Chapter 3 - Control techniques of molecular dynamics simulation

Abstract

This chapter presents the main control techniques involved in a classical molecular dynamics (MD) simulation. Proper MD simulation control begins with the selection of a thermodynamic ensemble, which is typically the microcanonical (NVE) ensemble, canonical (NVT) ensemble, isothermal–isobaric (NPT) ensemble, or grand canonical (µVT) ensemble. For the NVT and µVT ensembles, temperature control is achieved by coupling the system with a thermostat. For the NPT ensemble, in addition to the thermostat, a barostat is also indispensable to achieve efficient control of the system pressure. For temperature control, various classical thermostats, such as a thermostat based on simple velocity rescaling, the Gaussian thermostat, Berendsen thermostat, Andersen thermostat, Langevin thermostat, and Nosé–Hoover thermostat, are introduced in this chapter with their respective merits and drawbacks highlighted. The effects of these thermostats on the dynamics of equilibrium MD simulations are also discussed. For nonequilibrium MD simulations, a stochastic boundary condition technique is utilized to achieve efficient temperature control and minimal dynamic disturbance. Four typical barostats are introduced for pressure control, including the Berendsen barostat, Andersen barostat, Parrinello–Rahman barostat, and Martyna–Tuckerman–Tobias–Klein barostat. To better simulate the conditions under which experiments are conducted, nonperiodic boundary conditions are usually invoked in MD simulations. These conditions include the repulsive boundary condition, fixed boundary condition, and semirigid atomistic boundary condition, which are illustrated herein. The rigid bond constraint, which accelerates the simulation of molecules involving light atoms, is also discussed.