[gmx-users] Restrain the COM of a reference group in Umbrella sampling

[zifeng li]

Hi All,

I'm calculating the PMF of pulling a surfactant molecule from a polymer
slab in aqueous phase. Therefore, my pull group is the surfactant and the
reference group is the slab. To prevent the reference group (slab) from
being dragged by the pull group (surfactant), I would like to restrain the
COM of my reference group, instead of restraining all atoms in the
reference group.

I did this by removing the COM motion of my reference group using
"comm_grps" mdp options.
In addition, I also removed the COM motion of other atoms, including waters
and ions in the system so that there are no external forces applying to the
system.
Therefore, I have two "comm_grps": slab  Waters and ion.

As you may notice, my "comm_grps" does not include the pull group
(surfactant), which triggers a warning by grompp, saying " Some atoms are
not part of any center of mass motion removal group.
  This may lead to artifacts."
I could have included the surfactant in the group of waters and ions but
the fact that this surfactant is moving by an umbrella potential refrains
me from doing so.

My question is:
1. Does this seem to be a reasonable choice to restrain the COM of a
reference group?
2. I have two groups in the removal of COM motion. Does this suggest I
should also use the same two groups for temperature coupling?
3. For a MD step using a leapfrog algorithm, when does the COM removal
happen? Does it occur after applying corrections due to constraints and
before re-scaling the coordinates and box due to pressure coupling?

cheers,