[gmx-users] output unimaged trajectory?

Berk Hess gmx3 at hotmail.com
Wed Jul 14 11:16:07 CEST 2004

>From: david.evans at ulsop.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] output unimaged trajectory?
>Date: Wed, 14 Jul 2004 10:00:22 +0100
>Thanks for replying, probably these questions are insignficant, but I
>would like to understand anyway.
> >
> >I don't know what unimaged means exactly.
>I mean with the coordinates of all atoms propagated continuously in
>3D space, not wrapped around to the other side of the periodic box
>when they cross a boundary.

OK, this is what trjconv -nojump does.

>This is where Amber and Gromacs seem to give different results. I'm
>removing (according to my understanding of the docs  of both
>programs) the motion of the centre of mass of the whole system
>(water+solute). Amber, which outputs 'unimaged' trajectories, gives
>an untranslating solute molecule, with the waters moving gradually
>away from it in all directions. In gromacs the solute translates.

So you are sure you are removing only the center of mass of
the whole system in both Amber and Gromacs?
(just one com group in Gromacs and com mode linear,
I have no idea about Amber)

> >Any type of imaging (in mdrun or trjconv) will not make a molecule
> >which is a center of mass motion removal group move
> >(unless you let trjconv center on a moving group).
> >
>This being the case, it suggests to me that there is some difference
>I don't understand between how the COM motion is removed in the two
>programs. Probably this makes no difference to any actual result, but
>I 'd like to understand everything before I start my huge batch of
>simulations :)

Well, I hope it does not make a difference to the actual results,
but that depends on where the difference comes from.
If you only remove the com motion of the whole system, each molecule,
including the solute can translate. How much depends on the length
of your simulation, the weight of your molecule, the solvent viscosity,
the temperature (and temperature coupling algorithm) and the initial 

> >The only choice according to me (unless Amber does something strange)
> >is what you do with molecules that move out of the primary cell,
> >do you put them back on the other side (which causes jumps) or not.
> >But as I understood that is not the issue here.
>Actually that is the precise issue of my original question! Amber by
>default doesn't put them back on the other side (what I'm calling
>unimaged trajectories), whereas Gromacs does. Amber has an option to
>change to the other behaviour, but does Gromacs ?.... This might help
>me see what is going on.

No, in mdrun there is no option to change wrapping.


Play online games with your friends with MSN Messenger 

More information about the gromacs.org_gmx-users mailing list