Thread Subject: svd for a sparse matrix (not svds)

Hi, I'm trying to do the svd descomposition for a very big matrix which I have in sparse mode for saving memory, but svd function does not work in sparse matrix, so I have to take the full matrix to get the U S and V outputs, but the computing time is to high in this case; I tried to use the svds function to the sparse matrix, but the dimensions of the U S and V outputs that I get are not the same as with svd function, so what can I do in this case, which function can I use to get the same results as svd but for a sparse matrix? can I pad with zeros the resulting matrices in svds to obtain the same dimensions as svd function? Please help

"landon donovan" wrote in message <j6ob73$bm7$1@newscl01ah.mathworks.com>...
> Hi, I'm trying to do the svd descomposition for a very big matrix which I have in sparse mode for saving memory, but svd function does not work in sparse matrix, so I have to take the full matrix to get the U S and V outputs, but the computing time is to high in this case; I tried to use the svds function to the sparse matrix, but the dimensions of the U S and V outputs that I get are not the same as with svd function, so what can I do in this case, which function can I use to get the same results as svd but for a sparse matrix? can I pad with zeros the resulting matrices in svds to obtain the same dimensions as svd function? Please help

I'll be frank here: Trying to get SVD of large (sparse) matrix is foolish.

Bruno