Question about Preparing Flexible/Rigid Hybrid Simulations

Dear LAMMPS users,

I’ve been working on a project to study the dynamic behavior of a rigid solute in various compositions of ionic liquid/water mixtures. In order to model the system, I am using a united atom ionic liquid model developed by Cao and coworkers with TIP4P. Thus far I have been able to reproduce the densities/diffusion/radial distribution functions reported by this group. My problem lies in adding a rigid body probe to these mixtures.

Simulation boxes are prepared using packmol with a rigid probe molecule. Following minimization, I run NPT to equilibrate the system, and effectively the only difference in the submission scripts is as follows:

fix heatnpt all npt temp 10.0 400.0 100.0 iso 1.0 1.0 1000.0

becomes

fix heatrig probe rigid/nvt single temp 10.0 400.0 100.0
fix heatmob mobile npt temp 10.0 400.0 100.0 iso 1.0 1.0 1000.0 dilate all

During equilibration, the simulations develop a net velocity over the course of the simulation. I’ve considered using fix momentum to remove the center of mass velocity, but was concerned that it would ruin the dynamics of the system. Furthermore, I don’t understand why center of mass motion would occur using a Nose-Hoover thermostat, because my understanding was that this is typically a byproduct of stochastic thermostats. I think that I may be overlooking something when going from a single integrator to using two, one each for the rigid and mobile components of my simulation.

Even after equilibration, if I reset the velocities using the create velocity command, it picks up the same motion. This is not present in the mixtures not using the rigid integrator.

My questions are, am I implementing a rigid molecule in the proper manner and what are the implications of using a function like fix momentum to rezero the center of mass velocity?

Thank you very much for your help, and let me know if I need to provide any clarification.

Regards,

Brian

I would not advise using fix momentum with npt or nvt.

I would also not use rigid/nvt with a single rigid body. I don’t

know what it means to thermostat a single body. I would just

thermostat the solvent and let the rigid body evolve via NVE.

Do you get a momentum drift w/out a rigid body? Are you driving
the system (e.g. the probe body) in any other manner?

Do you get a momentum drift if you make the rigid body as

immovable?

Steve