Search area Expanding Strategy and Dynamic Priority Setting Method in the Improved 2-opt Method

Abstract

We propose a new 2-opt base method in a Memetic algorithm, that is, Genetic Algorithms(GAs) with a local search. The basic idea is from the fast 2-opt⁽¹⁾ method and the improved 2-opt method⁽²⁰⁾. Our new search method uses the “Priority" employed in the improved 2-opt method. The “Priority" represents the contribution level in exchange of genes. Matayoshi's method exchanges genes based on previous contribution to the fitness value improvement. We propose a new search method by using the concept of the Priority. We call our method the search area expanding strategy method in the improved 2-opt method. Our method escalates the search area by using “Priority". In computer experiment, it is shown that the computation time to find exact solution depends on the value of the Priority. If our method does not set an appropriate priority beforehand, then we propose the method to adapt to suitable value. If improvement does not achieved for certain generations, our dynamic priority method tries to modify the priority by the mutation operation.
Experimental results show that the search area expanding strategy method embedded with the dynamic priority setting method can find the exact solution at earlier generation than other methods for comparison.