I see. Thanks. Doing two SVD is too stupid.


________________________________
From: Ted Dunning <[email protected]>
To: [email protected]; Zhengguo 'Mike' SUN <[email protected]>
Cc: Dan Brickley <[email protected]>
Sent: Friday, February 25, 2011 1:38 PM
Subject: Re: Understanding Mahout's Hadoop SVD results - eigenvectors/values vs 
decomposed matrices

No.

You can multiply the original matrix by the transpose of the available
eigenvectors and the inverse of the eigenvalues to get the missing
eigenvectors.  No iterations or fancy math needed.

On Fri, Feb 25, 2011 at 10:08 AM, Zhengguo 'Mike' SUN <[email protected]
> wrote:

> Thus, in order to get all the singular vectors, I have to do two mahout
> SVD. One with the original matrix, the other with the transpose A'. Is that
> right?
>


      

Reply via email to