Performance study of mode-pursuing sampling method

Since the publication of the authors' recently developed mode-pursing sampling method, questions have been asked about its performance as compared with traditional global optimization methods such as the genetic algorithm and when to use mode-pursing sampling as opposed to the genetic algorithm. This work aims to provide an answer to these questions. Similarities and distinctions between mode-pursing sampling and the genetic algorithm are presented. Then mode-pursing sampling and the genetic algorithm are compared via testing with benchmark functions and practical engineering design problems. These problems can be categorized from different perspectives such as dimensionality, continuous/discrete variables or the amount of computational time for evaluating the objective function. It is found that both mode-pursing sampling and the genetic algorithm demonstrate great effectiveness in identifying the global optimum. In general, mode-pursing sampling needs much fewer function evaluations and iterations than the genetic algorithm, which makes mode-pursing sampling suitable for expensive functions. However, the genetic algorithm is more efficient than mode-pursing sampling for inexpensive functions. In addition, mode-pursing sampling is limited by the computer memory when the total number of sample points reaches a certain extent. This work serves the purpose of positioning the new mode-pursing sampling method in the context of direct optimization and provides guidelines for users of mode-pursing sampling. It is also anticipated that the similarities in concepts, distinctions in philosophy and methodology and effectiveness as direct search methods for both mode-pursing sampling and the genetic algorithm will inspire the development of new direct optimization methods.