[gmx-developers] moving frozen atoms

[Berk Hess]

>2) I find that under certain conditions frozen atoms DO MOVE
>in my simulations (NOT GOOD). This happens when I calculate a PMF
>using constraints with the reference group consisting
>of frozen atoms. The pull code seems to do something to the
>the velocities or forces of the frozen atoms (evidenve
>herefore is that the frozen atoms move only in the z-dimension
>which is the pull dimension in my case). A while ago
>Justin MacCallum wrote that the positions of the atoms ref group
>are not modified by the pull code. Did this change, or is this maybe
>a feature that generally occurs when constraining atoms in gromacs ??

I never read something from Justin saying that frozen atoms are not modified
in the pull code. As far as I know the pull code does not take frozen atoms
into account and there are just "pulled" as normal. For normal constraints
frozen atoms are treated correctly, which is done by setting their inverse
mass to zero. I guess the same thing could be done for the pull code,
but it would need to be implemented.
A workaround for now would be to use extremely high masses for the frozen
group.

Berk.

_________________________________________________________________
Protect your PC - get McAfee.com VirusScan Online 
http://clinic.mcafee.com/clinic/ibuy/campaign.asp?cid=3963