Large & sparse SVD

From: fijoy (tofijoy_at_yahoo.co.in)
Date: 12/26/04 

Date: Sun, 26 Dec 2004 15:10:07 -0500

Hello,

I need to compute the SVD of a large and sparse matrix: approximately, 7.7
million rows, and 20 columns with 10% non-zero elements. I will not need all
the eigenvectors; only the first 50 or so. Does anyone know of an existing
software or easily implementable algorithm which can do this?

I did some simple experiments for doing this on Matlab, but ran out of
memory everytime I tried.

Thanks
Fijoy

Relevant Pages

  • Meschach - SVD on a sparse matrix
    ... I'm trying use meschach, compiling some programs, etc. because I need ... compute svd on a sparse matrix. ... There are, of course, sparse matrix packages that are ...
    (sci.math.num-analysis)
  • Re: Meschach - SVD on a sparse matrix
    ... I'm trying use meschach, compiling some programs, etc. because I need ... compute svd on a sparse matrix. ... There are, of course, sparse matrix packages that are ...
    (sci.math.num-analysis)
  • Re: Counting non-zero elements in sparse matrix
    ... I have a large NxM sparse matrix. ... number of non-zero elements in sub-matrices of this matrix. ... The sub-matrices are index by row and column indices of the original ... An alternative is to store your sparse matrix in a hierarchical format like ...
    (comp.programming)
  • Re: Counting non-zero elements in sparse matrix
    ... I have a large NxM sparse matrix. ... number of non-zero elements in sub-matrices of this matrix. ... An alternative is to store your sparse matrix in a hierarchical format like ...
    (comp.programming)
  • Re: Meschach - SVD on a sparse matrix
    ... compute svd on a sparse matrix. ... There are, of course, sparse matrix packages that are ... matrices are self-contained data structures, ...
    (sci.math.num-analysis)