On 2014-01-29 12:44, Jed Brown wrote:
"Garth N. Wells" <[email protected]> writes:
On 2014-01-29 12:11, Patrick Farrell wrote:
On 28/01/14 19:49, Garth N. Wells wrote:
A plus of this approach is that if your system is symmetric you can
use CG or
MINRES, which you can't with NonlinearDiscreteProblem.
I don't really understand your comment. Perhaps I've misunderstood.
You can use
CG or MINRES perfectly well without subclassing NonlinearProblem
directly. Try
it out in the nonlinear-poisson demo. That's using
NonlinearDiscreteProblem
under the hood. Am I missing something?
NonlinearDiscreteProblem breaks symmetry when applying bcs.
CG is usually robust to this form of weak asymmetry.
Even with preconditioning, say ILU?
Garth
MINRES is probably
okay since it iterates in a benign space due to the first application
of
a preconditioner fixing the boundaries. It's still worth implementing
symmetric BCs.
_______________________________________________
fenics mailing list
[email protected]
http://fenicsproject.org/mailman/listinfo/fenics
