The global optimization of Morse clusters by potential energy transformations

Jon P. K. Doye (jpkd1cam.ac.uk)
Robert H. Leary (learysdsc.edu)
Marco Locatelli (locatelldi.unito.it)
Fabio Schoen (schoening.unifi.it)

Abstract: The Morse potential is a simple model pair potential that has a single parameter $\rho$ which determines the width of the potential well and allows a wide variety of materials to be modelled. Morse clusters provide a particularly tough test system for global optimization algorithms, and one that is highly relevant to methods that are to be applied to find the optimal configuration of a biomolecule. In particular, large $\rho$ values are very challenging and, until now, no unbiased global optimization method has been able to detect all the (putative) global minima at $\rho=14$ for clusters with up to $N=80$ atoms. In this paper we introduce some techniques for transforming the original Morse potential that allow us to considerably increase the efficiency in locating the known global minima and also to discover some new optimal clusters. These methods are promising candidates for application to the optimization of biomolecules.

Keywords: Morse potential; global optimization;basin-hopping; two-phase local search

Category 1: Global Optimization (Stochastic Approaches )

Category 2: Applications -- Science and Engineering (Basic Sciences Applications )

Category 3: Applications -- Science and Engineering (Biomedical Applications )

Citation: Report DSI 7/2003, Dipartimento di Sistemi e Informatica, Università degli Studi di Firenze, July/2003

Download: [PDF]

Entry Submitted: 07/15/2003
Entry Accepted: 07/15/2003
Entry Last Modified: 07/17/2003

Modify/Update this entry