This is the mail archive of the
gsl-discuss@sourceware.cygnus.com
mailing list for the GSL project.
Re: matrix multiplication broken
- To: Paul Walmsley <pwalmsley at iee dot org>
- Subject: Re: matrix multiplication broken
- From: Brian Gough <bjg at network-theory dot co dot uk>
- Date: Wed, 16 Feb 2000 12:23:15 +0000 (GMT)
- Cc: gsl-discuss at sourceware dot cygnus dot com
- References: <Pine.GSO.4.21.0002151522240.22673-200000@abed.eng.cam.ac.uk>
Thanks for the bug report! I have incorporated the tests, which are
very useful. I have fixed the multiply code in a different way, as
there was also a matrix size/physical dimension issue involved -- it
should all be working now anyway.
regards
Brian Gough
Paul Walmsley writes:
>
> I've recently downloaded and started using the gsl library (primarily with
> C++ wrappers for matrix classes) but I've found that the generalised
> multiply operation gsl_la_matmult_mod_impl() (in multiply.c) is broken.
> It works fine for NxN matrices but not MxN.
>
> I've attached a patch for this. Many apologies: it isn't very elegant, and
> has a separate subroutine for each combination of A*B, A'*B, A*B' and
> A'*B', and only deals with the case for matrices of doubles. I've also
> included a patch to the test_la.c program to test MxN multiplication,
> which will fail on the original (release 0.5) source.
>
> The GSL is a great resource -- I'm very glad to have come across it. Keep
> up the good work.
>
>
> Paul Walmsley
> --------------------------------------------------------------------------
> Signal Processing Group
> Cambridge University Engineering Department
> pwalmsley@iee.org pjw42@cam.ac.uk http://www-sigproc.eng.cam.ac.uk/~pjw42