RE: AMBER: Andersen thermostat: target T?

From: Arantxa Sanz <aranzazu.sanz.uab.es>
Date: Mon, 24 Jan 2005 11:18:05 +0100

> -----Mensaje original-----
> De: owner-amber.scripps.edu [mailto:owner-amber.scripps.edu]
> En nombre de David A. Case
> Enviado el: sábado, 22 de enero de 2005 1:23
> Para: amber.scripps.edu
> Asunto: Re: AMBER: Andersen thermostat: target T?
>
>
> On Fri, Jan 21, 2005, Arantxa Sanz wrote:
> >
> > I have been running the following Sander script to run MD
> dynamics at
> > constant T with Andersen thermostat
> > Discarding the first 200ps, it seems that average T oscillates
> > around 310K. That is, 10K above the given target T.
>
> Try setting vrand to a lower value than its default value of
> 1000. Also, run some tests with a small value of ntpr and
> look closely at the temperature profile....maybe that will
> give some hints.
>

I had tried vrand=500, but it did not seem to change results noticeably.

> Also: be sure that you really want/need this sort of
> thermostat. It is implemented in only a very naive way.
>
> ...dac
>

I had been told that I should avoid Berendsen because it is not sampling
a proper
Canonical ensemble, and recommended to use Langevin or Andersen. So I
had decided to
Use Langevin for the heating period and then Andersen for
equilibration/production, as
Constant P was a no-no for methane as a solvent, and Langevin seemed to
be slightly unreliable
For production.

Any further suggestion on ntt choices? My goal is to obtain a free
energy profile via umbrella
Sampling method

Thanks

Arantxa

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Mon Jan 24 2005 - 10:53:00 PST
Custom Search