Estimation of Sparse Jacobian Matrices and Graph Coloring Blems

Abstract

Given a mapping with a sparse Jacobian matrix, we investigate the problem of minimizing the number of function evaluations needed to estimate the Jacobian matrix by differences. We show that this problem can be attacked as a graph coloring problem and that this approach leads to very efficient algorithms. The behavior of these algorithms is studied and, in particular, we prove that two of the algorithms are optimal for band graphs. We also present numerical evidence which indicates that these two algorithms are nearly optimal on practical problems.

A scalable multi‐level preconditioner for matrix‐free µ‐finite element analysis of human bone structures

“…This requires that the columns belonging to the same group have all their nonzero elements on different rows. Coleman and Moré [17] discussed the ordering in which the columns should be considered, in order to minimize the number of groups. They showed that, for a general sparse pattern, the problem is equivalent to a certain coloring problem on a suitable graph, and proposed the use of graph coloring to obtain a small number of groups.…”

Discovering human immunodeficiency virus mutational pathways using temporal Bayesian networks

“…In this sense, a common measure of complexity in graph theory is density [32] edges of the graph. The maximal density is 1 for complete graphs, when the graph is fully connected and the minimal is 0 if all nodes are isolated.…”

Speeding up Parallel Graph Coloring

“…Supported by the U.S. National Science Foundation grant ACI 0203722 However, in practice greedy sequential coloring heuristics have been found to be quite effective [2].…”