Re: Possible way of performing a temperature gradient??

From: Patrick Yee (
Date: Tue Jun 15 2010 - 07:26:45 CDT


If you need a temperature gradient, one possible idea would be to use
langevin dynamics to set one temperature on a fixed surface in a group
of atoms in one physical location and have another a different
temperature on a fixed surface in another group of atoms. A natural
gradient would be created between them. You would have to look at
issues with periodicity in this case.

This idea is fairly specific and would probably only work in a few
temperature gradient cases. What sort of simulation are you trying to
perform and why do you want a temperature gradient?


On 6/15/10, Jérôme Hénin <> wrote:
> Hi,
> The only way that I think would work is to write your own custom
> thermostat based on TclForces of (probably better) TclBC.
> Jerome
> On 14 June 2010 06:19, Ranyere Deyler <> wrote:
> > Dear NAMD users,
> > the first time I asked if there was some way of performing a temperature
> > gradient in a system using NAMD. Didn't have an answear, just that would be
> > better if i used LAMMPS.
> >
> > Well, I was thinking and saw a possible way, but I don't know if it would
> > work. And i would like to know your opinions.
> >
> > So, I have a group of atoms that interact with each other, so first i thougt
> > of applying the langevin dynamics in a group of atoms, and by the
> > interaction of the atoms it would create a natural gradient of temperature,
> > that cannot be controled, but in my case i don't need to control it anyway,
> > i jsut need a gradient.
> > This i think will work, but what i want to know if would work is this: Apply
> > the langevin to one group of atoms, and apply the coupling in other group of
> > atoms, so i could set different temperatures to the langevin dynamics and to
> > the coupling (is that possible, or the temperature has to be the same to
> > both??), so i think that would be more efficient (if it worked) to create a
> > gradient.
> >
> > So can it be done? You think it would work? I know i could test it, but my
> > system is relatively large, and perform a simulation in the dark is not an
> > option. Thanks a lot.
> >
> >
> > Ranyere Deyler Trindade
> > Universidade de Campinas
> > Brasil
> >

Pat Yee - ChEg 2011
Vice President of Am. Inst. of Chemical Engineers at ND
Director of Interview Preparation for ND Consulting Consortium

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:54:14 CST