[gmx-developers] Non-periodic interaction in a periodic box

Carsten Kutzner ckutzne at gwdg.de
Thu Aug 21 09:55:32 CEST 2014


On 21 Aug 2014, at 09:11, Berk Hess <hess at kth.se> wrote:

> Hi,
> But there is another issue here. With domain decomposition different parts of the molecules are on different MPI ranks, so you will have to decide where to calculate the interaction and do extra communication (in general), since the molecule is probably larger than the cut-off distance.
> The pull code solves both issues by calculate the COMs using so-called "pbc" atoms. The distance for each COM is determined, with pbc, relative to a chosen "pbc" atom. The COM is then the coordinate of the pbc atom plus the COM distance from this atom. All ranks then do the, identical, pull force calculation. But this only works, relatively, efficiently when there are few pull coordinates.
> If you have few of those interactions, you could use a similar mechanism, but that will require some coding.
I believe the mechanisms in mdlib/groupcoord.h already do what is needed without any restraining.
If you define a group of atoms on which these special interactions should work, e.g. the protein,
then the ‘communicate_group_positions()’ routine will put these positions in an array
readable on all nodes. The PBC representation of these atoms will always fit the starting 
structure that is found in the .tpr, none of the group’s atoms will do a PBC ‘jump’, but
have a continuous trajectory in time (at least in the ‘xcoll’ array provided by the routine).

It does not matter how tight the box is then.

These mechanisms parallelize quite well (they are used in essential dynamics and the rotation code).


> Cheers,
> Berk
> On 08/20/2014 05:21 PM, David van der Spoel wrote:
>> Hi,
>> I wonder whether it is possible to define special interactions, say between all atom-pairs in a protein, not considering PBC. That is, to have a box which is slightly larger than the protein but not twice as large in all dimensions, and define these special interactions without considering PBC. This then assumes that the molecule is non-periodic itself.
>> Previously this would have been possible using the graph mechanism, but with domain decomposition molecules are split into atoms when they move over the box boundary, right?
>> An ugly way could be to restrain the protein in the center of the box and ignoring PBC for the special interactions, but this would not conserve energy, and be dependent on the location of the protein which one can not be sure of when computing the interaction.
> -- 
> Gromacs Developers mailing list
> * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_List before posting!
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-developers or send a mail to gmx-developers-request at gromacs.org.

Dr. Carsten Kutzner
Max Planck Institute for Biophysical Chemistry
Theoretical and Computational Biophysics
Am Fassberg 11, 37077 Goettingen, Germany
Tel. +49-551-2012313, Fax: +49-551-2012302

More information about the gromacs.org_gmx-developers mailing list