[Trilinos-Users] ML: -\Delta u + i omega u = f

Nico Schlömer nico.schloemer at gmail.com
Fri Mar 22 16:01:07 MDT 2013


Hi all,

I'm solving this complex-valued linear PDE problem

-\Delta u  + i omega u = f

with omega>0, and, let's say, homogeneous Dirichlet boundary. After
discretization with linear finite elements, this problem should be
very treatable with AMG according to
<https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=877730>.
Unfortunately, they don't specify the AMG parameters too clearly and I
cannot reproduce their results in one aspect: While small for omega=1,
the number of GMRES iterations increases with omega, and quite quickly
so.
I'm thinking that I might not be on the correct set of parameters.

Does anyone have any experience with this?

--Nico



More information about the Trilinos-Users mailing list