Optimising reaction coordinates for crystallisation by tuning the crystallinity definition

Abstract

AbstractWe apply maximum likelihood analysis to optimise crystallisation measures based on Steinhardt bond order parameters. Assuming that the size of the largest cluster of crystalline particles serves as a good reaction coordinate for the freezing transition, we write down the likelihood to observe the committor values computed for a large number of configurations. We then maximise the likelihood function by varying the parameters that enter the definition of crystallinity. For the crystallinity definition considered here this parameter set consists of the thresholds for the next-neighbour distance, the strength of the crystalline bonds and the number of crystalline connections. The optimum parameter set found by the likelihood maximisation differs considerably from the parameters that are commonly used, but leads only to a marginal improvement of the quality of the reaction coordinate.Keywords: crystallisationorder parameterreaction coordinatecomputer simulationsLennard–Jones system AcknowledgementsThe simulations discussed in this article were carried out on the Vienna Scientific Cluster (VSC).FundingThis work was supported by the Austrian Science Fund (FWF) within the SFB ViCoM [Grant F41] as well as Project P 24681-N20.