You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@mahout.apache.org by Young Y Kim <yo...@gmail.com> on 2010/04/15 07:52:29 UTC
lanczos problem
i'm testing my LanczosSolver and i'm having problems validating the result.
if i'm not mistaken, V^T * A * V should equal T (from
http://en.wikipedia.org/wiki/Lanczos_algorithm) but my
LanczosSolver is producing a completely different result.
how do i check if the resulting eigenvalues are correct?
I've used the basis as V and triDiag as T in solver function. is that
correct?
Re: lanczos problem
Posted by Grant Ingersoll <gs...@apache.org>.
Can you share your unit test?
On Apr 15, 2010, at 1:52 AM, Young Y Kim wrote:
> i'm testing my LanczosSolver and i'm having problems validating the result.
> if i'm not mistaken, V^T * A * V should equal T (from
> http://en.wikipedia.org/wiki/Lanczos_algorithm) but my
> LanczosSolver is producing a completely different result.
> how do i check if the resulting eigenvalues are correct?
>
> I've used the basis as V and triDiag as T in solver function. is that
> correct?
--------------------------
Grant Ingersoll
http://www.lucidimagination.com/
Search the Lucene ecosystem using Solr/Lucene: http://www.lucidimagination.com/search
Re: lanczos problem
Posted by Jake Mannix <ja...@gmail.com>.
Hi Young,
The test we do currently, to verify correctness, is to take the purported
eigenvectors
and simply check how far off from eigen-ness they are, as well as verify all
of the
eigenvectors are orthonormal.
See org.apache.mahout.math.hadoop.decomposer.EigenVerificationJob for how
we do cleanup and verification of a eigen extraction.
-jake
On Wed, Apr 14, 2010 at 10:52 PM, Young Y Kim <yo...@gmail.com> wrote:
> i'm testing my LanczosSolver and i'm having problems validating the result.
> if i'm not mistaken, V^T * A * V should equal T (from
> http://en.wikipedia.org/wiki/Lanczos_algorithm) but my
> LanczosSolver is producing a completely different result.
> how do i check if the resulting eigenvalues are correct?
>
> I've used the basis as V and triDiag as T in solver function. is that
> correct?
>