Global Optimization on an Evolving Energy Landscape

Abstract

Locating the global minimum of a complex potential energy surface is facilitated by considering a homotopy, namely a family of surfaces that interpolate continuously from an arbitrary initial potential to the system under consideration. Different strategies can be used to follow the evolving minima. It is possible to enhance the probability of locating the global minimum through a heuristic choice of interpolation schemes and parameters, and the continuously evolving potential landscape reduces the probability of trapping in local minima. In application to a model problem, finding the ground--state configuration and energy of rare--gas (Lennard--Jones) atomic clusters, we demonstrate the utility and efficacy of this method.