# Temperature without thermostat

Dear all, I am conducting a simulation of fluid transport nanochannels, an external force needs to be added to generate a stable density gradient in the system, further generating a stable flow.
To maintain isothermal conditions, need to dissipate the energy generated by external forces, the method I referred to in the literature is to “only use thermostat in the fluid part, not in the nanopore atom part”. The heat is dissipated through the collision of the two, and the temperature of the nanopore atom in this case is calculated. I am very confused that If the “fix command” is not used, the nanopore atom should not update the speed (although there is interaction between the fluid and the channel atom). So, shouldn’t the nanopore atom temperature be 0?

Hi @Tifo,

without the paper reference, it is impossible to tell. The extract you show says nothing about what is applied to the atoms of the nanopore but only to the methane atoms going through it. The paper should state how the equations of motion are solved on the nanopore.
The sentence you quote might suggest that the equations of motion for the nanopore are solved without coupling to a thermostat (which would be equivalent to using `fix nve` command). But this is only a guess from my side. This would require the definition and application of `fix nve` to a separate group containing only the nanopore atoms.

1 Like

Another possible way this could be realized in LAMMPS is to use fix nvt throughout and a temperature bias with compute temp/region applied to the reservoir region.

2 Likes

To me it sounds like you aren’t being precise in describing your simulation, which is a huge problem when you are performing a non-standard simulation.

Collisions transfer heat and equalize the temperatures of two interacting components. The thermostat acting on one component dissipates generated heat.

And one last comment: if there is net flow throughout the system, your thermostat needs to remove that component of kinetic energy to calculate the temperature correctly for the methane gas. `temp/com` might do it, but you’ll have to verify.

3 Likes

Thanks for everyone’s reply, the paper I referenced has been attached. Corresponding to “scheme 6”, in the temperature graph on the right, both the fluid and nanopore atoms have temperatures, while the nanopore atoms do not use the “fix command”. I am confused about this, at least in my opinion, the article did not mention the use of additional ensembles for nanopore atoms in scheme 6, such as “nve”. In my understanding, in this case, if the nanopore atoms do not have an initial velocity, even if there is interaction with fluid atoms, there should be no temperature

2017_A non-equilibrium molecular dynamics study of methane transport in clay nano-pores_Guan Qin.pdf (1.9 MB)

The documentation of `fix nve` states that it performs velocity Verlet integration. If you search the paper for “velocity Verlet” you will see how `fix nve` is to be used.

2 Likes