Re: Interaction energy difference

From: Nicholas J Musolino (musolino_at_mit.edu)
Date: Thu Jul 29 2010 - 10:42:35 CDT

On Thu, 2010-07-29 at 01:08 -0400, Axel Kohlmeyer wrote:
> On Wed, Jul 28, 2010 at 11:42 PM, <m.raviprasad_at_ndsu.edu> wrote:
> > Hi
> >
> > I calculated electrostatic interaction between two group of atoms with and
> > without periodic boundary condition. But the values are different.
> >
> > Ex: Elec. without PBC = 155 kcal/mol
> > Elec. with PBC = 300 kcal/mol
> >
> >
> > According to the unit of the interaction energy, Interaction energies are
> > calculated for mole. so Why this difference?
>
> short of knowing any other details, would say the difference is
> because of the periodic boundaries.
>
> a.
>
> > Does any one have an answer.

Can I suggest a question related to your e-mail, Mr. Raviprasad? What
is the "mol" in the kcal/mol counting? Is the interaction energy:
A. kcal/(mol atoms in system)
B. kcal/(mol molecules in system)
C. kcal/(mol of systems)
D. kcal/(mol of systems) = 0.166E-23 kcal for a single system

To avoid any confusion, I would point out that choices A and B are NOT
what NAMD prints. Choice "C" might be a good way to think of the units
of energy in the PBC case, while choice "D" might be a good way to think
of the units of energy for the isolated case.

Of course, I'm assuming that your confusion is with the units. If your
confusion is with the numbers (155 versus 300 kcal/mol), try to
calculate (by your own program if you want) the energy of a single
system of point charges.

Then you can calculate the energy of that same system, repeated in a
periodic fashion, by using the original system image plus the point
charges in its nearest neighbor images. Then all those charges plus the
second-nearest neighbors. The energies will be different, no?

This difference is certainly a physical one: NAMD is simulating two
different things in your two different calculation. The first is a
nanodroplet (or nanocrystal) in vacuum, while the other is a liquid (or
infinite-extent crystal). Sorry to be a little vague here, but we don't
really know what your system comprised.

Of course, depending on what your question really is, you may be
interested in reading about Ewald summation (since electrostatic
potential decays as 1/r^2, it's non-trivial to add up all pairwise
interactions in an infinite system).

And if you want to get very detail-focused, you can venture into tinfoil
boundary conditions, etc., but that is going far afield.

As Alex implied, without more information, it's hard to know exactly
what you're asking.

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:54:21 CST