[gmx-users] Rigid structure and flexible structure present in sing system

[Ignacio Fernández Galván]

<chris.neale at utoronto.ca> wrote:

> If you simply want to know how to freeze one structure, then use freeze groups
> and energy exclusions without pressure coupling, or use position restraints 
and
> refcoordscaling=com with position restraints, or create an elastic network of
> restraints.

Is it really not possible (or conducting to wrong results) to do NPT simulations 
with freeze groups?

I did some tests with just a freeze group (a single molecule with no bonds with 
the rest of the system), and it caused a blow-up, indeed. But excluding all 
intramolecular interactions, for example using the "5" bond type and a large 
enough "nrexcl" value, or explicitly listing all exclusions in the topology, 
seemed to fix it. The coordinates of the frozen molecule do not change when the 
box grows or sinks...


So, my conclusion was that NPT with a frozen molecule is possible, as long as no 
intramolecular terms are included. Am I wrong?

Ignacio