[Amath-seminars] J. Shadid's talk 02/23 (Room GUG 415L)

Ulrich Hetmaniuk hetmaniu at u.washington.edu
Tue Feb 23 06:35:54 PST 2010


Dear All,

Please note the room change (GUG 415L)

On Tuesday 02/23, at 10:30 am, in GUG 415L, we will have a special
talk from John Shadid (SNL).
John will spend the day on campus. Please let me know if you would
like to meet with him.
Feel free to forward to this announcement.


Progress on the development of a scalable fully-implicit stabilized
unstructured
finite element (FE) capability for low-Mach-number resistive MHD.


John N. Shadid
Computational Science R&D Group
Sandia National Laboratories

Abstract:

This talk describes recent progress on the development of a scalable
fully-implicit stabilized unstructured finite element (FE) capability
for low-Mach-number resistive MHD. The brief discussion considers the
development of the stabilized FE formulation and the underlying fully-
coupled preconditioned Newton-Krylov nonlinear iterative solver. To
enable robust, scalable and efficient solution of the large-scale
sparse linear systems generated by the Newton linearization, fully-
coupled multilevel preconditioners are employed. The multilevel
preconditioners are based on two differing approaches. The first
algebraic multilevel technique employs a graph-based aggressive-
coarsening aggregation method applied to the nonzero block structure
of the Jacobian matrix. Initial results for a second approach that
utilizes approximate block decomposition methods and physics-based
preconditioning approaches will also be presented.

The performance of the multilevel preconditioners is compared to
standard variable overlap additive one-level Schwarz domain
decomposition type preconditioners. Parallel performance results are
presented for a set of challenging prototype problems that include the
solution of an MHD Faraday conduction pump, a hydromagnetic Rayleigh-
Bernard linear stability calculation, and a magnetic island
coalescence problem. Initial results that explore the scaling of the
solution methods are presented on up to 4096 processors for problems
with up to 64M unknowns on a CrayXT3/4. Additionally, a large-scale
proof-of-capability calculation for 1 billion unknowns for the MHD
Faraday pump problem on 24,000 cores is also presented.


Best regards,

Ulrich Hetmaniuk






More information about the Amath-seminars mailing list