[gmx-users] COM-Pull along a direction

[Pengzhi Zhang]

Dear Gromacs users,

I am trying to pull an ion out of a protein using Gromacs/5.0.4. I pull 
the ion along specific pre-determined paths.

 From my understanding, pulling along a direction _V guarantees that the 
dumb bead would move only at direction _V. The position of the dumb bead 
is simply: _r(t) = _r(0) + _V*speed*time (_ means vector). And if the 
spring constant is sufficiently large, the pulled atom (a single ion) 
would also follow the dumb bead (with fluctuations). However, I found 
that no matter how large the spring constant is (I vary it from 100 to 
1,000,000 kJ/mol/nm**2), the pulled atom deviates from the direction _V 
in a certain amount of time (usually after the pulled bead unbinds), 
sometimes it even goes perpendicular to _V and moves out of the box, 
terminating the simulation.

Does the direction _V (pull-coord1-vec) mean the direction of 
instantaneous pull force or the direction of the path that the dumb bead 
follows?

Thank you very much in advance!


Here is the pull part of my mdp file.
; COM PULLING
; Pull type: no, umbrella, constraint or constant-force
pull                            = umbrella
pull-geometry            = direction
pull-ngroups              = 2
pull-ncoords              = 1
pull-group1-name      = binding_site_atoms
pull-group2-name      = pull_ion           ; pull group
pull-start                    = yes
pull-coord1-vec          = -0.07 -0.04 0.99
pull-coord1-groups    = 1 2
pull_coord1_rate        = 0.001        ; nm/ps
pull-coord1-k             = 10000       ; kJ/mol/nm^2
pull-nstxout               = 10         ; 0.02 ps
pull-nstfout                = 10
pull-print-reference    = yes


Sincerely,
Pengzhi