Sparsity-Oriented Network Reduction

Abstract: W.F. Tinney, W.L. Powel N.M Peterson, PAS 93 No.1, January/February 1974, p.6

The conventional approach to network reduction in which all non essential nodes are limited usually results in an equivalent that is so densely interconnected that sparse matrix methods cannot be effectively used on it. In order to obtain an equivalent that is suitable for sparsity exploitation, it is usually necessary to retain certain non essential node that would normally be eliminated. This paper identifies the main factors affecting sparsity of network equivalents, describes a practical algorithm for determining the non essential nodes that should be retained and makes suggestions for better algorithms. This paper is of importance in any network application in which sparse matrix method are used.