[gmx-users] Umbrella sampling: issue with COM distance far from initial value

[Wes Barnett]

On Tue, Oct 17, 2017 at 8:05 AM, Giuseppe Léonardo Licari <
Giuseppe.Licari at unige.ch> wrote:

> Dear all,
>
> I have troubles with umbrella sampling. During the umbrella sampling
> simulation the COM distance between the two solutes is going too far from
> the set initial value. I tried to increase a lot the force constant of the
> harmonic potential. I also tried to change "umbrella" to "contraint", with
> no success. How can I fix the COM distance between the two solutes? It
> seems that increasing the force constant does not work. Am I making some
> mistake in the input? I have Gromacs v. 5.1.2 and here you can find the
> files of a test simulation:
>
> https://www.dropbox.com/s/cl7au3twx8d6iip/test.zip?dl=0
>
> As you see from the trajectory the two solutes go very far from the
> initial COM distance. Does somebody have an idea about this?
>
> Thanks for the help and have a nice day.
> Regards,
> Licari
> Giuseppe
>

I viewed the trajectory, but I can't see anything wrong. The two solutes
seem to maintain a COM to COM distance that they started with. Note you are
only doing the pull code on the z-vector so they are allowed to move as far
as they want in the x and y directions from each other. Also your are using
direction-periodic, which is probably not what you want.

Use a bigger box, and use the distance method, or if you want to just do it
along the z-vector (with direction, not direction-periodic) you will
probably have to use a flat-bottom restraint to keep your solutes along the
z-axis.

-- 
James "Wes" Barnett
Postdoctoral Research Scientist
Department of Chemical Engineering
Kumar Research Group <http://www.columbia.edu/cu/kumargroup/>
Columbia University
w.barnett at columbia.edu
http://wbarnett.us