[gmx-users] Problem with coulombtype=user and 1,4 interactions

[Christian Mücksch]

Dear Gromacs experts,

I've been trying to use a tabulated potential for the 
coulombic-interactions with all Gromacs versions from 4.0.7 to 4.5.3.

In combination with OPLS/AA force field and SPC/E water I used a 
potential by Gezelter for a simple system consisting of a protein in 
Water+Ions:
Fennell, Christopher J. ; Gezelter, J. D.: Is the Ewald summation still 
necessary? Pairwise alternatives to the accepted standard for long–range 
elec- trostatics. In: Journal of Chemical Physics 124 (2006)

When I run mdrun I specify sth. like this:
mdrun_mpi_s -table table_gezelter_s.xvg -tablep table_gezelter_s.xvg -v 
-deffnm file.tpr

My problem is that the 14 Coulomb-energy is approximately half the value 
of that compared to PME-switch.
The Coulomb-SR energy is approximately at the same value as PME and 
everything else works alright.

So there's sth. wrong with the 1,4 energies. Since OPLS/AA uses a 
scaling factor of 1/2 for the 1,4-interactions changing it will mess up 
the whole simulation since the 1,4 LJ energies will get wrong. If I use 
another table for the 1,4 interactions multiplied by a factor of 2 
everything gets worse.

As a consequence this potential fails to reproduce densities etc.

So what am I missing? Is it the potential that's wrong?

Thanks alot for any ideas!
Christian

-- 
Christian Mücksch
Department of Physics
TU Kaiserslautern
Erwin Schrödinger Straße
67663 Kaiserslautern
Germany

Phone: +49 (0)631 205 4287
Fax:   +49 (0)631 205 4965
Email: muecksch at rhrk.uni-kl.de