Thispaper presents multiDE, an extension of Price and Storn’s differential evolution (DE) algorithm that consistently outperforms state-of-the-art search techniques for identifying multiple global optima in multidimensional, discontinuous solution spaces. MultiDE automatically determines appropriate values for control parameters, and periodically updates those values at run-time. MultiDE requires little expert knowledge of the solution space, and is capable of searching both discontinuous and differentiable solution spaces. Innovative use of multiple subpopulations, minimum spanning distances, subpopulation expiration, and precision control contributes to multiDE’s speed and effectiveness. Results from several benchmark problems reveal MultiDE’s extraordinary power.
Published Date: May 2004
Registration: ISBN 978-1-57735-201-3
Copyright: Published by The AAAI Press, Menlo Park, California.