ALGORITMOS PRIMAIS E DUAIS PARA O PROBLEMA DAS P-MEDIANAS / PRIMAL AND DUAL ALGORITHMS FOR THE UNCAPACITED P-MEDIAN PROBLEM

AUTOR(ES)
DATA DE PUBLICAÇÃO

2009

RESUMO

A facility is any center that offers services to a set of clients. It may be, among others, a school, a factory or a depot. Facility location problems are combinatorial optimization problems that handle decisionmaking in respect to the positioning of those services, optimizing some defined criteria. The measures often used to assess the quality of a solution for this class of problems relate to which clients are served by which facility. An immediate consequence is the strong relationship between location problems and data clustering. One of the widely studied facility location problems is the uncapacited p-median problem (UPM), the main subject of this thesis. Given a set of possible facility locations, the UPM consists in determining a subset of locations at which the facilities shall be established, minimizing the sum of distances from each client to its closest open facility. The UPM belongs to the class of NP-hard problems and is a central problem of data clustering. This thesis presents primal, dual and exact algorithms for approaching the UPM, focusing on the development of dual and exact algorithms. Five constructive heuristics and one local search method were implemented. Furthermore, three new dual methods and one exact method were proposed. The result is the analysis of a set of techniques to solve the problem. The choice of best technique is strongly dependent of the configuration of the treated instance. We obtained the optimum for some instances and for others the difference between the value of the lower and upper bounds in the best cases do not exceed 3%.

ASSUNTO(S)

programacao inteira combinatorial optimization integer linear programming otimizacao combinatoria

Documentos Relacionados