On Mon, Apr 06, 2009 at 12:13:53PM -0500, Andy Ray Terrel wrote: > I usually overload my own NonlinearProblem for this purpose, it makes > it much easier to switch out different types of newton solvers as > well. Even the demos go back and forth on how to define a non-linear > solver some using Variational Problem others building their own. It > would be nice to have a way to just give a VariationalProblem a newton > solver. > > +1 for adding getters and setters here.
The NewtonSolver is now parametrized by the VariationalProblem so any
parameters set for a problem should be inherited by the Newton solver:
problem.set("Newton relative tolerance", 1e-10);
I've also made the Newton solver itself directly accessible. I haven't
made any modifications to NewtonSolver itself though. I'll leave that
to Garth if he thinks it's appropriate (to allow modifying the linear
solver and preconditioner).
--
Anders
signature.asc
Description: Digital signature
_______________________________________________ DOLFIN-dev mailing list [email protected] http://www.fenics.org/mailman/listinfo/dolfin-dev
