Some Aspects of the Influence of Population Diversity on the Performance of Differential Evolution

This paper presents a study about some aspects of the influence of population diversity on the performance of the Differential Evolution technique. In order to accomplish this, the referred algorithm is tested with different benchmark functions widely used in the literature, and the performance results are analyzed and discussed by associating changes in the population diversity with changes in the evolution of the best solution over the generations. The objective of this work is to investigate the pattern of diversity behavior throughout the optimization process through graphs results and, then, evaluate how sensitive is the technique performance when associated with the population diversity behavior. This work can assist the implementation of new operators and strategies, which will permit the Differential Evolution technique to have a better performance.