[gmx-users] Constraints without any meaningful potential

[jgoetze at uni-potsdam.de]

Dear gmx-users,

Short story: How to freeze the internal coordinates of a molecule?

Long story:
For a new project, we need to evaluate the van-der-Waals interactions  
between two frozen molecules.

For this, we have frozen one of them (a carbonanotube), constrained  
the other molecule's internal coordinates and wanted to apply a simple  
energy minimization to the system. However, there are severe  
difficulties arising, since we cannot make the [ constraints ] part  
working. The molecule always explodes (since we have not defined any  
physical bonds and/or angles, we do not have the necessary parameters  
so far).

What we have are the vdW parameters of the Amber99 force field for our  
molecule, charges are set to zero since we are only interested in the  
vdW interactions.

We tried in two ways:
a) The [ bonds ], [ pairs ], [ angles ] and [ dihedrals ] list of our  
molecule is empty. The [ constraints ] list looks like this:
1 2 2 0.140078
1 3 2 0.241381
and so on. However, our molecule exploded regardless.

b) We defined a list of three bonds per atom, leaving the other lists  
empty. The bond's parameters were set to freeze-distance and 573000  
(some random value). Then, we defined in the .mdp file
constraints=all-bonds
Again, our molecule didn't bother and exploded regardless.

I assume there must be an simple way to constrain a molecule to a  
fixed internal geometry while retaining it's overall translational and  
rotational degrees of freedom. If anybody could help me on that, I  
would be very grateful.

Thank you in advance,
Jan Götze

University of Potsdam
Institute of Chemisty
Theoretical Chemistry Workgroup