[Changeset] Re: BUG: invert a square sparse matrix.

[Julien MICHOT]

David Bateman a écrit :
> David Bateman wrote:
>   
>> What is happening in your case is that the fallback to
>> the LU factorization is not taking place (I'll figure out why and send a
>> patch later).
>>     
>
> Ok, in the cholesky factorization code in the sparse_base_chol class the
> "info" flag was being reset by the rcond calculation. The mldivide code
> doesn't have an issue as it called cholmod directly. Changeset attached
>
> D.
>   
Ok, thank's a lot.

-- 

Julien MICHOT, PhD
julien.michot at cea.fr
CEA/SAC/DRT/LIST/DTSI/SARC/LSVE