[gmx-users] Pull code with dummy atom as reference - not working

[Stella Nickerson]

I want to pull a carbon particle (170 atoms in a lattice) across an
interface between Liquid A and Liquid B.  The particle starts out in Liquid
A with the center of mass at about x = 2.5 y = 2.5 z = 2.5.  The
liquid-liquid interface is at about z = 5.  I want to pull the particle in
a straight line in the positive z direction.  So I put a dummy atom (no
interaction perameters) in Liquid B at x = 2.5 y = 2.5 z = 7.5 to use as a
reference.  I put the dummy atom in a freeze group to keep it from moving.
 This is my pull code in my .mdp file, plus the freeze group code (DMY is
the dummy atom, DNP is the carbon particle):

; Pull code
pull            = umbrella
pull_geometry   = distance      ; simple distance increase
pull_dim        = N N Y
pull_start      = yes           ; define initial COM distance > 0
pull_ngroups    = 1
pull_group0     = DMY
pull_group1     = DNP
pull_rate1      = -0.01          ; 0.01 nm per ps = 10 nm per ns
pull_k1         = 1000          ; kJ mol^-1 nm^-2

; Define dummy atom as freeze group
freezegrps = DMY
freezedim = Y Y Y

At first I used a negative pull rate because this tutorial (
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin/gmx-tutorials/umbrella/05a_pull_tips.html
)
says to use a negative pull rate if you want to decrease the distance
between pull_group0 and pull_group1.  But instead of moving in a straight
line in the positive z direction, the particle moved in the negative z
direction at an angle (could this be because of periodic boundary
conditions?)

When I tried to use a positive pull rate the particle just shuffled around
in a little x-y circle.

What am I doing wrong?