Graph Partitioning Algorithms with Applications to Scientific Computing Graph Partitioning Algorithms

Identifying the parallelism in a problem by partitioning its data and tasks among the processors of a parallel computer is a fundamental issue in parallel computing. This problem can be modeled as a graph partitioning problem in which the vertices of a graph are divided into a speciied number of subsets such that few edges join two vertices in diierent subsets. Several new graph partitioning algorithms have been developed in the past few years, and we survey some of this activity. We describe the terminology associated with graph partitioning, the complexity of computing good separators, and graphs that have good separators. We then discuss early algorithms for graph partitioning , followed by three new algorithms based on geometric, algebraic , and multilevel ideas. The algebraic algorithm relies on an eigenvector of a Laplacian matrix associated with the graph to compute the partition. The algebraic algorithm is justiied by formulating graph partitioning as a quadratic assignment problem. We list several papers that describe applications of graph partitioning to parallel scientiic computing and other applications. Finally we discuss the application of graph partitioning to obtain nested dissection orderings for solving sparse linear systems of equations in parallel.