Title: Interpolated differential evolution for global optimisation problems

Authors: Musrrat Ali, Millie Pant, Atulya K. Nagar

Addresses: Department of Paper Technology, Indian Institute of Technology Roorkee, Saharanpur campus, Uttrapradesh 247001, India. ' Department of Paper Technology, Indian Institute of Technology Roorkee, Saharanpur campus, Uttrapradesh 247001, India. ' Intelligence and Distributed Systems Lab, Department of Computer Science, Liverpool Hope University, Hope Park Liverpool, L16 9JD, UK

Abstract: Differential Evolution (DE) is a popular metaheuristics for global optimisation, but little research has been done on its initial population generation. The selection of the initial population is important, since it affects the search for several iterations and often has an influence on the final solution. In this study, quadratic interpolation is used in conjugation with pseudorandom numbers to generate initial population for DE. The proposed algorithm named Quadratic Interpolation DE (QIDE) is validated on a set of 20 benchmark problems. Numerical results show the competence of the proposed scheme in terms of convergence rate and average CPU time.

Keywords: metaheuristics; differential evolution; crossover; initial population; random numbers; global optimisation; quadratic interpolation.

DOI: 10.1504/IJCSM.2010.037450

International Journal of Computing Science and Mathematics, 2010 Vol.3 No.3, pp.298 - 315

Published online: 13 Dec 2010 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article