Thermostatting on both sides

Hi, everyone:

I want to bulid a sandwich structure with two solids on both sides and liquid in the moddile. The two solid walls need to be thermostated as 100 K and 140 K, and I want to see the tempureture distribution of the liquid region. I wonder which thermostatting method is best.

  1. Firstly, Itried temp/rescale, and the enegry is output. However, the linear temperature distribution can not be obtained after 3000000 steps running. What is more, unger this situation, the momentum conversion can be remained or not ?

fix 1 all nve
run 200000
fix 2 cold temp/rescale 10 100 100 0.5 1.0
fix_modify 2 energy yes
fix 3 hot temp/rescale 10 140 140 0.5 1.0
fix_modify 3 energy yes
thermo_style custom atoms step temp pe ke etotal press vol f_2 f_3

  1. I tried

fix 2 cold nvt temp 100 100 100 drag 0.2
fix_modify 2 energy yes
fix 3 hot nvt temp 140 140 100 drag 0.2
fix_modify 3 energy yes
thermo_style custom atoms step temp pe ke etotal press vol f_2 f_3

Just off the top of my head, I wouldn’t think 200000 timesteps would be anywhere close to long enough to get good averaging. If your system is small there should be lots of fluctuation in temperature. I’d run for at least 5x longer and look at the average temperature distribution then.

Point 1) What sort of a temperature distribution do you get after 3 million steps? Is it very noisy, or just nonlinear?

Point 3) What do you mean by "without the NVE command"? You need to have an NVT fix for each of the walls, and an NVE fix for the fluid, if you want to do this. I'd recommend sticking with temp/rescale or Berendsen/Langevin for now, though.