Langevin thermostat in local region

Dear Steve and users,

In in.langevin, it looks like we control the temperature through the entire system and “modify” the temperature control regions for Thot and Tcold later. For energy in/out to/from system, we need to multiply 8000 (total number of atom).

Now, in my system, my liquid atoms are sandwiched by two solid plates, and I would like to create temperature gradient in the liquid atoms through the two solid plates.

Initially, I thought that we could control the two different temperatures of the solid plates using

“fix hot plate1 langevin …”

“fix cold plate2 langevin…”.

Then, when I calculate “dQ”, I considered the number of atoms of plate 1 and 2 only without including liquid atoms, and the total heat flux seems much larger than expected.

As far as I understand, I do not manipulate anything else, and I just want to check my energy balance.

Am I missing something here? Any comments/help would be greatly appreciated.

If this is not right approach, I think I need to follow the approach given in the in.langevin input file, i.e., initially include all the atoms in langevin fix and then modify it later for plates.

Thanks, Gisuk.