# Time integration without thermostatting.

Hi all,
I am doing Poiseuille flow simulations using Argon as fluid atoms and Platinum as wall atoms. I use velocity rescaling method to maintain the average temperature of wall at 300 K. I do not want to apply any thermostat or any other ensemble to the fluid domain but to merely integrate the equations of motion. In other words, I just need a time integrator without any ensemble. For that, I used the following set of commands on the fluid domain.
(Group id of fluid domain is flow)
compute flt flow temp/partial 0 0 0
fix 2222 flow nvt temp 300 300 100
fix_modify 2222 temp flt

I wish to know whether it is right or not. What I intended to do by the above commands is to first compute a temperature which excludes all the velocity components (which gives zero temperature) and thermostatting the same compute so that I do not apply any thermostat to the fluid but making use of the integrator.
I wish to know whether my approach is right, whether I can use the above set of commands or any changes to be done.

Of course not. Why not integrate the motion of fluid atoms using fix nve?

Ray

Hi all,
I am doing Poiseuille flow simulations using Argon as fluid atoms and Platinum as wall atoms. I use velocity rescaling method to maintain the average temperature of wall at 300 K. I do not want to apply any thermostat or any other ensemble to the fluid domain but to merely integrate the equations of motion. In other words, I just need a time integrator without any ensemble. For that, I used the following set of commands on the fluid domain.
(Group id of fluid domain is flow)
compute flt flow temp/partial 0 0 0
fix 2222 flow nvt temp 300 300 100
fix_modify 2222 temp flt

I wish to know whether it is right or not. What I intended to do by the above commands is to first compute a temperature which excludes all the velocity components (which gives zero temperature) and thermostatting the same compute so that I do not apply any thermostat to the fluid but making use of the integrator.
I wish to know whether my approach is right, whether I can use the above set of commands or any changes to be done.

Ranjith

Dear Ray, Thanks for the reply. First, I do not want any ensemble on the fluid domain but only to integrate. If I use fix nve, I apply nve ensemble, right? Second, could you please explain where I am wrong in the previous post? Thanks.

Fix NVE, simply refers to the velocity verlet integration where for Hamiltonian systems the energy is conserved. No extra terms from thermostats. This is consistent with the NVE ensemble.

Ranjith,

fix nve simply integrates and adds/subtracts no energy to the motion
of atoms. I do apologize that I was mistaken previously. Your three
lines of commands are correct and have the exact same effect as "fix
nve"

Ray