and Perl's built-in "abs()" for
scalars.
'>' Greater than
As with the '<' and '<=' operator, this
if ( $xn - $x0 > 1E-12 ) { # ... }
is just a shortcut for:
if ( abs( $xn - $x0 ) > abs(1E-12) ) { # ... }
Uses the "one"-norm for matrices and Perl's built-in "abs()" for
scalars.
'>=' Greater than or equal
As with the '<', '<=' and '>' operator, the following
if ( $LR >= $A ) { # ... }
is simply a shortcut for:
if ( abs($LR) >= abs($A) ) { # ... }
Uses the "one"-norm for matrices and Perl's built-in "abs()" for
scalars.
SEE ALSO
Math::VectorReal, Math::PARI, Math::MatrixBool, Math::Vec, DFA::Kleene,
Math::Kleene, Set::IntegerRange, Set::IntegerFast .
VERSION
This man page documents Math::MatrixReal version 2.08.
The latest code can be found at
https://github.com/leto/math--matrixreal .
AUTHORS
Steffen Beyer <sb@engelschall.com>, Rodolphe Ortalo <ortalo@laas.fr>,
Jonathan Leto <jonathan@leto.net>.
Currently maintained by Jonathan Leto, send all bugs/patches to me.
CREDITS
Many thanks to Prof. Pahlings for stoking the fire of my enthusiasm for
Algebra and Linear Algebra at the university (RWTH Aachen, Germany),
and to Prof. Esser and his assistant, Mr. Jarausch, for their
fascinating lectures in Numerical Analysis!
COPYRIGHT
Copyright (c) 1996-2011 by Steffen Beyer, Rodolphe Ortalo, Jonathan
Leto. All rights reserved.
LICENSE AGREEMENT
This package is free software; you can redistribute it and/or modify it
under the same terms as Perl itself.
perl v5.14.0 2011-06-17 Math::MatrixReal(3)