===affil2:
Tufts University
===firstname:
Thomas R.
===firstname4:
Scott P.
===firstname3:
Eric C.
===lastname2:
Adler
===lastname:
Benson
===firstname5:
Raymond S.
===affil6:

===lastname3:
Cyr
===email:
thomas.benson@tufts.edu
===lastname6:

===affil5:
Sandia National Labs
===otherauths:

===lastname4:
MacLachlan
===affil4:
Tufts University
===lastname7:

===affil7:

===firstname7:

===postal:
Department of Mathematics
Tufts University
503 Boston Avenue
Medford, MA 02155
===firstname6:

===ABSTRACT:
Magnetohydrodynamic models are used to model a wide range of plasma
physics applications. The system of partial differential equations
that characterizes these models is nonlinear and strongly couples
fluid interactions with electromagnetic interactions. As a result, the
linear systems that arise from discretization and linearization of the
nonlinear problem can be difficult to solve. In thistalk, we
consider a multigrid preconditioner for GMRES as a solver for these
systems. We compare three potential smoothers for this system, two of
which are motivated by well-known smoothers for the incompressible
fluids system and the other is a new smoother that splits the physics
into a magnetics-velocity operator and a Navier-Stokes operator. While
we examine these smoothers in the context of geometric multigrid, they
can be extended to relaxation schemes for algebraic multigrid. Results
for a two-dimensional, steady-state test problem are shown.
===affil3:
Sandia National Labs
===lastname5:
Tuminaro
===affilother:

===title:
Multigrid Smoothers for Magnetohydrodynamics
===firstname2:
James H.