[gmx-users] coupling to water bath

Justin A. Lemkul jalemkul at vt.edu
Sun Dec 16 14:20:33 CET 2007


Quoting Xavier Periole <X.Periole at rug.nl>:

>
> You do not want to couple the whole system to one bath. Due to the
> exchange of kinetic energy between water and protein you'll end up
> with a frozen protein and warm water.
>
> This is simply due to the different behavior of the water and the
> protein relative to the integration of the equation of motion.
> Integration errors are much more on the solvent, which warms it up;
> the use of constraints removes kinetics energy on the direction of
> the bonds and proportionally more on the protein, which cools down.
> Thus, even if these effects are very small they add up to make the
> solvent warmer than the protein cooler. Coupling the whole system
> to one bath would thus generate energy exchange to compensate the
> temperature difference. On top of this the bath would tend to decrease
> the temperature of the whole system although only the solvent is
> getting warmer. The protein ends up frozen.

I am going to analyze a few of my simulations to see this effect.  I have not
done much aqueous MD (and thus perhaps I should hold my tongue instead of
giving bad advice!)  In my membrane protein MD, my T-coupling groups have been
divided (i.e., protein, bilayer, solvent/ions), in line with your suggestion
above.  In aqueous, however, I have only worked with very small peptides and
protein fragments.  I was wondering then, doesn't one need to have a large
enough group of atoms to get proper statistics?  And how small can a group be
and still be considered large enough to have its own T-coupling group?

Thanks for any insights; your recommendations above are very interesting.

-Justin

>
> I would suggest to Justin to check the temperature of your protein and
> your solvent separately. You should see this effect.
>
> The suggestion of Mark to couple the ion to the part of the system it
> belongs to (protein or solvent) is certainly the best.



========================================

Justin A. Lemkul
Graduate Research Assistant
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul at vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin/

========================================



More information about the gromacs.org_gmx-users mailing list