COMB3 energy drain

Hi Seth,

Using fix qeq/comb breaks energy conservation – this is inevitable due to a damping term. As for the system drifting with NVT, I would use a larger supercell and see if the same problem persist.


We have observed the linear momentum buildup for a more complicated interfacial system that contained ~8500 atoms. I will run it using a larger system of bulk alumina crystal and see if the behavior changes though.

Below is a plot of the temperature and total energy over the course of 4 picoseconds for the small bulk crystalline alumina described in the first email. I knew that the damped charge dynamics led to the energy not being conserved, but I didn’t expect for the energy to drain nearly this rapidly. Is this to be expected? If so, how do you combat the energy drain; does this potential require a thermostat that can feed in energy to reach a steady state?

Thanks for your help Ray.

Seth Martin