Gene-pool based genetic algorithm for TSP

Based on the analysis of previous genetic algorithms (GAs) for TSP, a novel method called Ge_GA is proposed. It combines gene pool and GA so as to direct the evolution of the whole population. The core of Ge GA is the construction of gene pool and how to apply ii to GA. Different from standard GAs, Ge_GA aims to enhance the ability of exploration and exploitation by incorporating global search with local search. On one hand a local search called Ge_LocalSearch operator is proposed to improve the solution quality, on the other hand the modified Inver-Over operator called Ge_InverOver is considered as a global search mechanism to expand solution space of local minimal. Both of these operators are based on the gene pool. Our algorithm is applied to 11 well-known traveling salesman problems whose numbers of cities are from 70 to 1577 cities. The experiments results indicate that Ge_GA has great robustness for TSP. For each test instance, the average value of solution quality, found in accepted time, stays within 0.001% from the optimum.