Re: [boost] Re: ublas & custom containers? (2)

From: Joerg Walter (jhr.walter_at_[hidden])
Date: 2002-04-08 14:42:37

Hi Kresimir,

you wrote:

[snip]

> > As the very_long_class_names should indicate, classes like
vector_of_vector
> > and sparse_vector_of_sparse_vector were mainly intended as a proof of
> > concept, that one is able to adapt traditional C++ container of
container
> > (like std::vector<std::vector<> >) as a matrix format.
>
> > Your proposed vector_of_sparse_vector is the only form of these, I could
> > vaguely imagine to use in a production environment. But wouldn't you
really
> > prefer the traditional compressed storage format?
>
> You are probably right. But maybe there are some libraries
> (sparse solvers in particular) which expect vector_of_sparse_vector
> format. I still have old Roldan Pozo's TNT library (version 0.7 from 1997)
> where sparse matrices are implemented that way.

Interesting. This is somewhat a question of priorities: after some tests
with CLAPACK I'm currently focussing to borrow a test suite for sparse
matrices. The best opportunity I've found so far is SuperLU (thanks, Toon
;-).

> (Newer versions
> have only dense vectors and matrices, but the stated goal is to
> integrate `Lapack++', `Sparselib++' and `IML++'. And although there
> were no new developments for more than a year and a half, it
> seems that Pozo will finally resume the work -- citation from
> `http://math.nist.gov/tnt/index.html': ``Note: TNT 1.0 release scheduled
> for Summer 2002.'')

May be, we should also take a rest ;-)

Best regards

Joerg