[gmx-users] Langevin dynamics

[Berk Hess]

>From: "syma" <syma.khalid at bioch.ox.ac.uk>
>Reply-To: syma.khalid at bioch.ox.ac.uk,Discussion list for GROMACS users 
><gmx-users at gromacs.org>
>To: "'Discussion list for GROMACS users'" <gmx-users at gromacs.org>
>Subject: RE: [gmx-users] Langevin dynamics
>Date: Tue, 4 Jul 2006 13:59:06 +0100
>
>Hi,
>
>Thanks.
>
>Ok so some more info:
>
>I am developing a coarse-grained model of peptides based on an elastic
>network approach. Rather than run the simulations in vacuum I wanted to use
>Langevin dynamics to dampen the motions.
>
>Also, at the moment if I simulate very long, thin polymers I get big
>fluctuations in my energies (have tried NVT, NPT and NVE). I have tried to
>reduce the time step, but this does not solve the energy problem.
>
>So the simulation starts off ok- the energy is fine- but after a time, the
>ends of my polymer start to 'waggle' a lot and this is when the energy
>begins to drift. I thought using Langevin dynamics might prevent this.
>
>Any ideas would gratefully received,

Read some literature on polymers.
Polymer dynamics can be nicely decribed by normal modes.
Starting from a "perfect" straight conformation will cause
the modes to start syncronized and you should indeed see large
energy fluctuations.

Berk.