Parallel and adaptive finite element algorithms for solving partial differential equations

Peter Jimack

(University of Leeds)

Our present research at Leeds is looking at efficient parallel and adaptive algorithms for the solution of differential equations on domains of a potentially complex geometry. This talk will focus mainly on our work into scalable distributed memory finite element algorithms for the solution of elliptic PDEs on unstructured grids. In particular the issues of parallel mesh generation and parallel domain decomposition preconditioning will be addressed. The talk will also consider related issues such as suitable measures of parallel efficiency for large problems, and different possibilities for adaptive remeshing in a parallel environment.
Tuesday 22nd November 1994, 14:30
Seminar Room 322
Department of Computer Science