Exploiting Structural Symmetry in Unsymmetric Sparse Symbolic Factorization

Abstract

This paper shows how to exploit structural symmetry in determining the nonzero structures of the lower and upper triangular factors L and U of an unsymmetric sparse matrix A. Two symmetric reductions of the graphs of L and U are introduced and used to formulate symbolic factorization algorithms. Experimental results demonstrate the effectiveness of these algorithms versus other schemes in the literature.