dolfin team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

Great. How does it compare to solving with a dense uBlas matrix (for a
reasonably small system) and using the "built-in" uBlasMatrix::solve()?

/Ander

On Mon, Jul 10, 2006 at 06:37:06PM +0200, Garth N. Wells wrote:
> On Mon, 2006-07-10 at 18:32 +0200, DOLFIN wrote:
> > One or more new changesets pushed to the primary DOLFIN repository.
> > A short summary of the last three changesets is included below.
> > 
> > changeset:   2054:0691435ba8b47358c9c50a363ac3424c24ca8420
> > tag:         tip
> > user:        "Garth N. Wells <g.n.wells@xxxxxxxxxx>"
> > date:        Mon Jul 10 18:13:24 2006 +0200
> > files:       configure configure.ac src/kernel/la/dolfin/uBlasLUSolver.h src/kernel/la/dolfin/ublas.h src/kernel/la/uBlasLUSolver.cpp src/test/Makefile src/test/main.cpp
> > description:
> > Use UMFPACK (if installed) for LU solution of uBlas matrices.
> > 
> 
> The direct solver for uBlas data types now uses UMFPACK (which is very
> fast) if it's installed, otherwise it uses a Krylov solver. When UMFPACK
> is installed, the demos in src/demo/pde run nice and fast with uBlas
> matrices and vectors.
> 
> If you have a Debian-based distribution, you can install UMFPACK by
> doing
> 
>     apt-get install libumfpack4 libumfpack4-dev 
>     
> Packages are available for most common Linux distributions under the
> name "umfpack" or "ufsparse".
> 
> Garth
> 
> 
> > 
> > changeset:   2053:10663473e69cf9ea1824380b0ef1181744696623
> > user:        "Garth N. Wells <g.n.wells@xxxxxxxxxx>"
> > date:        Fri Jul 07 20:18:29 2006 +0200
> > files:       src/kernel/la/dolfin/uBlasMatrix.h
> > description:
> > Use iterators in uBlasMatrix::disp().
> > 
> > 
> > changeset:   2052:9ab764260e78856ffab0b4cb74f90c8815896478
> > user:        "Anders Logg <logg@xxxxxxxxx>"
> > date:        Fri Jul 07 19:59:38 2006 +0200
> > files:       src/kernel/la/dolfin/uBlasMatrix.h
> > description:
> > Fix bug in uBlasMatrix::disp(), remove unused variable N
> > 
> > 
> > -------------------------------------------------------
> > For more details, visit http://www.fenics.org/hg/dolfin
> > _______________________________________________
> > DOLFIN-dev mailing list
> > DOLFIN-dev@xxxxxxxxxx
> > http://www.fenics.org/mailman/listinfo/dolfin-dev
>