Re: AMBER: temperature fluctuation in REMD

From: David A. Case <>
Date: Fri, 20 Jul 2007 09:16:46 -0700

On Fri, Jul 20, 2007, priya priya wrote:
> I am doing replica exchange method using implicit solvent, looking into the
> results summary.TEMP created by using reminfo file i found that the
> temperature is fluctuating a lot even though i am using langevin
> thermostat.I started from temp range 273K to 373K but in the output file
> there are temperature 402 K and above also.

> ntr = 0, gamma_ln = 1.0,offset = 0.09

Just a few quick points:

1. The subject of temperature fluctuations comes up a lot on the mailing

It makes a big difference how big your system is, with flucutuations being
larger for smaller systems.

2. gamma_ln = 1.0 is pretty small, although others might want to give their
experiences here. You might try some tests with gamma_ln = 5.0 to see what

3. I would think that the act of swapping conformations might also increase
temp. fluctuations as well, since a "swapped-in" conformation will take some
amount of time to re-equilibrate at the new temperature. When all the
ensembles are converged, this should not happen, as everything should then be
at equilibrium. But before everything is really converged, this result seems
possible. Others with much more experience in REMD might want to give their
experience here.


The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Sun Jul 22 2007 - 06:07:48 PDT
Custom Search