[gmx-users] Re: High density after NPT

[Justin Lemkul]

On 12/21/12 12:46 PM, zugunder wrote:
> Justin Lemkul wrote
>> Depends on the algorithm.  Pressure is a fickle metric (see
>> http://www.gromacs.org/Documentation/Terminology/Pressure and previous
>> discussions on this list).  It is generally advisable to run equilibration
>> using weak coupling (i.e. Berendsen) methods, then switch to a more robust
>> thermostat and barostat for further equilibration and data collection.
>>
>> -Justin
>
> Thanks for the explanations, Justin. Talking of thermostats, after some
> reading I realized (correct me please if I miss something), that Nose-Hoover
> and V-rescale are similar in a sense they introduce corrections into kinetic
> energies (though they use different approaches). From this prospective, is
> there any preference for any of them for a simple protein-water system?

Probably not in most cases.  I can't recall any existing side-by-side comparison 
that would make a recommendation for such a "standard" system.

> I am asking because in some publications they recommend Nose-Hoover as the
> only alternative to simple Berendsen, but when I tried to use it in

Depending on how old the paper is, that may have been true at one time.  It is 
certainly not true in a general sense that the Nose-Hoover method is the only 
alternative to Berendsen.  One can, of course, use V-rescale (Bussi thermostat) 
or use stochastic dynamics as a means to control temperature.

> production MD I got a warning on incompatibility of Nose-Hoover with a
> leap-frog integrator, so it switched to a different mode. It doesn't seem to
> be a real problem, but since you used V-rescale for both NPT and production
> in your tutorial on lysozyme, I think you had certain reason for that.
>

Not really.  During equilibration, one should not use Nose-Hoover until the 
system is reasonably stable, so V-rescale represents a good thermostat to use 
during that stage, but so is Berensen in that case.  For the tutorial, I wanted 
to avoid a whole bunch of parameter switching that would lead to a bunch of 
questions about why I did it, so I stuck with V-rescale.

-Justin

-- 
========================================

Justin A. Lemkul, Ph.D.
Research Scientist
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================