Assessments of the implementation of the minimum degree ordering algorithms

Abstract

The minimum degree ordering is one of the most widely used algorithms to preorder a symmetric sparse matrix prior to numerical factorization. There are number of variants which try to reduce the computational complexity of the original algorithm while maintaining a reasonable ordering quality. An in-house finite element solver is used to test several minimum degree algorithms to find the most suitable configuration for the use in the Finite Element Method. The results obtained and their assessments are presented along with the minimum degree ordering algorithms overview.