Combining Local Surrogates and Adaptive Restarts for Global Optimization of Moderately Expensive Functions.

This study combines the ideas of using local surrogates and a restart mechanism to improve the algorithm runtime and optimization efficiency of iterative surrogate global optimization frameworks. The proposed framework, LSOR (Local Response Surface based Surrogate Optimization with Adaptive Restarts), is a modification of the DYCORS framework proposed by Regis and Shoemaker [1], and uses locally fitted RBF (Radial Basis Function) surrogates, candidate search and restarts during the optimization. The key purpose of LSOR is to enable many function evaluations without increasing algorithm run-time complexity. Hence, LSOR is suitable for parallel optimization settings and for moderately expensive problems. Numerical results on ten test problems with a budget of 5000 evaluations show that LSOR has comparable accuracy to DYCORS with a significantly lower algorithm runtime. [ABSTRACT FROM AUTHOR]

Copyright of AIP Conference Proceedings is the property of American Institute of Physics and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)