Re: [AMBER] constant pH md

From: Jason Swails <>
Date: Mon, 15 Jan 2018 21:33:10 -0500

On Thu, Jan 11, 2018 at 2:44 AM, Maciej Maciejczyk <> wrote:

> Dear amber users,
> I have a question regarding explicit solvent constant pH simulation of
> protein. The problem is that mean temperature printed in output file is
> well below requested temperature of 300 K. I assume that it has
> something to do with solvent relaxation step during which the protein is
> frozen.
> The question is whether it is only a problem with temperature showed in
> the output file, or is it a deeper problem of incorrect coupling with
> the thermal bath. Can I just ignore this issue?

‚ÄčIf I recall correctly, the problem is, as you point out, the fact that the
temperature is computed at a time when the store velocities of the frozen
solute is 0, but they're still included in the degrees of freedom count so
the temperature is artificially depressed.

Of course, you can check this pretty easily by printing a velocity
trajectory and computing the temperature of *just* the solvent for each
snapshot. The temperature should come out to close to 300 (subject to
typical statistical mechanical fluctuations, of course). If this is the
case, then you can be pretty confident that solute is also at the correct
temperature as well (courtesy of the equipartition theorem).


Jason M. Swails
AMBER mailing list
Received on Mon Jan 15 2018 - 19:00:02 PST
Custom Search