Monday , June 18 2018

The Influence of Experimental Designs on the Performance of Surrogate Model Based Costly Global Optimization Solvers


Dedicated to Dr. N. Andrei on the occasion of his 60th birthday

Abstract: When dealing with costly objective functions in optimization, one good alternative is to use a surrogate model approach. A common feature for all such methods is the need of an initial set of points, or “experimental design”, in order to start the algorithm. Since the behavior of the algorithms often depends heavily on this set, the question is how to choose a good experimental design. We investigate this by solving a number of problems using different designs, and compare the outcome with respect to function evaluations and a root mean square error test of the true function versus the surrogate model produced. Each combination of problem and design is solved by 3 different solvers available in the TOMLAB optimization environment. Results indicate two designs as superior.

Keywords: Black-box, Surrogate model, Costly functions, Latin Hypercube Designs, Experimental Design.

N.-H. Quttineh got a M.Sc. in Optimization at Linköping University in 2004 and has been a graduate student at Mälardalen University since 2005, with main subject Algorithms for costly global optimization.

K. Holmström got a Ph.D. in Numerical Analysis at Umeå University 1988 and did industrial research and development work for ABB between 1990 and 1993. Besides his academic career, he has been consultant in more than 100 industrial and scientific projects. Since 2001 he has been Full Professor in Optimization at Mälardalen University with main research interests algorithm and software development for industrial and financial optimization problems, in particular costly global mixed-integer constrained nonlinear optimization. Holmström is CEO of Tomlab Optimization Inc. and creator of TOMLAB, an advanced MATLAB optimization environment distributed since 1998.

>>Full text
Nils-Hassan QUTTINEH, Kenneth HOLMSTRÖM, The Influence of Experimental Designs on the Performance of Surrogate Model Based Costly Global Optimization Solvers, Studies in Informatics and Control, ISSN 1220-1766, vol. 18 (1), pp. 87-95, 2009.