EXTENDED DEPENDENCY ANALYSIS OF LARGE SYSTEMS∗ Part II: Static Analysis

This paper continues the investigation started in Part I, describing the extension of Dependency Analysis by the use of heuristic methods. The extended methods are capable of detecting and analyzing the structure of large systems containing hundreds or thousands of variables. The methods in this Part discover “static” relations, that is, relations in which the time dimension is immaterial and causality is not imputed. In the analysis of static relations, complications arise which do not appear in the analysis of dynamic relations, and these are described and appropriate heuristics for dealing with them are proposed. The methods are illustrated by a 100-variable example. The results of many tests with large systems are discussed, showing the method to be nearly free of errors. It is argued on theoretical grounds that the method may actually become less prone to error as the size and complexity of the system increases