Heuristics using a variable neighborhood approach for the clustered traveling salesman problem

In this work are proposed new heuristic algorithms to solve the Clustered Traveling Salesman Problem (CTSP). The CTSP is an extension of the Traveling Salesman Problem (TSP), where the vertices are partitioned into disjoint clusters and the goal is to find a minimum cost Hamiltonian cycle, such that all vertices of each cluster must be visited consecutively. Two algorithms using Iterated Local Search with the Variable Neighborhood Random Descent were implemented to solve the CTSP. The proposed heuristic algorithms were tested in the types of different granularity instances with the diversified number of vertices and clusters. Computational results show that the heuristic algorithms reached the best performance in comparison to the literature algorithms and they are competitive to an exact method using the Parallel CPLEX software