P-Median: A Performance Analysis

Abstract

This work approaches the P-median problem with a partitioning around medoids methodology. This problem has been extensively studied because of the multiple applications and its NP-hard nature; therefore several are the efforts to find optimal solutions. In this paper we consider two case studies: 1) instances from OR-Library and 2) geographical objects from the metropolitan zone of Toluca, Mexico. The methods we use to work with these data are: Partitioning Around Medoids (PAM), Variable Neighborhood Search (VNS), Simulated Annealing (SA), P-Median with MATLAB (PML) and finally Tabu Search (TS).