[gmx-users] periodic moleules

[Mark Abraham]

Andrei Neamtu wrote:
> Hi,
> 
> I am trying to build a hydrogel polymeric system and I want to make
> use of the option periodic_molecules.
> In this regard I am a little bit confused..... are there any other
> keywords except the option 'peridic molecules' that I have to set to
> define a periodic polymer chain e.g. in the molecular topology file?
> Because in the case  I define a bond between the last atom of the
> chain and the first one, how does the grompp and mdrun know that the
> molecule is periodic and not cyclic?

As you observe, the *topology* is the same in both cases. The coordinate 
file supplied in a well-formed example of each case will differ. The 
.mdp file option merely changes the implementation of PBC during the 
setup and/or simulation. (See manual 7.3.9) As an exercise for the 
reader, presumably you could construct a coordinate file with a cyclic 
version of the same topology, and observe that with and without 
"periodic_molecules" you'd get the same simulation slower and faster, 
respectively.

> I also suppose that the angle and dihedrals that cross the periodic
> border should also be defined.

Yes, in the same manner as normal versions of such interactions.

> Is there a place where I can find how to define periodic molecules in Gromacs?

There are some tidbits here 
http://oldwiki.gromacs.org/index.php/Carbon_Nanotube, some of which 
pre-date the use of "periodic_molecules" I think.

If any of this is bewildering, then doing some tutorials of simpler 
GROMACS simulations might be a good idea, to get an idea of the workflow :-)

Mark