Re: [Bug-gsl] bug in file eigen/jacobi.c

[Brian Gough]

At Wed, 24 Dec 2008 14:59:29 +0100,
stefan wrote:
> I have found a bug in the function:
> 
> int gsl_eigen_jacobi (gsl_matrix * a, gsl_vector * eval, gsl_matrix * 
> evec, unsigned int max_rot, unsigned int *nrot)
> 
> The Problem is that the routine does not stop if a good solution is 
> found! That means if I set the parameter max_rot for example to 1000 and 
> i call the function then nrot is set to 1000, so nrot is always max_rot. 
> I figured out the the function:

Hello,

This function only works for symmetric matrices.  The example program
has an nonsymmetric matrix so it will not converge.  For non-symmetric
matrices use gsl_eigen_nonsymm instead.

-- 
best regards

Brian Gough
(GSL Maintainer)

Support freedom by joining the FSF!
http://www.fsf.org/news/fall-2008-fundraiser