Re: [AMBER] Langevin Dynamics on selected atoms?

From: Adrian Roitberg <>
Date: Fri, 3 Jan 2014 10:52:29 -0500

Dear Jose

I think you are technically correct, but the ensemble that you create by
using a differential langevin bath is weird at best. I think one 'could'
figure it out, but it would resemble no known experiment I cam imagine.
If you truly believe that the 'rest' of the system is harmonic, you
could just run NVE for everyone and then do onw of two things

1. project out the normal modes for the modes that do not include methyl
rotations. If you can get the normal modes, this should work.
2. simply apply a filter to your FFT, which I believe is exactly the
same as applying a damping term to the dynamics.


On 1/3/14 10:46 AM, Jose Borreguero wrote:
>> Correlation functions of what?
> The incoherent structure factor I(Q,t) of the hydrogen atoms (the Fourier
> transform of the dynamic self distribution function g(r-r0,t) )
>> Why not just run NVE?
> I am running a relatively small molecule in vacuum at T=200K, thus the
> molecule behaves as a harmonic solid except for the methyl rotations. I
> wanted to measure I(Q,t) for these rotations but the low-frequency harmonic
> motions of the rest of the molecule degrade the signal coming from the
> methyl rotations. Unfortunatley, running in the NVE at low T and in vacuum
> preserves these harmonic motions.
> The harmonic motions can be removed by including interactions with external
> particles, either real (solvent) or ghost (Langevin dynamics). In
> particular, if you use Langevin dynamics with damping coefficient w_0, you
> will decorrelate any harmonic motions with frequencies under w_0, thus
> cleaning the I(Q,t) signal. Also, the external interactions will affect the
> methyl rotations weakly if they are applied only to non-hydrogen atoms. If
> you "bombard" the hydrogens with ghost particles, you will also decorrelate
> the methyl rotations.
> On Fri, Jan 3, 2014 at 12:34 AM, Jason Swails <>wrote:
>> On Thu, Jan 2, 2014 at 3:21 PM, Jose Borreguero <
>>> wrote:
>>> Dear AMBER users,
>>> Is it possible to apply the Langevin dynamics a subset of the atoms?
>> No.
>>> I need
>>> to calculate correlation functions for the hydrogens of my system,
>> Correlation functions of what? Bond vectors? I can't imagine any
>> measurements you would make in which Langevin dynamics targeted to a subset
>> of the atoms would not impact their correlation functions.
>> and it
>>> would be handy to apply Langeving dynamics only to the heavy atoms so
>> that
>>> the correlation functions are unnafected by the addition of the thermal
>>> bath.
>> Why not just run NVE? For a properly equilibrated simulation it should be
>> equivalent to an NVT ensemble as long as the settings were optimized for
>> good energy conservation (constraint tolerance, etc.). The upside is that
>> you can compute temporal properties from an NVE simulation (which seems
>> like it may be of interest to you).
>> Good luck,
>> Jason
>> --
>> Jason M. Swails
>> BioMaPS,
>> Rutgers University
>> Postdoctoral Researcher
>> _______________________________________________
>> AMBER mailing list
> _______________________________________________
> AMBER mailing list

                             Dr. Adrian E. Roitberg
Colonel Allan R. and Margaret G. Crow Term Professor.
Quantum Theory Project, Department of Chemistry
University of Florida
AMBER mailing list
Received on Fri Jan 03 2014 - 08:00:03 PST
Custom Search