dolfin team mailing list archive

["Garth N. Wells" <gnw20@xxxxxxxxx>]

On 13/09/10 08:09, Anders Logg wrote:
> Question #125019 on DOLFIN changed:
> https://answers.launchpad.net/dolfin/+question/125019
>
>      Status: Open =>  Answered
>
> Anders Logg proposed the following answer:
> On Mon, Sep 13, 2010 at 06:11:13AM -0000, Evan Lezar wrote:
> > Question #125019 on DOLFIN changed:
> > https://answers.launchpad.net/dolfin/+question/125019
> >
> >      Status: Answered =>  Open
> >
> > Evan Lezar is still having a problem:
> > Hi
> >
> > I think Anders is correct. I have just had a look at the SystemAssembler
> > code, and it does seem to also place ones on the diagonals of the matrix
> > to which the boundary conditions are being applied. In addition, I am
> > not solving a linear system, but eigenvalue problems and as such b is a
> > Matrix not a Vector.
> >
> > At present I am manually removing the rows and columns for the matrices
> > that correspond to (zero) Dirichlet conditions and then reconstructing
> > the correct solution vector from the eigenvector that I obtain. This is
> > fine for testing and getting the rest of the system running.
> >
> > Thanks for the help so far.
>
> How do you remove the rows and columns?
>
> If there is good support for doing this in the backends (PETSc,
> Trilinos), we might consider adding a wrapper for it.

I doubt that there is good support (or in fact any support) for this. 
It's really nasty, especially in parallel.

What would be good is incorporating things like this in the sparsity 
pattern and not assembling the entries in the first place. Related to 
this, I added a Blueprint a while ago on modifying the sparsity pattern 
for periodic bcs to overcome the horrible performance we have at the 
moment due to manipulation an assembled sparse matrix.

Garth

> --
> Anders