Funnel hopping: Searching the cluster potential energy surface over the funnels

Abstract

We designed a cluster surface smoothing method that can fast locate the minimum of the funnels in the potential energy surface (PES). By inserting the cluster surface smoothing approach into the gradient-based local optimization (LO)-phase and the global optimization (GO)-phase as a second LO-phase, the GO-phase can focus on the global information oWalesf the PES over the various funnels. Following the definition of “basin-hopping” method [D. J. and J. P. K. Doye, J. Phys. Chem. A 101, 5111 (1997)], this method is named as “funnel hopping.” Taking a simple version of the genetic algorithm as the GO-phase, the funnel-hopping method can locate all the known putative global minima of the Lennard-Jones clusters and the extremely short-ranged Morse clusters up to cluster size $N=160$ with much lower costs compared to the basin-hopping methods. Moreover the funnel-hopping method can locate the minimum of various funnels in the PES in one calculation.