Renumbering sparse matrices by simulated annealing

Abstract

This paper deals with Simulated Annealing algorithm and their application to reduce simultaneously the band of a sparse matrix and their total number of null terms between the first non-null term in a row of the lower triangle and the diagonal term. The t arget of this application is to increase in practice, performance of ILU factorization as preconditioner. This problem is a multi-objective optimization problem with two terms in the utility function cost.