[gmx-users] temp and press coupling

[David Mobley]

Maria,

I remember looking into this on the list at one point, too, and I concluded
that as far as I can tell, Parrinello-Rahman probably doesn't work properly
at present. I seem to recall seeing a bunch of discussion of this on the
developers list, and I found this message from Michael Shirts particularly
concerning:
http://www.gromacs.org/pipermail/gmx-developers/2004-May/000877.html. So I
concluded I should avoid Parrinello-Rahman.

The problem with Nose-Hoover is that with the leapfrog integration method
that GROMACS currently uses, it's not reversible. I'm not sure this is
necessarily that big of a problem; it depends on what you want to do. If you
are doing something which requires reversibility obviously this is a
problem. There are other problems with Nose-Hoover, of course, as in this
message you point to (
http://www.gromacs.org/pipermail/gmx-users/2004-November/013017.html) -- but
those aren't problems with the GROMACS version of Nose-Hoover, but with
Nose-Hoover generally. See especially the Toda demon paper -- it's
particularly informative, and probably you can get an idea from there some
checks you could do to see whether it is working "OK" for the system you're
using.

I've had some experience where it works OK, but it also (as you might
expect) seems to work particularly poorly when you are applying it to
anything with a relatively small number of degrees of freedom. I've finally
ended up using Langevin (stochastic) dynamics for my calculations as a
"thermostat" because I needed something that could cope with decoupling
calculations (turning off interactions between a ligand and a protein)
without giving crazy results. I don't really trust either Berendsen or
Nose-Hoover in that case, and the Anderson thermostat isn't quite
implemented yet.

Someone else may be able to comment in nore detail on the Rahman-Parrinello
issue. I only know what I read on the list, which seemed concerning.

David Mobley


On 1/1/06, maria papadopoulou <maria_papadopoulou55 at hotmail.com> wrote:
>
> Hi
>
> As someone interested in using gromacs I have been looking at the mailing
> list and trying to decide on which  temp and presure coupling method to
> use.
> My preference would be to use Nose Hoover for temp and  Parrinello Rahman
> for pressure coupling, but I have read some posts that suggest that there
> are problems with these methods in gromacs:
>
> NH
> http://www.gromacs.org/pipermail/gmx-developers/2005-March/001062.html
> http://www.gromacs.org/pipermail/gmx-users/2004-November/013017.html
> Parrinello Rahman
> http://www.gromacs.org/pipermail/gmx-users/2003-October/007426.html
> http://www.gromacs.org/pipermail/gmx-users/2003-October/007426.html
> and in paticular the fix mentione in the following link, for which i can
> find no further detials of:
> http://www.gromacs.org/pipermail/gmx-users/2004-January/008670.html
>
> Could someone please advise me as to how serious these problems are in
> paricular when using parrinello rahman with an octahedral box and
> isotropic
> pressure coupling.
>
> I have v 3.2.1 installed
>
> Thanks
>
> Maria
>
> _________________________________________________________________
> Be the first to hear what's new at MSN - sign up to our free newsletters!
> http://www.msn.co.uk/newsletters
>
> _______________________________________________
> gmx-users mailing list
> gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20060103/5dcae96b/attachment.html>