[gmx-users] PVT-simulation

[Mark Abraham]

Jiang ran wrote:
> Hello everybody,
> 
> Thank you all for your suggestions ,
> I red the Gromacs software package manual  but  found that PVT was not 
> implemented yet in the Gromacs software  package.  I want to conserve 
> both volume and pressure because my professor says that they are both 
> important for the dynamics of the protein. 

I don't think you can set up a non-trivial ensemble where both pressure
and volume are constant, and even if you can, no force fields have been
parameterized for it because there's no physical system in which it is
true. Consider an ideal gas in such an ensemble. PV = nRT, so with P, V
and n constant, then T is also constant, which means your gas molecules
can only interact in a way such that kinetic energy is constant.

Now for constant n and V you could define negative feedback algorithm
that returned P to a reference pressure, which will force T to vary
inversely. What's the point in doing that?

> So I would like to start 
> implementing it for the weekend.
> I head the idea to introduce langevin dynamics on the  atoms close to 
> the boundary so the magnitude of the random force satisfy the 
> fluctuation dissipation theorem, however the net inwarded direction of 
> the total force on each plane will be solved with a set of langrance 
> multipliers such that the net inwarded force on the plane gives the 
> correct reference pressure  such that  P_ref=-sum(F_plane)/A . In this 
> way both volume as pressure are conserved. In which file of the gromacs 
> software package I can best start implementing it ?
> I started looking in mdrun.c however this didn't make much sence to me.

There's no shortcut possible here, as I said some other time in the last
week. Start with the main function and follow through a typical
calculation - even use a debugger to make life easy on yourself. Learn a
bit about how gromacs works by running a simulation with it first.

Mark