Analysis and Computational Schemes for p-Median Heuristics

This paper is concerned with solution procedures for the p-median problem: the well-established heuristic of Teitz and Bart, and the GRIA (Global/Regional Interchange Algorithm) technique developed more recently by Densham and Rushton. A computational scheme is presented which facilitates efficient implementations in both cases. The mathematical basis for the computational scheme is explained concisely by means of set-theory notation, and implementation of the Teitz—Bart heuristic is discussed with particular reference to search and storage considerations in large networks and in trees. In addition, it is shown that the two procedures in general do not terminate at solutions of equivalent local optimality.