Is chaotic randomization advantageous for higher dimensional optimization problems?

Abstract

The focus of this work is the deeper insight into arising serious research questions connected with the growing popularity of combining metaheuristic algorithms and chaotic sequences showing quasi-periodic patterns. This paper reports analysis on the performance of popular and CEC 2019 competition winning strategy of Differential Evolution (DE), which is jDE, for optimization problems of higher dimensions. Experiments utilize ten chaos-driven quasi-random schemes for the indices selection and chaotic-driven crossover operations in the DE. All important performance characteristics are recorded and analyzed with simple descriptive statistics, Friedman rank tests and target-based comparisons analyzing distribution of hitting p% best minimum values over all versions and runs of jDE. The test suite was CEC 2015 in 50D. © 2020, Springer Nature Switzerland AG.