Re: Low temperature Molecular Dynamics

From: Mert Gür (
Date: Thu Sep 16 2010 - 14:49:01 CDT

In the light of the information and reference provided by Elsa, and some
more research, I would like to rephrase my question.

I am investigating the change of the dynamics of my protein in solvent, with
respect to temperature.

In other words I am cooling my system and look at the changes in protein

I am now at 273.2K (N,V,T ensemble) but I would like to go to lower

If I perform annealing in NAMD , I will obtain supercooled water (TIP3P)

Since the simulations I am performing are on nanosecond scale they are
actually very short.

In that light, how wrong would I be in investigating the dynamics of my
protein in the supercooled water?
Does TIP3P give meaningless result at these temperatures?

There are work performed in vacuum at lower temperatures than 0 celcius.
There are also many papers on TIP4P but I couldn't find any direct info
regarding my problem in TIP3P water.For example for TIP4P, Svishchev and
Kusalik have applied an external potential to transform the liquid into
solid state.

So I would really appreciate it if anyone can give me any suggestions or
tell me what might go wrong.



On Wed, Sep 15, 2010 at 11:27 AM, <> wrote:

> Quoting Mert Gür <>:
> Dear all,
>> During minimization the temperature drops to 0K.
>> So, how meaningful would it to do a simulation at a lower temperature than
>> 273.2K. Lets say at 100K.
>> I am asking because of the liquid/solid transformation that should happen
>> at
>> 273.2K.
>> Since I use water as a solvent,will it be supercooled?
>> ...
> Actually the minimization procedure doesn't involve any temperature at all,
> what it does is trying to find the minimum value for the potential energy
> function (no kinetics involved, ergo, no temperature) using some algorithm
> (NAMD uses conjugated gradient).
> Dynamics 'brings back' temperature, and here it's a question of how good
> are the force-field parameters for the particular temperature you want to
> simulate. The TIP3 water model (usually taken as default in NAMD), is OK for
> liquid phase but not for solid or gas (or for transitions between phases)..
> E.
> --
> Elsa S. Henriques e-mail:
> Investigadora Auxiliar
> Biologia Estrutural e Computacional (@CNC)
> e Grupo de Química Estrutural (@CCC)
> Cv15 - RMB Lab
> Departamento de Química - Universidade de Coimbra
> 3004-535 COIMBRA Portugal
> Tel.: +351 23 9852080 - ext.417
> Fax : +351 23 9827703

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:54:31 CST