Summary

Proceedings of the 2012 International Symposium on Nonlinear Theory and its Applications

2012

Session Number:A4L-A

Session:

Number:215

Continuous Global Minimization Method Based on Special Mathematical Structure of Objective Functions and Adjacent Local Minima Search

Hideo KANEMITSU,  

pp.215-218

Publication Date:

Online ISSN:2188-5079

DOI:10.15248/proc.1.215

PDF download (278.9KB)

Summary:
We introduce mathematical structure of local minima, and propose a concept of adjacent local minima in univariate multimodal functions. We rewrite the our previous mathematical structure using the concept. We propose an method for finding the global minimum of a multivariate function whose univariate function on line search is almost lower unimodal sequence. We show using a numerical example that the method effectively finds the global minimum with only a few function evaluations.

References:

[1] Dzemyda, G. ,Šaltenis, V. and Zilinskas, A. :“Stochastic and Global Optimization,” Kluwer Acad. Publ., Dordrecht, 2002.

[2] A. Hedar and M. Fukushima: “Minimizing multimodal functions by simplex coding genetic algorithm,” Optimization Methods and Software, vol.18, 265-282, 2003.

[3] Kanemitsu, H., et al.: An Algorithm for Globally Minimizing a Function with Local Minimal Values whose Sequence is Almost Lower Unimodal, Proc. of NOLTA'2009, 580-583, 2009.

[4] Kanemitsu, H.: “Studies on Continuous Global Optimization Methods using Mathematical Structure of Problems,” Doctoral dissertation, Hokkaido University, 2011(In Japanese).

[5] C.T. Kelly, “Iterative Methods for Optimization,”SIAM, 1999.

[6] N.-G. Kim, Y. Park, J.-W. Kim, E. Kim and S.W. Kim, “Digital Pattern Search and Its Hybridization with Genetic Algorithms for Bound Constrained Global Optimization,” IEICE-EA, vol.E92-A, 481-492, 2009.