Re: [Help-gsl] bug in gsl_linalg_LU_invert?

[Jordi Gutiérrez Hermoso]

On 28/04/2008, address@hidden <address@hidden> wrote:
> The multiplication of sigma and
>  sigma_inv should be 1 on diagonal and zero elsewhere. I am using
>  double and am expecting the epsilon of the result I get should be
>  small (< 1e-10) but actually some of them are very large (~0.001).

What is the condition number of your matrix? Can we see your matrix, please?

Thanks,
- Jordi G. H.