Configurational temperature and pressure molecular dynamics: review of current methodology and applications to the shear flow of a simple fluid

Abstract

Recently developed configurational methods for controlling the temperature and pressure in a molecular dynamics simulation are reviewed [C. Braga and K. P. Travis, J. Chem. Phys. 123, 134101 (2005); ibid. 124, 104102 (2006)]. Results from equilibrium molecular dynamics simulations of a WCA fluid are used to demonstrate that at equilibrium, these new methods are at least as good as their kinetic counterparts, and in some applications, may be advantageous. A finite difference based numerical scheme is presented for calculating the configurational temperature and pressure. This approach is advantageous in cases where the potential energy function may have many terms such as the kind used in general simulation codes like DL_POLY. Equilibrium molecular dynamics simulations are conducted in the isothermal–isobaric ensemble in which these numerical estimates of the pressure and temperature are used in the Nosé–Hoover feedback schemes. The results obtained demonstrate the validity of the numerical method. Non-equilibrium molecular dynamics (NEMD) simulations have been conducted at constant volume and constant pressure in order to compare kinetic and configurational methods of controlling temperature and a combination of temperature and pressure. We demonstrate that the new configurational thermostat does not give rise to a spurious string phase, unlike its kinetic temperature counterpart, but gives results in line with those obtained using a previously developed configurational thermostat. Furthermore we find that at constant pressure, there is no evidence for a string phase with our new configurational isothermal–isobaric method but a string phase is observed, albeit at a higher strain rate, using kinetic temperature and pressure control.

Acknowledgements

This work was funded by EPSRC grant number GR/R13265/01. Jerome Delhommelle (University of South Carolina, USA) is thanked for kindly providing us with his simulation data and Peter Daivis (RMIT, Australia) is thanked for useful discussions during the course of this work.