mahout-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jake Mannix <jake.man...@gmail.com>
Subject Re: Lanczos?
Date Mon, 22 Feb 2010 05:53:58 GMT
On Sun, Feb 21, 2010 at 9:46 PM, Ted Dunning <ted.dunning@gmail.com> wrote:

> Jake,
>
> This sounds like Krylov iteration rather than Lanczos' algorithm.  Is this
> just an issue with the description?
>

What I've implemented (modulo the timesSquared() vs. times() distinction) is
almost exactly what is in the basic wikipedia article linked to in the wiki
( http://en.wikipedia.org/wiki/Lanczos_algorithm the bolded part for
"algorithm" ).
Am I not describing it well?


> I was under the impression that Lanczos' and Arnoldi methods have much
> better numerical stability than raw Krylov iteration.  Is there some
> improvement to be had here?


Quite possibly!  Can you be more specific?  The code in LanczosSolver fits
on one screen, really, so check it out (don't even worry about
DistributedLanczosSolver, there's nothing new there) and tell me what you
think.

  -jake

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message