[gmx-users] implementing Ewald 3D correction algorithm

Rui Qiao ruiqiao at ews.uiuc.edu
Thu Jan 17 19:56:27 CET 2002

Dear All:
	I posted a question on dealing reduced periodicity for study of
channel flow where electrostatic interaction is important. Thanks for the
replies I've got, I think I do find a suitable algorithm, i.e. Ewald
summatation with correction. (J. of Chem Phys, vol 111, No, 7,
	The alogrithm is very simple:
	a. do Ewald summation, you can use whatever method you want,
e.g. PME, P3M and so on;
	b. Add a correction term to force and potential term. For the
force, it is simply (assume no periodicity in z-direction):
	Fx = Fx + 0 (no correction in x and y direction)
	Fy = Fy + 0
	Fz = Fz - 4*pi*q_i/(slab_volume)*M_z
	wherre M_z is total system dipole moment in z direction. i.e 
M_z = sum(q_i * z_i).

	One can define slab_volume to be a user-defined variable to pass
to Gromacs, and can access q_i(change of atom i) easily, and the only
thing left is calculation of M_z. Because the code is parralleled, I do
not know how to access the position and charge of ALL atoms because each
node probably know nothing about the other nodes.

	So I am wondering if someone has the experience of writing
parallel code can enlight me on how to access such variable. It must be a
simple thing to do, and if one do it in a proper way, 30 minutes is more
than enough.

	All the comments and suggestion are appreciated.



Rui Qiao
Research Assistant
Beckman Institute, UIUC

More information about the gromacs.org_gmx-users mailing list