Does "drag" in NVT affect to the NVE system following the NVT


I am calculating interfacial thermal conductance between two materials–let’s say two groups in LAMMPS–in an equilibrium system.

To set the equilibrium system, I do ‘fix nvt’ first and ‘fix nve’ follows.

I wonder whether “drag” keyword would affect to the following nve system or not.

For more details on this question,

I equilibrated the system by using ‘fix nvt’ command and ‘drag’ keyword, because it took too long time to make both groups at the same temperature. Even if the whole system’s temperature is equilibrated by controlling ‘Tdamp’ and ‘drag’ keyword, the temperature of both group has a gap. I used drag upto 2.0. After waiting both groups become similar temperature, I calculate the conductance from the data of nve system. ( I do unfix nvt and fix nve. ) Then, the results seem like they are not interacting each other and all the fluctuation is just random without any correlation.

Because it takes too much time for a bigger system to be equilibrated, I changed the equilibration technique by using Langevin and nvt.

I certainly got much faster equilibration with Langevin only, and nve following the Langevin gives a somewhat reasonable value. ( I do fix nve & fix Langevin, and unfix both and fix nve only ) One thing, that Langevin may not give a very rigorous nvt ensemble due to random change of velocity, made me add nvt followed by Langevin, without ‘drag’ this time though. This equilibration with sequential Langevin and nvt seems also to give a reasonable value.

Here comes the question. Then, why does that equilibrium with ‘nvt only’ and ‘drag keyword’ affect to the following nve ensemble?

My guess is that suppression on temperature fluctuation and energy in nvt system led the limited interaction in nve system, but it is not fully comprehended.

Or I may be obsessed in the drag’s role, and equilibration with ‘nvt only’ may be a problem. Still, I cannot fully comprehended the reason why using ‘nvt only’ and waiting for a long time gives non-interacting value.

Could you give any idea, if you know how ‘drag’ interferes the system’s energy and affects to the system?

Thanks in advance.