Application of Genetic Recombination to Genetic Local Search in TSP

In this paper, an approach based upon Genetic Recombination is proposed, and applied to the Traveling Salesman Problem (TSP). The algorithm is composed of two Sample Genetic Algorithms (SGAs) which only consist of the basic genetic operators such as selection, crossover and mutation. One of the SGAs is named as the Global Genetic Algorithm (GGA), and encompasses the main tours where it is designed to search for the global optimal solutions. Another one is named as the Local Genetic Algorithm (LGA) and traverses over the sub tours to find the local optimal solutions. The LGA is combined to the GGA as an operator. The local optimal solutions are recombined to the main tours for improving the search quality. To investigate the features of the proposed algorithm, it is applied to a small double circles TSP, and some interesting results are presented in our experiments. Keyword: genetic algorithm, local search, genetic recombination, traveling salesman problem