Boost logo

Ublas :

From: Rutger ter Borg (forms_at_[hidden])
Date: 2005-09-20 10:01:28


It could be that the matrices are in row_major format (uBLAS default). The
lapack bindings only work correctly with matrices using column_major
orientation.

Best regards,

Rutger

Peter Melchior wrote:

> Hello,
>
> I'm using LAPACK + bindings to do a Singular Value Decomposition of
> uBLAS matrices. For inverting a matrix, I need the U and the Vt matrix
> calculated explicitly.
>
> But whatever I give as Jobu and Jobvt, only the Vt matrix will be
> calculated correctly; the U matrix has only correct diagonal elements.
>
> I've tried the Jobu = 'A' setting for gesvd and the Jobz = 'A' setting
> for gesdd with exactly the same results as JobX = 'O'.
>
> I'm using the ublas bindings from CVS (checkout: 09.09.2005).
>
>
> Can anybody give me a hint, what might go wrong here.
>
>
> Best regards,
>
> Peter
>
> _______________________________________________
> ublas mailing list
> ublas_at_[hidden]
> http://lists.boost.org/mailman/listinfo.cgi/ublas