Subject: Nullspace(M) takes a lot of time for a big matrix
From: Alasdair McAndrew
Date: Thu, 10 May 2007 09:55:47 +1000
If the matrix M is entirely numeric, use the singular value decomposition
from lapack:
dgesvd(M,true,true)
which will produce the singular values, and the left and right singular
vectors. You can use these to find a basis for the nullspace.
-Alasdair
On 5/10/07, ptomaine <ptomaine at bk.ru> wrote:
>
> Hello.
> Thanks for the answer on previous question. That helped a lot.
>
> Now i'm trying to find a nullspace of a 243x270 matrix. It has a lot
> of zeros. Actually, it has nonzeros in 5 place in each row. Some of
> nonzero elements are 1, -1. Some are symbolic a, b, c. ( a, b, c are
> declared constants. )
>
> The problem is a that calculating takes a lot of time, while nullspace
> for the same matrix in Matlab is calculated ( symbolically ) for like
> several seconds. In Maxima it takes > 15 minutes and i didn't have
> enough courage to wait for finish.
>
> Is there any algorithm or workaround to fasten the process of finding
> nullspace?
>
> Thanks in advance, ptomaine//
> _______________________________________________
> Maxima mailing list
> Maxima at math.utexas.edu
> http://www.math.utexas.edu/mailman/listinfo/maxima
>