[lammps-users] fix NVE problem

Dear lammps users,

Although similar thread appeared couple of days ago, I would like your opinions regarding my problem which is confusing me a bit.

I am simulating an atomistic system with lj/cut/coul/cut 14.0 cutoff. When I am running NVT and NPT everything seems to be working fine, the volume and temperature are nicely equilibrated. However, when I am using NVE after that, the temperature is constantly increasing and hence total energy is not constant. Moreover, there are fluctuations in the total as well as potential energy.

As there could be many reasons which could lead to energy shift in NVE simulations as discussed before on this forum, the one which seams most predominating as of now is due to missing interactions.

In the last NVE thread, it was suggested to use pair_modify shift yes to take care of Vdwl interactions beyond cutoff. I was wondering how to take care of or neglect electrostatic interactions beyond this cutoff value.

There is one way to keep temperature constant my using Langevin thermostat. I did and it worked too. However, as it assigns random forces to the atoms which in turn will modifies their velocities and in turn final outcome.

If you're cutting off Coulombic interactions, you are not conserving
energy by definition. LAMMPS does not shift those to 0. So you
will see fluctuations in total E as atom pairs move across the cutoff.

This can be hidden by various thermostats, a long cutoff, small
timestep, etc, but a gradual drift up in KE is normal.

Steve