A New Co-evolutionary Genetic Algorithm for Traveling Salesman Problem

The paper presents the use of a new co-evolutionary genetic algorithm (CGA) for solving a traveling salesman problem. The genetic algorithm co-evolves individuals and schemata. Current genetic algorithm approaches are computationally intensive and may not produce acceptable tours within the time available. The CGA is inspired by the idea which effectively use of symbolized information on solution space can be useful for GA-based on search for solutions. The schemata have high average fitness values. In the algorithm, a new good point set crossover operator is utilized. The simulation results indicate that the use of the CGA is proven to be highly efficient and stable in comparison to the results given by a traditional genetic algorithm.