[gmx-users] shell water and epsilon

David spoel at xray.bmc.uu.se
Fri May 13 20:17:47 CEST 2005

On Fri, 2005-05-13 at 13:18 -0400, David L. Bostick wrote:
> Hi David,
> Thanks for the itp file.  Should I be running the shell models with double
> precision mdrun? At least that's what I gather from looking at previous
> postings to the list. Can one get by with single precision?
It depends on the energy surface, sometimes one does get faster
convergence (fewer iterations) in DP which compensates somewhat for the
slower calculations. In other cases we have seen that it is difficult to
get convergence at all. Try it.

> Sorry to be unclear about my question.  I suppose in comparing your paper's
> figure 5 to Lamoureux's figure 2, it seems that somehow, Lamoureaux's
> epsilon appears to converge after 500 ps, and the SW epsilon is still
> changing after 1500 ps. I meant to ask why there is this difference, even
> if you had to speculate. Also, the dielectic constant for SW, is high (or
> appears to be high), and
> in your paper, it was mentioned that it was possibly due to LJ interactions
> not making the water "sticky" enough ... connecting the diffusion
> coefficient with the high epsilon. Has there been any exploration to change
> the SW parameters a bit?
No, without changing charges (i.e. going away from the perfect monomer)
one can only modify LJ params and that we have tried without success. If
you are interested in our present plans then please contact me off the

> Is it really possible to run the Lamoureux model in the same way as was
> published?  Is it possible to have a Nose-Hoover heat bath on the
> oxygen-Drude pair's relative motion in the water's local reference frame?
> This heat bath is to be uncoupled from the global heat bath on the
> hydrogens and the center of mass of the drude pair. Their argument was that
> if the oxygen-Drude pair's heat bath was held at 1 K, the model stays close
> to the self-consistent-field regime.
Here again, please try it. I have recently run SW, the Lamoureux model
and the COS models from the Van Gunsteren group. They all work happily
in GROMACS. The thing with the shell with a mass is of course for speed.
Apparently they refer to an older paper which I haven't read to justify
it, but haven't tested it themselves. I'm pretty sure that it will
influence the properties at least somewhat, again, the proof of the
pudding is in the eating.

David van der Spoel, PhD, Assoc. Prof., Molecular Biophysics group,
Dept. of Cell and Molecular Biology, Uppsala University.
Husargatan 3, Box 596,          75124 Uppsala, Sweden
phone:  46 18 471 4205          fax: 46 18 511 755
spoel at xray.bmc.uu.se    spoel at gromacs.org   http://xray.bmc.uu.se/~spoel

More information about the gromacs.org_gmx-users mailing list