[lammps-users] Grain boundary energy oscillations - how to reduce?


I’m trying to measure a grain boundary energy (GBE) and a surface energy (SE) of ZnO.

With the proper structure built I’m running an equilibration at npt condition (Nose-Hoover thermostat). Then switching to another equilibration at nvt (also Nose-Hoover thermostat) for the last part to try to correct some periodic energy oscillations. At first, we thought were associated only with volume oscillation, but even with a fixed volume it is the potential energy that oscillated at a similar amplitude.

Does someone have any suggestions or contributions to understand and avoid these periodic energy oscillations? Some ideas we had were to choose another thermostat for the 2nd equilibration.


Such fluctuations are normal and their magnitude is coupled to both, material properties and system geometry.

The canonical way to reduce fluctuations is to average, either over space or over time. That is, you either have to simulate a larger system or look at time averaging with fix ave/time or both.

You want as little impact as possible of the thermostat since you want to measure the properties of your system and not those of your thermostat. Thus suppressing fluctuations via a different choice of thermostat is counterproductive.