Re: SVD for a large sparse matrix
- From: "John D'Errico" <woodchips@xxxxxxxxxxxxxxxx>
- Date: Fri, 24 Feb 2006 01:34:03 -0500
vivek wrote:
Hi everyone,
I would like to know if there is a way to calculate the singular
value decomposition (SVD) of a large matrix.
The matrix dimension is of the size 17640 x 254016. It's currently
a
sparse matrix.
When i use the matlab's SVD function its gives "out of memory"
error
as it tries to convert the sparse matrix into a full one.
Then i tried using SVDS which is svd for sparse matrices, but the
problem with that one is that if matrix A is of size m by n then
U is of m by k
S is of k by k
V is of n by k
where k is an integer that you specify.
So if i do [U,S,V] = svds(A,10);
I'll get U = 17640 by 10, s = 10 by 10 and V = 254016 by 10 where
as
in a real SVD, if A is m by n,
U will be m by m
S will be m by n
V will be n by n
Can anyone help me ?
What is wrong with that?
So how will you expect to store a full version
of U and V? Remember they will be dense matrices
even if A was sparse.
HTH,
John D'Errico
.
- Follow-Ups:
- Re: SVD for a large sparse matrix
- From: vivek
- Re: SVD for a large sparse matrix
- References:
- SVD for a large sparse matrix
- From: vivek
- SVD for a large sparse matrix
- Prev by Date: hi..
- Next by Date: Re: matlab can't multiply?? HELP!!!
- Previous by thread: Re: SVD for a large sparse matrix
- Next by thread: Re: SVD for a large sparse matrix
- Index(es):
Relevant Pages
|
