Boost logo

Ublas :

From: Karl Meerbergen (Karl.Meerbergen_at_[hidden])
Date: 2007-11-08 03:45:38


Hi Per,

You can indeed use the boost-sandbox bindings for LAPACK (dense
matrices) or UMFPACK or MUMPS for sparse matrices. All are available in
the boost-sandbox module boost/numeric/bindings.
Examples are available from the libs directory in boost-sandbox.

Some examples can also be found on http://tinyurl.com/2h4j5f
or http://www.cs.kuleuven.be/publicaties/rapporten/tw/TW506.abs.html

It is some work to get everything working. Hope it is fine.

Karl

Gunter Winkler wrote:

>Am Dienstag, 6. November 2007 14:05 schrieb Per Abrahamsen:
>
>
>>Hi ublas,
>>
>>We currently use lu_factorize and ublas::matrix A to solve an
>>equation system A x = b, and it works fine for the small matrixes we
>>have tested it with.
>>
>>What binding / library do you suggest?
>>
>>
>
>The AFAIK simplest way is to try
>
>http://guwi17.de/ublas/ublas_csparse.hpp
>
>together with
>
>http://www.cise.ufl.edu/research/sparse/CSparse/
>
>Alternatively you can use the boost-bindings library from the
>boost-sandbox repository which provided UMFPACK binding. (@Karl:
>correct me if I'm wrong)
>
>in both cases you should prepare you matrix as
>
>compressed_matrix<double, row_major> for C-libraries
>
>or
>
>compressed_matrix<double, column_major, 1> for FORTRAN libraries
>
>mfg
>Gunter
>
>
>------------------------------------------------------------------------
>
>_______________________________________________
>ublas mailing list
>ublas_at_[hidden]
>http://lists.boost.org/mailman/listinfo.cgi/ublas
>
>