An improved genetic algorithm based on gene pool for TSP

Abstract

Traveling salesman problem is a typical representative of combinatorial optimization problems. An improved Genetic algorithm is proposed for solving Traveling Salesman Problem (TSP). This Partheno-genetic algorithm employs only mutation and selection operators to produce the offspring, A new combinatory operator is designed combining the gene pool operator with inversion operator which ensures its strong searching capability. The gene pool directs the single-parent evolution and enhances the evolutionary speed. This algorithm simulates the recurrence of nature evolution process. Experiments based on 4 instances selected from TSPLIB are used to test the performance of this algorithm. They prove that it can reach the satisfying optimization at a faster speed. Especially, for the KroA100, the best path it finds is better than any other available one. © 2014 Springer International Publishing.