A Fully Anisotropic Formulation of Stochastic Cell Rescaling

TL;DR

This paper introduces a new anisotropic stochastic cell rescaling method for molecular dynamics that improves volume fluctuation control and reduces oscillations during equilibration, applicable in production runs.

Contribution

It develops the first-order anisotropic stochastic barostat that overcomes limitations of second-order algorithms and can be integrated with existing codes.

Findings

Robust across various input parameters.

Efficient control of volume autocorrelation time.

Produces correct physical cell fluctuations.

Abstract

Anisotropic barostats are employed to carry out Molecular Dynamics simulations where the volume is allowed to fluctuate with no constraints on the shape of the simulation cell. Most of these algorithms are based on second-order differential equations and share some common drawbacks, namely they can lead to slowly damped oscillations in the equilibration phase, and they do not allow to control efficiently the volume autocorrelation time. This work develops the anisotropic version of stochastic cell rescaling, a first-order stochastic barostat that overcomes these limits and can also be employed in the production phase, resulting in the correct physical fluctuations of the cell. The algorithm can be easily implemented in the existing codes on top of the anisotropic Berendsen barostat. The validation tests, performed on a number of crystal systems, show that the method is robust against wide variations of the input parameter, which allows an efficient control of the volume autocorrelation time.