Parallel Adaptive Multigrid Software for
Elliptic PDEs and Eigenvalue Problems
William F. Mitchell
NIST/MCSD/MSG
Tuesday, March 19, 2002 15:00-16:00,
Room 145, NIST North (820)
Gaithersburg
Tuesday, March 19, 2002 13:00-14:00,
Room 4550
Boulder
Abstract:
The solution of elliptic partial differential equations (PDEs) is the
most time consuming part of the solution of mathematical models for
many physical applications, such as fluid dynamics applications,
electromagnetic propagation, and semiconductor device simulation.
Over the years, much research has focused on the development of better
methods to solve these problems. Currently, the fastest methods
involve the use of adaptive grid refinement to concentrate the effort
in the most important part of the physical domain, and multigrid
solution methods that can solve the discretized system with an optimal
number of operations. Recent research has explored approaches to
implementing these methods on parallel computers. This talk describes
an approach to parallelizing adaptive grid refinement and multigrid
methods, and the design of an implementation in Fortran 90. These
methods can also be applied to eigenvalue PDEs, with applications in
quantum physics. Ongoing research in this field will be presented.
Contact: A. J. KearsleyNote: Visitors from outside NIST must contact
Robin Bickel; (301) 975-3668;
at least 24 hours in advance.
|
