[gmx-developers] nose hoover thermostat induces resonance

[Berk Hess]

Yes, I would guess you are getting coupling here.

1 ps is also very short for the coupling period.
And maybe it would also be better to not have the T and P coupling times
equal.
Since the coupling "efficiency" does not really depend on the coupling time
with second order coupling, you can make them much longer.
I would try tau_t=5 ps and tau_p=10 ps.

Probably Nose-Hoover chains, as available with the new VV integrator,
would avoid such problems.

Berk

David van der Spoel wrote:
> Hi,
>
> we are simulating some liquids with OPLS and GAFF using Leap frog
> integrator, Nose Hoover (tau_T = 1 ps) and Parrinello Rahman (tau_P =
> 1 ps) and find that in some cases the Angle potential energy couples
> to the Temperature. This leads to a resonance where the temperature
> fluctuates with an amplitude of roughly 150 K (for 216 molecules) and
> the system freezes.
>
> Could this be a side effect of the coupling constants?
>
> Maybe the angle frequency matches the tau_T?
>
> Any other clues?
>