Re: Lowe-Andersen, Langevin and velocity rescaling thermostats (lipid bilayers)

From: Jeff Comer (
Date: Wed Apr 06 2016 - 08:49:16 CDT

Dear Claire LOISON,

I would suggest if you have flows that you only use the Lowe-Andersen
thermostat. The Langevin thermostat has a preferred reference frame—it
does not conserve momentum and unrealistically suppresses flows.

It's interesting that you say "lipidic diffusion is perturbed by the
thermostat"? What leads you to this conclusion? Do you calculate
different diffusion coefficients for the lipids with and without the
thermostat? Keep in mind that lipid diffusion is very slow and you
usually need hundreds of nanoseconds of simulation (or a very large
system) to get reasonable statistics. A better solution might be to
decrease loweAndersenRate until the effects on quantities of interest

Honestly, I've never seen artifacts from the Lowe-Andersen thermostat
with default parameters (vs. NVE simulations) in any of the tests I've
done. There are probably some better momentum-conserving thermostats
out there (see,
but in NAMD, Lowe-Andersen is the only choice for simulating flows.


Jeffrey Comer, PhD
Assistant Professor
Institute of Computational Comparative Medicine
Nanotechnology Innovation Center of Kansas State
Kansas State University
Office: P-213 Mosier Hall
Phone: 785-532-6311

On Mon, Apr 4, 2016 at 7:16 AM, Loison Claire
<> wrote:
> Dear NAMD-users and developers,
> I simulate all-atom hydrated fluid lipidic bilayers, eventually with some
> flows,
> and I would like to use a thermostat which perturbs not too much the flows
> in the systems.
> I have tried the Lowe-Andersen thermostat and it seems to be
> reasonable for equilibrium diffusion in the water
> (similar diffusion in NVT and NVE ensembles),
> but the lipidic diffusion is perturbed by the thermostat.
> Therefore, I would like to try using the Lowe-Andersen thermostat for water
> atoms only.
> Would it be possible to have,
> as for Langevin or T-coupling thermostats,
> a pdb file where a given column provides
> coupling factors for each atoms individually ?
> Would that be difficult to implement in NAMD ?
> Similarly, is it possible to thermostate in chosen directions only (X,Y or
> Z...) ?
> Typically, with an electric field, I do not want to thermostate in that
> direction, but
> the results remains not too false if one thermostates in the other
> directions
> From the UG, I had the impression that it is not implemented.
> I understand that for Lowe-Andersen this is not possible, but
> maybe this would be already helpfull using Langevin thermostat...
> Naively, it seems to me that a direction-selective thermostat
> should be relatively easy to implement
> (having a zero coupling for selected directions for instance)
> but I have no experience at all with NAMD development,
> would somebody have some hint where to begin with, and
> maybe what not to forget...
> Thank you in advance for any help.
> Best regards,
> Claire Loison
> --
> =========================================
> Claire LOISON
> Chargée de Recherches CNRS
> Theoretical Physical Chemistry Group
> =========================================
> Institut Lumière Matière
> Bat. Alfred Kastler, RdC
> 43, bd du 11 Novembre 1918
> Tel +33 4 72 43 12 57
> Fax +33 4 72 43 15 07
> Cellular +33 6 31 30 11 72
> =========================================

This archive was generated by hypermail 2.1.6 : Tue Dec 27 2016 - 23:21:59 CST