[gmx-users] Re: how is the pulling force measured

Thomas Schlesier schlesi at uni-mainz.de
Tue Apr 30 11:50:15 CEST 2013

I never looked into the code, but i understanded it the following way 
(for pull_geometry = distane, direction, position).
For the pulling the reference group is held fixed and the force is 
applied to the pulled group. Since the reference group an move during 
the rest of the simulation (all steps not involving pulling), the 
pull-stuff is updated in every step (meaning if you use 'pull_geometry = 
position' you're pulling to a point relative to the reference group. If 
the referene group moves, the point to where you pull also moves -> so 
pulling is always relative to the reference group, which leads to the 
fact the reference group is fixed for the pulling).
If you simulate with 'pbc' you should separate the movement of the 
center of mass (com) of the system. If you look into a video of the 
trajectory (vmd or similar program), it will look like that both groups 
will be pulled. But this is from the separation of the com-movement. If 
you want a 'nice' movie you can use 'g_trjconv' (i think) to define a 
group which is always in the enter of the box.

One important side note:
If you pull only in one dimension (x,y or z), there will be no 
pulling-force acting on the other two directions. In your case your 
pulled-group could move freely in the xz-plane. In my opinion it's safer 
to pull in all three dimensions.


Am 30.04.2013 09:37, schrieb gmx-users-request at gromacs.org:
> Hi Alex,
> I read the manual but I got confused with the details. My pull_geometry =
> position with pulling direction along y. I am trying to dissociate two
> interacting proteins. So I set the last amino acid of one protein as my
> reference group and the last amino acid of the other protein as the pulled
> group. Now, I don't know exactly how to interpret what reference group and
> pulled group mean, and what is their difference; since in the manual it says
> 'there is no difference in treatment of the reference and pulled group
> (except with the cylinder geometry).' Also, it says that measured force
> output is the force of the pulled group.
> So basically, I am trying to visualize how the system works. Using the
> pull_geometry stated above, are the two groups (reference and pulled)
> attached to two virtual springs? Will this statement be correct?
> "Force was applied by moving the clamped ends of the two springs (i = 1, 2)
> in opposite directions with constant velocity v to positions Zi(t) = zi(0) +
> vt, where zi(0) is the initial position of the end of the molecule. The
> forces fi(t) at the two ends were measured at every time step using the
> relationship fi(t) = k (zi(t) ??? Zi(t))."
> I hope my questions make sense. Thank you.

More information about the gromacs.org_gmx-users mailing list