Annotation

Annotation: This paper considers population-based optimization algorithms and the dependence of its performance on the geometry of the search area shape in the solution space. A defect inherent in some algorithms is described, due to which, when the search area is disproportionately reduced, the optimization accuracy deteriorates, which is an undesirable effect and makes the use of such algorithms impractical. To detect it, a simple two-step approach is proposed, which uses a one-tailed Mann-Whitney U test and relative characteristics that make it possible to numerically assess the severity of the defect. To validate the proposed approach, more than 150 different population-based optimization algorithms are tested in this paper. The results obtained from the experiments are discussed and analyzed.