Partitioning Algorithms and Their Application to
Massively Parallel Computations of Multiphase
Fluid Flows in Porous Media

Abstract

In this EPA funded project (joint between Texas A&M and Rice University) we are developing algorithms for simulating multi-phase flows in porous media on parallel distributed memory computers. The computational problems arising in such simulations are very large and require appropriate mechanisms for distributing the data among the processors. The goal is that the load be evenly divided among the processing elements and that the communication between processors be minimized.
Graph separators provide natural tool for finding such partitions and one of the goals of the project is to implement and make experiments with load balancing algorithms based on our research on graph separation.