Abstract
A cluster routing problem on graphs is considered in which the vertex set is partitioned into several clusters and each cluster must be served exactly once by a single server. The objective is to minimize the total cost, involving the traveling cost and service cost, of a service tour. The problem is one of variants known as the Generalized Traveling Salesperson Problem (GTSP) in the literature. For the NP-hard problem, a heuristic algorithm is designed which employs a genetic search based procedure to seek for a sequence of clusters, and a known dynamic programming based shortest path procedure to choose a vertex to be served in each cluster. A part of numerical results for test instances with forty vertices and ten clusters is reported.
