scaling velocity + NVE = NVET?

Dear All

I have a question about velocity rescaling. According to LAMMPS manual:

Dear All

I have a question about velocity rescaling. According to LAMMPS manual:

fix temp/rescale command

IMPORTANT NOTE: Unlike the fix nvt command which performs Nose/Hoover thermostatting AND time integration, this fix does NOT perform time integration. It only modifies velocities to effect thermostatting. Thus you must use a separate time integration fix, like fix nve to actually update the positions of atoms using the modified velocities

==============================
If we use fix temp/rescale command combined with NVE, what parameters are kept constant? NVET?

You cannot see it like that. Using fix nve does not keep anything constant, but rather it does not adjust anything. So if there is nothing else, a simulation (with fully periodic boundary conditions) would be in NVE ensemble. Similarly fix nvt only adjusts the kinetic energy to create an NVT ensemble.

Using fix nve and fix temp/rescale combined does not put you in any defined statistical mechanical ensemble. At best it could be a called a very crude approximation of an NVT simulation.

Axel

Just to make sure, with that combination (velocity scalling + NVE) energy is not kept constant. Am I right?

Just to make sure, with that combination (velocity scalling + NVE) energy
is not kept constant. Am I right?

just to re-iterate. *no* integrator *enforces* constant (total) energy.
(total) energy conservation is a desirable feature in an integrator
and also depends on the choice of time step and potential, since
you approximate differentials with differences and use floating
point numbers with a finite accuracy.

on top of that, velocity rescaling is never a good
choice for adjusting kinetic energy if you care about
ensembles (like you seem to be doing).

axel.