Библиографическое описание:An Investigation of the Hybridization of DE and BFGS Algorithms : доклад, тезисы доклада / A. A. Sopov, P. P. Sherstnev. - [S. l. : s. n.], 2022. - Текст : непосредственный // HYBRID METHODS OF MODELING AND OPTIMIZATION IN COMPLEX SYSTEMS : Proceedings of the International Workshop “Hybrid methods of modeling and optimization in complex systems” (HMMOCS 2022) / Krasnoyarsk Regional Science and Technology City Hall, Russia Siberian Federal University, Reshetnev Siberian State University of Science and Technology ; International Workshop “Hybrid methods of modeling and optimization in complex systems” (HMMOCS 2022) (2022 ; 22.11 - 24.11 ; Krasnoyarsk). - London, United Kingdom : European Proceedings, 2022. - P. 336-342. - ISBN 9781802969603, DOI 10.15405/epct.23021.41.
Аннотация:Many real-world global optimization problems are presented by a black-box model, therefore there is no information on properties of the objective function and its derivatives. Differential evolution (DE) is one of search algorithms, which demonstrates the high performance in solving global black-box optimization problems. Usually DE shows good global convergence, but it is slow in the local convergence. A local search can efficiently find a local optimum with high accuracy but cannot identify a basin of the global optimum. In the study we have proposed a hybrid of DE and the Broyden-Fletcher-Goldfarb-Shanno (BFGS) algorithm. In BFGS derivatives are substituted by their approximations using the finite difference method. We have tested 3 different schemes of the cooperation of DE and BFGS. Using a set of benchmark optimization problems, the experimental results have shown that the hybridization can improve the performance of the standard DE algorithm. At the same time the choice of the hybridization scheme affects the results.
