DEVELOPMENT OF A PARALLEL METHOD OF ADAPTIVE MULTISTAGE REDUCTION |
6 | |
2013 |
scientific article | 519.853.4 | ||
216-222 | multiextremal optimization, adaptive decision rules, parallel computing, Lipschitz constant evaluation |
Multiextremal optimization methods based on an adaptive multistage dimensionality reduction scheme are con-sidered. These methods are extended for the generalized case of parallel computing and some effective implementa-tion problems are discussed. To speed up calculations, a new procedure is developed to estimate local Lipschitz con-stants using the additive convolution technique. The results of numerical experiments are given that prove the effec-tiveness of the proposed computational schemes. |
![]() |