[gmx-users] PMF via constraint forces and frozen/restrained atoms incompatible ?

[m b]

dear gmx-users,

I would like to calculate the PMF between a surface-slab
and a solvated molecule by integrating constraint
forces. Using the pull.ppa file:

runtype = constraint
group_1 = ped
reference_group = top
reftype = com
direction = 0.0 0.0 1.0

I find that the z-distance between the center of mass of
my group_1 (the solvated molecule) and the reference group 
(top, a single frozen atom in the (frozen) surface) 
is not constrained at all to its initial value.
In fact I find that the position of the FROZEN reference atom 
(as opposed to all the other atoms in the frozen surface) CHANGES
while it remains exactly at its starting position if I 
run the same tpr file without the -ip pull.ppa option

If I define the whole frozen surface as the reference
group the system blows up after a few timesteps
(again, without using the -pi option, the simulation is
perfectly well behaved)
If I use position restraints for the atoms in the reference 
group (the whole surface slab) the system also blows up,
- again only when using the -pi option.

are frozen or position restrained reference groups incompatible
with the pull code ??
... any work-arounds ?

and finally: when i start the run gromacs asks me to choose 
a group from the index file ... why ??
the two involved groups are already defined in the pull.ppa file,
aren't they ?

thanks for any help,
Michael


__________________________________
Do you Yahoo!?
Yahoo! Hotjobs: Enter the "Signing Bonus" Sweepstakes
http://hotjobs.sweepstakes.yahoo.com/signingbonus