Hi Ted, On 25 February 2011 18:36, Ted Dunning <[email protected]> wrote: > Generally the SVD in these sorts of situations does not return the entire > set of three matrices. Instead either the left or right (but usually the > right) eigenvectors premultiplied by the diagonal or the square root of the > diagonal element. > I can't comment specifically on your situation, but hopefully knowing that > the algorithm isn't supposed to return all three matrices will help you.
That does indeed help! Clearly plenty more to learn... Also I found a JIRA issue reporting some oddities in the lanczos output (eg. returning one less result than requested). So I'm creeping up on getting it. But there's certainly a gap still, I'd be grateful if anyone can help bridge things back to http://www.igvita.com/2007/01/15/svd-recommendation-system-in-ruby/ -ish territory, and if we get that far I'll put some time into making a Wiki writeup with friendly examples... Dan
