Re: [Bug-gsl] SV decomp failure

[Peter Lelievre]

Hi Rhys.
I increased that factor by 10 several times but this still lead to the same 
test failure for -O2. I got up to 100,000*N but went no further.
Peter

On 2012-09-18, at 2:18 PM, Rhys Ulerich wrote:

>> Looks to be that way. -O1 was fine. -O2 was not.
> 
> The "failed to converge" messages you reported come from
> linalg/svd.c:149 which is part of the following check:
> 
>        if (iter > 100 * N)
>          {
>            GSL_ERROR("SVD decomposition failed to converge", GSL_EMAXITER);
>          }
> 
> The N factor makes sense to me.  Can anyone comment on how 100 was
> chosen?  The documentation says that this is the "Golub-Reinsch SVD
> algorithm".
> 
> Peter, if you tweak that 100 to be some higher value can you get the
> linalg 'make check' tests to pass at -O2 on your OS X system?
> 
> - Rhys


This electronic communication is governed by the terms and conditions at
http://www.mun.ca/cc/policies/electronic_communications_disclaimer_2012.php