Dear lammps users,
I have a question regarding how accurate the Nose-Hover thermostat implementation is in very low temperature (near zero [LJ units]). I am using fix deform to shear a system in a very low temperature via nvt/sllod. I have observed interesting phenomena in the simulation, but I want to check if the implementation of nvt/sllod in lammps allows very low temperature. I once read that it does not allow absolute zero temperature.
Thanks in advance.
best regards.
There is nothing inherent in a Nose-Hoover thermostat (nvt or nvt/sllod)
that is different when the temperature is lo vs hi. So long as it is
non-zero. If you plot thermosttated temp as a function of time
for your dynamics, what does it look like?
Steve
I agree that the LAMMPS implementation of Nose-Hoover NVT dynamics will work equally well at any specified temperature greater than zero. However, if you attempt to thermostat the system at T=0, all of the particle velocities will approach zero and the dynamics of the system is not well-defined. Also, if you set the temperature to a very small value, the characteristic timescales of the system dynamics will become very large and you will have to be careful to scale all of the timescale-dependent simulation parameters (timestep, thermostat Tdamp, etc.) accordingly. Finally, there is the difficulty that at low temperatures, quantum effects become prominent and real systems are no longer well described by classical MD. For all of these reasons, people tend to avoid using MD in the regime kT/delPE << 1, where delPE is some characteristic potential energy difference in the system. Better to use non-dynamical LAMMPS methods, such as minimization and random atom displacements to sample curvature.
Aidan
p.s. One more thing. SLLOD was developed for fluid systems, which rules out low temperatures.