[Clp] Clp Digest, Vol 109, Issue 7

[Miles Lubin]

Hi Wolfgang,

If you provide a sparse problem to Clp, it will certainly use a sparse
factorization by default. In fact, it uses a highly specialized
Markowitz-type factorization that's more suited to the structural
properties of LP basis matrices than more well-known state-of-the-art
sparse linear algebra libraries (see
http://page.math.tu-berlin.de/~liesen/Publicat/lasimp09.pdf).

Best,
Miles

On Wed, Aug 6, 2014 at 3:57 AM, Wolfgang Hartmann
<cmat.wolfgang at gmail.com> wrote:
> Just to make sure that I solved this Problem.
> By some ridiculous mistake, I did specify not enough
> iterations in my Interface to Clp. Sorry, for bothering,
> with a little bit more patience I would have found that
> without contacting the community.
>
> Unortunately, compared with some other LP methods
> is my current Interface to Clp's primal and dual Simplex
> algorithm a Little bit slow. But that may have to do with
> the Default factorization (which I think is dense). I will
> enhance my Interface with some additional sparse factorizations.
>
> There are quite a number of methods set up, mostly by Tim Davis,
> (has moved from Florida to Texas A&M)?
> Could anybody point me to one or two of the "best"?
>
> Wolfgang in Heidelberg