[Haskell-cafe] Re: Purely funcional LU decomposition

[Rafael Gustavo da Cunha Pereira Pinto]

On Wed, Feb 4, 2009 at 05:15, Neal Alexander <wqeqweuqy at hotmail.com> wrote:

> Array is no good man! Quad Tree matrices perform much nicer from what I've
> seen.
>
> I wrote some matrix stuff based on D. Stott Parker's "Randomized Gaussian
> elimination" papers (http://www.cs.ucla.edu/~stott/ge/<http://www.cs.ucla.edu/%7Estott/ge/>).
> He presents some recursive block based methods of solving without pivoting.
>
> I want to upload the full library to Hackage, but it'd be nice to have some
> people look through it before - mainly because i never took linear algebra
> heh. QuickCheck seems to be holding things together though.
>
>

I read some of those papers and, yes, it is impressive. OTOH, I really need
pivoting (some ill-conditioned matrices are expected), which they all claim
to be hard to implement using quadtrees and I'm afraid that using RBTs might
introduce other sources of error.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.haskell.org/pipermail/haskell-cafe/attachments/20090205/1f14912d/attachment.htm