Global optimization of Morse clusters with a heuristic algorithm based on the Strongin method

Abstract

This article presents a new heuristic method for the global optimization of Morse clusters. The algorithm uses multiextremal optimization of a function of one variable, a known method developed by R. G. Strongin. The proposed approach to the parametrization of the problem of structural optimization of Morse clusters makes it possible to significantly reduce the complexity of the original multidimensional problem of global optimization of the Morse cluster and reduce it to the problem of one-parameter optimization.The proposed method is largely based on the ideas of constructing genetic algorithms. First of all, this refers to the proposed method of encoding information as well as to reducing the problem of structural optimization to the task of optimizing a function of one variable. In this case, instead of the usual genetic operations, such as mutation and crossover, we used the R. G. Strongin method for global optimization of a function of one variable. This significantly reduced the algorithmic complexity of the problem being solved.