thermo stat in Gay Berne model


  Recently I'm using Lammps for the Gay-Berne model to simulate liquid-vapor coexistence phase of uniaxial ellipsoidal particles. When measuring the density of the liquid phase, I found that only the Langevin thermostat can reproduce the density reported by the literature in which a rescaling thermostat is used. The NVT/ASPHERE, or the Noose-Hoover thermostat cannot match the literature. Could anyone give a remark on it?

The references are
Miguel et al, J. Chem. Phys, 105 (1996) 4234
Emerson et al, Chem. Phys. Lett. 271 (1997) 241-246
Martin del Rio, Phys. Rev. E 55 (1996) 3

Additionally, the reduction of the degree of freedom has an effect on the NVT system, but not the Langevin system. Because the particle is uniaxial, sometimes one rotational degree of freedom is removed to account for such symmetry.

Thank you in advance.


If you are using nvt/asphere, then you are setting
the density via your choice of box size. So not

sure what you mean that you aren’t reproducing the

Also I assume you are monitoring the temperature
with compute temp/asphere. It should use the correct

DOF for uniaxial. Not clear on what you are saying
the diffference is for nvt/asphere vs langevin. I assume
you are using the “angmom” option with fix langevin?


Dear Steve,

Thanks for your response. I’m simulating liquid-vapor coexistence system in which liquid and its vapor coexist-there is phase separation. So initially the particles fill half of the simulation box. As the system equilibrates, some of the particles escape the bulk and we start to see two phases. The densities of the liquid and vapor phase should relax to some values which are functions of the temperature. Thus thermostat matters. I’m trying to reproduce the reported densities of the liquid phase. And that’s why I raise this question concerning the thermostat.

About degree of freedom: I didn’t make it clear. Actually when I do Nose-Hoover thermostat, the reduction of DOF affects the equilibrium density of the liquid phase. Whereas if I use Langevin themostat, the reduction of DOF has no effect on the density of the liquid phase. I didn’t use “angmom” to modify my Langevin thermostat.

Thank you again for your interest in my question.



If you don’t use angmom on fix langevin then it is only
thermostatting the 3 translational DOF of each particle.
Which is not what fix nvt/aphere is doing. Either way you
should monitor the temp of the system via compute temp/asphere
if you want to include the rotational DOF in the temperature.