problem with langevin theromstat

Hi guys

I wanna simulate argon flow in a nano channel.I used fix temp/rescale which uses Nose_Hoover thermostat and got good results.
But when I change only this thermostat to fix/langevin I get abnormal results.This is my code for thermostat:

Noose Hoover:
compute mobile flow temp
velocity flow create 1.1 482748 temp mobile #units box
fix 1 all nve
fix 2 flow temp/rescale 10 1.1 1.1 0.005 1.0
fix_modify 2 temp mobile

Langevin:
compute mobile flow temp
velocity flow create 1.1 482748 temp mobile #units box
fix 1 all nve
fix 2 flow langevin 1.1 1.1 1 699483 zero yes
fix_modify 2 temp mobile

I simulated couette flow,but velocity slip is so high and velocity profile is not linear.
Could you help me with this thermostat?
By the way I notified that Nose Hoover thermostat keeps temp much better than Langevin?!!!

Langevin is a notoriously bad thermostat for flow simulations, and
Nosé-Hoover behaves usually better, as long as hydrodynamic velocities
remain reasonably small compared to thermal velocities.

However fix temp/rescale *does not* implement Nosé-Hoover thermostat
but a simple rescaling algorithm (have a closer look at the
documentation). You want to use fix nvt instead. For any of those
thermostats, one can reduce the effect of the thermostat on the
velocity profile by applying it only to the degrees of freedom
perpendicular to the flow, using fix_modify together with compute
temp/partial.

Best,
Laurent

To add to Laurent's comments, see section
6.16 of the doc pages about the temp computes
subtracting a bias from the particles as part
of being used in a thermostat. You can use
compute temp/partial or compute temp/profile
in this way with a flow to thermostat the degrees-
of-freedom that you want, i.e. to not affect the flow.

Steve