On Oct 27, 2010, at 8:21 AM, Jed Brown wrote:
> On Wed, Oct 27, 2010 at 14:47, Barry Smith <bsmith at mcs.anl.gov> wrote:
> I've considered it but have not seen a specific need. Do you have examples
> where it really matters?
>
> The natural one would be Dirichlet boundary conditions for (non-Galerkin)
> geometric multigrid with either variable coefficients or non-uniform mesh
> size. I still prefer modifying the element matrices before assembly (which
> also maintains symmetry), but if you want to use MatZeroRows to impose the
> boundary condition, you would also need a non-constant diagonal.
>
Right, but how do you know what values you should put on the diagonal?
Should there be an option to just leave the value that is currently there?
Should we change it so we always just leave the value that is already there?
Damn C for not having named arguments.
Barry
> Jed
