[gmx-users] Help about 2 D periodic boundary condition

[David L. Bostick]

Hi,

I have found that for systems with a large asymmetry in one dimension like
a macro-dipole along the bilayer normal in membrane-channel/bilayer systems,
periodic images in the asymmetric direction can affect the polarization of
solvent (that is, if you have a polar solvent).  The implementation of this
sort of geometry makes a huge difference if one wishes to calculate the
dielectric constant of bulk water (see JCP 110(16):7935-7942) or observe
how water behaves in confined spaces (see JCP 117(17):8050-8062).

This is an old issue in simulations of liquid/liquid interfaces, but has
not carried over to biological simulations much.  One cannot just include
vacuum in the asymmetric dimension or add extra water.  A 2D Ewald sum must
be used (which is expensive) or one can opt to simulate with a 3D sum and
add a correction term for slab geometry plus the addition of vacuum (making
the simulation box around 3x larger in the asymmetric direction than in
the other two).

All in all however... it depends on what you want to look at..  If you are
interested in things that are rather relatively insensitive to the large
scale polarization of solvent (i.e. protein shape etc.) then it may be
perfectly fine to use tinfoil boundary conditions.

Yes ... it only affects the surface term, but the essence of the
conditional convergence of the Ewald sum is encapsulated in this term.  If
you take a spherical sum you get one answer.  If you take a slab-wise sum
you get another.  We are dealing with the behavior of the system at
infinity, then again, with periodic boundaries, we are simulating an
infinite system.  .. Seems philosophical, but it can make a difference.

David

-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
David Bostick					Office: 262 Venable Hall
Dept. of Physics and Astronomy			Phone:  (919)962-0165
Program in Molecular and Cellular Biophysics
UNC-Chapel Hill
CB #3255 Phillips Hall				dbostick at physics.unc.edu
Chapel Hill, NC 27599	           		http://www.unc.edu/~dbostick
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

On 13 Jan 2003, David wrote:

> On Mon, 2003-01-13 at 17:35, David L. Bostick wrote:
> >
> > Hello,
> >
> > For treatment of coulomb interaction in "slab geometry" i.e. the treatment
> > of 3D systems with 2D periodicity there is a variable that can be set in
> > the mdp file..
> >
> > ewald_geometry
> >
> > you can set ewald_geometry = 3d
> >
> > for a normal spherical ewald sum and
> >
> > ewald_geometry = 3dc
> >
> > for the addition of a correction term for slab geometry.
> I recently looked into the code for this option. It only affects the
> surface term, so if you set epsilon_surface = infinity there is no
> difference. I assume that is the correct implementation (but haven't
> read the paper), but I still have my doubts, since the PBC is computed
> in three D just as usual. So I still would like to ask whether people
> have tried this?  Is this really suitable for studying e.g. a vacuum
> interface?
>
>
> > > As for 2D PBC, it depends on what you want in the third direction.
> > > You can for example make a two 'walls' of frozen atoms above and
> > > below your system, or perhaps simply a vacuum boundary. The system
> > > will still be treated in 3D PBC, but the third dimension will not
> > > be relevant.
> So what does it mean: not relevant?
>
> --
> Groeten, David.
> ________________________________________________________________________
> Dr. David van der Spoel, 	Biomedical center, Dept. of Biochemistry
> Husargatan 3, Box 576,  	75123 Uppsala, Sweden
> phone:	46 18 471 4205		fax: 46 18 511 755
> spoel at xray.bmc.uu.se	spoel at gromacs.org   http://xray.bmc.uu.se/~spoel
> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> _______________________________________________
> gmx-users mailing list
> gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
>