Title: Global minimum structure optimisation of Lennard-Jones clusters by hybrid PSO

Authors: Yongjing Chen; Zhihua Cui; Jian Yin; Ying Tan

Addresses: Complex System and Computational Intelligence Laboratory, Taiyuan University of Science and Technology, 030024, China. ' Complex System and Computational Intelligence Laboratory, Taiyuan University of Science and Technology, 030024, China. ' Complex System and Computational Intelligence Laboratory, Taiyuan University of Science and Technology, 030024, China. ' Complex System and Computational Intelligence Laboratory, Taiyuan University of Science and Technology, 030024, China

Abstract: Structural optimisation of Lennard-Jones clusters (LJ) is a classical NP-hard problem due to the exponential increased local optima. In this paper, a hybrid particle swarm optimisation (PSO) is designed to solve this problem. To increase the escaping probability from local optimum, each particle maintains two different phase motion: attraction and repulsion in which attraction phase aims to exploitation, while repulsion motion tends to make the exploration capability. To further increase the population diversity, one-Gaussian mutation is applied to the best location found by entire swarm. Then, a well-known local search strategy, limited memory Broyden-Fletcher-Goldfarb-Shanno (L-BFGS) is employed to make an efficient local search. Simulation results show this new hybrid algorithm is effective for LJ2-LJ17 when compared with standard version and original attractive and repulsive PSO.

Keywords: limited memory BFGS; Broyden-Fletcher-Goldfarb-Shanno; L-BFGS; Lennard-Jones clusters; hybrid PSO; attractive PSO; repulsive PSO; ARPSO; Gaussian mutation; particle swarm optimisation; simulation.

DOI: 10.1504/IJMIC.2011.043154

International Journal of Modelling, Identification and Control, 2011 Vol.14 No.4, pp.303 - 309

Published online: 21 Mar 2015 *

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