Experiments with Local Search Heuristics for the Traveling Salesman Problem

Abstract. In this paper, the experiments with local search heuristic algorithms for the traveling salesman problem (TSP) are described. Since the ordinary local search heuristics very seldom yield solutions of high quality, we investigate the enhanced local search algorithms using the extended neighbourhood structures. We also examine the performance of the local search heuristics in an iterated local search paradigm based on the deterministic descent-random ascent methodology. Our new heuristic algorithms are tested on the symmetric TSP instances taken from the publicly available electronic library of the TSP instances (TSPLIB). The results from the experiments demonstrate that our heuristics appear to be superior to traditional types of local search algorithms.