Hybrid Algorithms for Electromagnetic Problems and the No-Free-Lunch Framework

Author

Quevedo-Teruel, Óscar ; Rajo-Iglesias, Eva ; Oropesa-García, Alicia

Author_Institution

Dept. of Signal Theor. & Commun., Univ. Carlos III of Madrid

Volume

55

Issue

3

fYear

2007

fDate

3/1/2007 12:00:00 AM

Firstpage

742

Lastpage

749

Abstract

A popular approach to electromagnetic design is based on generic optimization algorithms which mimic natural phenomena. These algorithms have been proved feasible through multiple experiments even though there is no standard procedure to their design for a particular problem. Hybrid schemes allow an insightful and quite efficient approach towards their design. Furthermore, an intrinsic characteristic of this class of algorithms is their randomness, which is precisely the base the no-free-lunch theorem exploits to measure their fit to a particular problem. This characterization permits a reproducible comparison between algorithms through their fitness density functions

Keywords

computational electromagnetics; electromagnetic design; generic optimization algorithm; no-free-lunch framework; Algorithm design and analysis; Cost function; Density functional theory; Design optimization; Entropy; Genetics; Histograms; Particle measurements; Proposals; Robustness; Algorithm comparison; hybrid algorithms; no-free-lunch (NFL) framework;

fLanguage

English

Journal_Title

Antennas and Propagation, IEEE Transactions on

Publisher

ieee

ISSN

0018-926X

Type

jour

DOI

10.1109/TAP.2007.891569

Filename

4120295