This is the mail archive of the
gsl-discuss@sources.redhat.com
mailing list for the GSL project.
Re: SVD bugs in gsl version .9.1?
- To: Devin Balkcom <devin at ri dot cmu dot edu>
- Subject: Re: SVD bugs in gsl version .9.1?
- From: Brian Gough <bjg at network-theory dot co dot uk>
- Date: Tue, 28 Aug 2001 23:02:59 +0100 (BST)
- Cc: gsl-discuss at sources dot redhat dot com
- References: <3B8BE5C9.FA9DE2F@ri.cmu.edu>
Devin Balkcom writes:
> I'm using gsl to take singular value decompositions. I have had
> problems with all three of the
> implementations. I have written a test program.
>
> - gsl_linalg_SV_decomp() does not return for the test matrix.
> - gsl_linalg_SV_decomp_mod() also does not return.
> - gsl_linalg_SV_decomp_jacobi() does return, although I have had it fail
> in other circumstances,
> with an incorrect result. (I will send test code for this case
> later.)
Thanks for the bug report.
It is a bug in the SVD.... which is not handling the 2x2 case [0,X;0,0]
correctly, so it never converges for that matrix. I'll fix that.
regards
Brian Gough