[gmx-users] NVT + frozen groups

[Erik Lindahl]

Valentin Gogonea wrote:

> Thanks Erik for you quick answer.
>
> Does it make sense to make NPT simulations with frozen groups? I mean 
> if the coordinates of the frozen atoms are not scaled what happen with 
> the bonds between atoms in the moving part and those in the frozen part?
>
> In a hypothetical situation that the moving region is totally buried 
> in the frozen one, how does the pressure coupling works? It seems to 
> me that the box size cannot be adjusted in this case because the 
> moving part is wrapped in a hard wall. Am I wrong?

No, it could probably lead to quite strange results, especially if you 
freeze a large part of the system, so in general it's probably easier to 
freeze in NVT.

Are you sure you need to freeze things completely? In most cases it's 
gentler on the dynamics to just apply position restraints (but even then 
the coordinates will be constrained to the starting positions, while 
everything else scales if you use pressure coupling).

There's no such thing as a free lunch :-)

Cheers,

Erik