[gmx-users] temperature coupling strength

[Omer Markovitch]

On Mon, Nov 10, 2008 at 17:42, Seunghyun Chung <chungsh at umich.edu> wrote:

> Deal all,
> Is there any practical approach to choose the right temperature coupling
> strength for a simulation? For example, if a system behaves differently with
> weak coupling and strong coupling, which result I should trust?

In general, the coupling lifetime should not interfere with any fast degrees
of motion you might have.
For example, for pure water (where hydrogen bond lifetime is in the ps
region, and OH vibration is in the fs lifetime), the thermostat coupling
might be ~0.5 ps, but for a protein, perhaps a value in the area of 1 ps and
above might be more suites.
Plotting the temperature devations and/or averages might direct you to the
better value.



> Also, I thought giving a very weak nose-hoover thermostat coupling would be
> similar to the NVE simulation. However, when I run the simulation, the
> system behaves very weirdly by constantly accelerating and slowing down.
> will there be any explanation on this behavior?
>
I agree. But did you start applying the weak coupling before or after the
system was already equilibrated?


  --Omer.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20081110/f696119d/attachment.html>