On Tue, Jan 20, 2009 at 12:49 PM, Derek Gaston <[email protected]> wrote:
> I know there was a lot of discussion that led up to adding the
> ShellMatrix object to libMesh.... and looking back through it I see
> that preconditioning was mentioned... but now that I'm looking at the
> code I don't see a straight forward way to add a ShellMatrix as a
> preconditioner.... but maybe I'm missing it.

All I was able to finish was checking in Tim's patch, I didn't have a
chance to redo the whole NumericMatrix hierarchy yet.  It's on my todo
list but there are day-job things taking up time currently.


> What I mean is... there needs to be a call to PCSetType(pc,PCSHELL)
> (which there is) and a call to PCShellSetApply.... which there
> isn't... or am I just missing it?
>
> Currently I'm just using my own hacked up code that doesn't even use
> ShellMatrix... but it would be nice to use the interfaces (for easier
> transition to Trilinos).
>
> Also... on a related question.... what do you guys think about
> providing a Preconditioner base class in libMesh?  Concrete
> implementations of this class would be HyprePreconditioner,
> MLPreconditioner, LinearSolvePreconditioner, etc...  I guess they are
> similar to linear solvers... but not exactly the same....

I guess I don't see the need for this, but that's not to say I can't
be convinced.  There's nothing like that in PETSc AFAIK.

-- 
John

------------------------------------------------------------------------------
This SF.net email is sponsored by:
SourcForge Community
SourceForge wants to tell your story.
http://p.sf.net/sfu/sf-spreadtheword
_______________________________________________
Libmesh-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-devel

Reply via email to