# [lammps-users] ensemble problem

Dear all,

I am doing a heat transfer problem between water and graphite recently. In my case, my water is inside the graphite channel. One side of graphite wall has higher temperature and the other graphite wall has lower temperature. In the simulation, I let my graphite wall interact with each other and with water, and I am doing thermostat to the wall in order to get temperature gradient distribution in water. So I apply thermostat to the wall by maintaining the temperature of the graphite walls to be unchanged.

I can assign NVT to graphite wall. But I am not sure about the ensemble for my water. Because when I maintain the temperature of my graphite walls, I add or substract energy into my system. So I can’t assign NVE to water only when water reaches to be steady-state.

So my first question is that in my case, How should I assign the ensemble in an appropriate way?

And another question is also about ensemble. Our MD group has compiled code by ourselves to simulate the same heat transfer problem between fcc wall and argon liquid by only considering LJ potential before. In our case, we calcuate force based on LJ potential, and we update the atoms’ position using Verlet based on Newton second law. And we do thermostat by rescaling the velocity on the wall every timestep to maintain the temperature on the wall. We didn’t assign ensembles like NVE or NVT on it and we got reasonable results. Also, in gromacs, we don’t have to assign ensemble too. And we do time integration and thermostat seperately.

So my question is that why in lammps ensemble has to be related to time integration?

Best wishes!

Ziyuan Shi

If you thermostat both walls you should be able to just run the
water inbetween with NVE. You can also try other kinds of
thermostats on the walls, e.g. fix langevin, fix temp/rescale, or
even fix heat to create two reservoirs (the walls) that you