[gmx-users] Inconsistent shifts using multiple bonds type 6

[Berk Hess]

>From: TJ Piggot <t.piggot at bristol.ac.uk>
>Reply-To: Discussion list for GROMACS users <gmx-users at gromacs.org>
>To: Discussion list for GROMACS users <gmx-users at gromacs.org>
>Subject: Re: [gmx-users] Inconsistent shifts using multiple bonds type 6
>Date: Tue, 06 Mar 2007 00:22:00 +0000
>
>Hi Mark thanks for your reply,
>
>I'm not sure the problem is that these restraints are inappropriate, but 
>the problem might be the sheer number. For example if i restrain a fairly 
>large number of atom distances (say 100 Calpha distances) to the same 
>length they are in the structure after minimisation and position restrained 
>MD then mdrun still crashes, even if i use a very weak force for the 
>restraint.
>
>I have also been playing round trying to fix the problem and have noticed 
>from the mailing list that inconsistent shifts are seen when people are 
>using infinite systems and have not set pbc=full. I know i am not using the 
>kind of system but setting pbc=full does not result these problems that i 
>see when i use pbc=xyz. Is pbc=full reasonable to use for production runs 
>and does anyone know roughly how much slower it is (time is quite important 
>as i have a large system)
>
>Thanks again
>
>Tom

Inconsistent shift should only occur with infinite molecules,
or when distances between atoms connected by bonded
interactions exceed half a box dimension.
Is the latter the case for your system?

You can safely use pbc=full for production runs (I only noticed
recently this option is missing in the mdp option manual).
The performance difference should be a few percent.

Berk.

_________________________________________________________________
Live Search, for accurate results! http://www.live.nl