Parallel Refinement of Tetrahedral Meshes on Distributed-Memory Machines

S. Blazy and O. Marquardt (Germany)


Cluster Computing, Algorithms, Parallel Refinement, Nu merical Software Framework, padfem2 AMS subject classification 65M50, 65Y05


In this paper we present a scalable parallel refinement al gorithm for tetrahedral meshes which is based on a regular refinement strategy with closure tetrahedra. This algorithm works efficiently on distributed-memory machines. We fo cus on the parallel implementation of the boundary updates of each mesh partition. Additionally we show that it is sufficient to exchange only boundary edge information be tween neighboring mesh partitions to obtain a conforming distributed three-dimensional tetrahedral mesh. The algo rithm is implemented using the padfem2 framework.

Important Links:

Go Back