Dear Derek,
On Thu, 20 Aug 2009, Derek Gaston wrote:
> On Aug 20, 2009, at 2:15 AM, Tim Kroeger wrote:
>
>> Can it reuse a precoditioner between successive solves without setting it
>> up again?
>
> It can if you create a Preconditioner that acts that way. It sounds like
> what you might want to do is create your own Preconditioner class... maybe
> inheriting from PetscPreconditioner and then override the init() method so
> that it doesn't clear the preconditioner. Then you might also want to create
> your own clear() method that does clear the preconditioner... and call this
> manually whenever you want to reset it.
Thank you very much; works great!
Anyway, just to be curious, I have two more questions about this:
1. Why does PetscPreconditioner::clear() not do anything? I guess, it
should at least call PCDestroy(), or is that called elsewhere?
2. I wonder what the suggested way is to cancel the effect of
LinearSolver::attach_precondition(), in particular if my
Preconditioner object goes out of scope. Currently, I'm just calling
LinearSolver::clear();
LinearSolver::attach_preconditioner(NULL);
LinearSolver::set_preconditioner_type(ILU_PRECOND);
to do that. (That's sort of an academic question because no solves
will be performed after that anyway, but it avoids dangling pointers.)
Best Regards,
Tim
--
Dr. Tim Kroeger
[email protected] Phone +49-421-218-7710
[email protected] Fax +49-421-218-4236
Fraunhofer MEVIS, Institute for Medical Image Computing
Universitaetsallee 29, 28359 Bremen, Germany
------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now. http://p.sf.net/sfu/bobj-july
_______________________________________________
Libmesh-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-users
