NetNews Usenet Archive 1992 #27/NN_1992_27.iso/spool/comp/sys/sgi/16634

Internet Message Format | 1992-11-18 | 10.6 KB

Path: sparky!uunet!ogicse!uwm.edu!zaphod.mps.ohio-state.edu!usc!news!benton.prepress.com!bob
From: [e-mail] (Robert Crowe)
Newsgroups: comp.sys.sgi
Subject: Re: matrix class in C++
Message-ID: <[e-mail]>
Date: 18 Nov 92 23:57:38 GMT
Article-I.D.: prepress.1992Nov18.235738.12736
References: <[e-mail]> <[e-mail]>
Organization: Pre-Press Technologies, Carlsbad Ca.
Lines: 292

In article <[e-mail]> [e-mail] (Dave Ciemiewicz) writes:
>In article <[e-mail]>, [e-mail] (Rose Oguz) writes:
>|> I haven't seen any mention of a matrix class in C++.  I have no information
>|> that tells me that there should be.  I just assume that matrices/2D arrays
>|> are commonly used and there would a class definition.  All I have found is
>|> a complex class and a vector class.  Any chance that I missed the matrix
>|> class?
>|>
>|> Rose
>|>
>
>Nope, there isn't one.  Some people out there may have written some but there
>isn't one that comes with the C++ compiler. The Vector and Complex classes
>are just traditional C++ classes that were created by AT&T to demonstrate
>classes as much as anything else.  It is pretty obvious how a Complex class
>can be used to demonstrate overloading of operators such as operator+().
>Notice that the term "vector" is use more in a data structure sense by the
>Vector classes than in the sense of a vector used in linear algebra.  I think
>the name was chosen to avoid confusion with intrinsic "array" structures.
>
>Writing your own Matrix class was left as an exercise to the user by
>Stroustrup and company.
>

There is in fact a package called newmat, which is a matrix package
written in c++.  I believe it has been posted to one of the sources
newsgroups, but I'm not certain which one. Instructions for obtaining
follow below.

Following is the top portion of the documentation that comes with the
package. I didn't include the whole thing because it is quite lengthy,
and contains details that the average net-reader won't be interested in.


Please note that I have nothing to do with the production of this
package, I just use it occasionally.

P.S.  Im not sure if this is from the most recent version of newmat.
      I think the current rev is 04, not 03.


--------- Top portion of newmat.txt follows -----------
//$$ newmat.txt            Documentation file


   Documentation for newmat03, an experimental matrix package in C++.
   ==================================================================


MATRIX PACKAGE                                           25 November, 1991

Copyright (C) 1991: R B Davies and DSIR

Permission is granted to use but not to sell.


Contents
========

General description
Is this the package you need?
Changes
Where you can get a copy of this package
Compiler performance
Example
Detailed documentation
   Customising
   Constructors
   Elements of matrices
   Matrix copy
   Unary operators
   Binary operators
   Combination of a matrix and scalar
   Scalar functions of matrices
   Submatrix operations
   Change dimensions
   Change type
   Multiple matrix solve
   Memory management
   Output
   Accessing matrices of unspecified type
   Cholesky decomposition
   Householder triangularisation
   Singular Value Decomposition
   Eigenvalues
   Sorting
   Fast Fourier Transform
   Interface to Numerical Recipes in C
List of files
Notes on the design of the package
   What this is package for
   What size of matrices?
   Allow matrix expressions?
   Which matrix types?
   What element types?
   Naming convention
   Row and Column index ranges
   Structure of matrix objects
   Data storage - one block or several
   Data storage - by row or by column or other
   Storage of symmetric matrices
   Element access - method and checking
   Use iterators?
   Memory management - reference counting or status variable?
   Evaluation of expressions - use two stage method?
   How to overcome an explosion in number of operations
   Using const
   A calculus of matrix types
   Error handling
   Band and sparse matrices
Problem report form


---------------------------------------------------------------------------


General description
===================

The package is intented for scientists and engineers who need to
manipulate a variety of types of matrices using standard matrix
operations. Emphasis is on the kind of operations needed in statistical
calculations such as least squares, linear equation solve and
eigenvalues.

It supports matrix types

    Matrix                       (rectangular matrix)
    nricMatrix                   (variant of rectangular matrix)
    UpperTriangularMatrix
    LowerTriangularMatrix
    DiagonalMatrix
    SymmetricMatrix
    RowVector                    (derived from Matrix)
    ColumnVector                 (derived from Matrix).

Only one element type (float or double) is supported.

The package includes the operations *, +, -, inverse, transpose,
conversion between types, submatrix, determinant, Cholesky
decomposition, Householder triangularisation, singular value
decomposition, eigenvalues of a symmetric matrix, sorting, fast fourier
transform, printing and an interface with "Numerical Recipes in C".

It is intended for matrices in the range 4 x 4 to about 90 x 90 (125 x
125 for triangular matrices). The upper limit is imposed by the maximum
number of elements that can be contained in a single array (8192 doubles
in some machines).

A two-stage approach to evaluating matrix expressions is used to improve
efficiency and reduce use of temporary storage.

The package is designed for version 2 of C++. It works with Turbo C++,
Borland C++, Glockenspiel C++ (2.00a) on a PC and AT&T C++ (2.0) and Gnu
C++ on a Sun. It works with some problems with Zortech C++ (version 2).


---------------------------------------------------------------------------


Is this the package you need?
=============================

Do you

1.   need matrix operators such as * and + defined as operators so you
     can write things like

        X  = A * (B + C);

2.   need a variety of types of matrices

3.   need only one element type (float or double)

4.   work with matrices in the range 4x4 to 90x90

5.   tolerate a large and complex package


Then maybe this is the right package for you.

If you don't need (1) then there may be better options. Likewise if you
don't need (2) there may be better options. If you require "not (5)"
then this is not the package for you.


If you need (2) and "not (3)" and have some spare money, then maybe you
should look at M++ from Dyad or the Rogue Wave matrix package.


---------------------------------------------------------------------------


Changes
=======

Newmat03 - November 1991:

Col and Cols become Column and Columns. Added Sort, SVD, Jacobi,
Eigenvalues, FFT, real conversion of 1x1 matrix, "Numerical Recipes in
C" interface, output operations, various scalar functions. Improved
return from functions. Reorganised setting options in "include.hxx".


Newmat02 - July 1991:

Version with matrix row/column operations and numerous additional
functions.


Matrix - October 1990:

Early version of package.


---------------------------------------------------------------------------


How to get a copy of this package
=================================

I am putting copies on Compuserve (Borland library, zip format),
SIMTEL20 (MsDos library, zip format), comp.sources.misc on Internet
(shar format), and on the MsDos program library at Victoria University,
Wellington.


---------------------------------------------------------------------------


Compiler performance
====================

I have tested this package on a number of compilers. Here are the
levels of success with this package. In most cases I have chosen code
that works under all the compilers I have access to, but I have had to
include some specific work-arounds for some compilers. For the MsDos
versions, I am using a 386/387sx computer running MsDos 5, except that
Turbo is on an old XT. The unix versions are on a Sun Sparc station.

A series of #defines at the beginning of "include.hxx" customises the
package for the compiler you are using. Turbo, Borland and Zortech are
recognised automatically, otherwise you have to set the appropriate
#define statement.

The compilers are looking a bit old now. I do intend to test the package
against newer versions as they become available.

Borland C++ 2.0: Recently, this has been my main development platform,
so naturally almost everything works with this compiler. The library
manager "tlib" fails but you can use "zorlib" from Zortech instead.
Sometimes Borland crashes during a compiler or mis-compiles. You just
have to reboot and continue the compile.

Turbo C++ (? version): Almost works OK. My rather elderly version does
show a problem. Probably not worth tracking down - buy a newer version.
Haven't tried the linker.

Zortech C++ 2.12: "const" doesn't work correctly with this compiler, so
the package skips all of the statements Zortech can't handle. If you are
using a later version of Zortech you could probably re-activate these
statements. Zortech leaves rubbish on the heap. I don't know whether
this is my programming error or a Zortech error. It works better when
one doesn't optimise but there still are problems. The nric routines
don't work. Zortech does not support IO manipulators.

Glockenspiel C++ (2.00a for MsDos loading into Microsoft C 5.1): I
haven't tested the latest version of my package with Glockenspiel. I had
to #define the matrix names to shorter names to avoid ambiguities and
had quite a bit of difficulty stopping the compiles from running out of
space and not exceeding Microsoft's block nesting limit. A couple of my
test statements produced statements too complex for Microsoft, but
basically the package worked. This was my original development platform
and I still use .cxx and .hxx as my file name extensions. However,
Glockenspiel is no longer competitive for MsDos and I am not updating my
copy of the compiler.

Sun AT&T C++ 2.00: This works fine. Except aggregates are not supported.

Gnu G++ 1.37.1: This mostly works. You don't seem to be able to use
expressions like Matrix(X*Y) in the middle of an expression and
(Matrix)(X*Y) is unreliable. Gnu does not support IO manipulators. Gnu
leaves rubbish on the heap. This is from output statements and not my
package and may not be an error. The previous version of the package did
not work under Gnu 1.37 or 1.39.


--
============================================================
Bob Crowe                      Voice:   [phone]
Pre-Press Technologies.        Email:    [e-mail]
2443 Impala dr.               FAX:     [phone]