At Sat, 21 Mar 2009 18:10:07 +0100,
Christophe Dehais wrote:
> 
> I'm trying to compute the null space of a matrix A with M < N, using
> the residual singular vectors of the SVD decomposition of A.
> However, as the SVD code doesn't allow M < N, I have to work on A tanspose.
> I could use the fact that if A' = U.S.V' then A = V.S.U' and return
> the last vectors of U, but U is actually truncated because we compute
> a thin SVD,
> so some (or all in case M is full rank) vectors of M's null space are lost.
> 
> I'm guessing that either the SVD code should support the case M < N or
> there should be an option to compute the full SVD decomposition.

I agree that it would be good to add support for M<N to the SVD.  Any
help appreciated.  I believe it's possible to also get the null space
from a QR decomposition.

-- 
Brian Gough
(GSL Maintainer)

Support freedom by joining the FSF!
http://www.fsf.org/associate/support_freedom/join_fsf?referrer=37


_______________________________________________
Help-gsl mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/help-gsl

Reply via email to