Prev: Are all Webinars Down?
Next: Numerical matrix with text headers
From: Rational Adjustment on 12 Aug 2010 19:48
I have a 408,789 x 89,534 sparse matrix with 21,390,749 non-zero elements.

Any idea what the fastest method is that will compute (or estimate) the rank of this monster? (Hopefully using sparse matrix calculations.)

Most methods run out of memory - I have 16GB total to work with.

Anyone ever hear of a trick to use the LSQR method to estimate rank? I recall a paper years back that suggested how, but can't remember who wrote it.
 | 
Pages: 1
Prev: Are all Webinars Down?
Next: Numerical matrix with text headers