Multidimensional Global Search Using Numerical Estimations of Minimized Function Derivatives and Adaptive Nested Optimization Scheme
- 44 Downloads
This paper proposes a novel approach to the solution of time-consuming multivariate multiextremal optimization problems. This approach is based on integrating the global search method using derivatives of minimized functions and the nested scheme for dimensionality reduction. In contrast with related works novelty is that derivative values are calculated numerically and the dimensionality reduction scheme is generalized for adaptive use of the search information. The obtained global optimization method demonstrates a good performance, which has been confirmed by numerical experiments.
KeywordsMultiextremal optimization Global search algorithms Lipschitz condition Numerical estimations of derivative values Dimensionality reduction Numerical experiments
The reported study was funded by the RFBR under research project No. 19-07-00242.
- 22.Grishagin, V., Israfilov, R., Sergeyev, Y.D.: Comparative efficiency of dimensionality reduction schemes in global optimization. AIP Conf. Proc. 1776, 060011-1–060011-4 (2016). https://doi.org/10.1063/1.4965345
- 23.Gergel, V., Goryachih, A.: Multidimensional global optimization using numerical estimations of minimized function derivatives. Optim. Methods Softw. (2019). https://doi.org/10.1080/10556788.2019.1630624