[gmx-developers] more on electrostatic decoupling

[Erik Lindahl]

Hi,

First, it's certainly possible to do accurate free energy  
calculations in the current version - the decoupling is a neat  
optimization, but not crucially necessary.

> ... not really, if the Ewald sum (or PME) is implemented
> correctly - for a good overview of ALL terms see e.g.:
> http://www.earth.ox.ac.uk/~keithr/moldy-manual/node11.html
> There are two terms called "Uniform Sheet Correction"
> and "Surface Dipole Term" - I am not sure if the former is
> in gromacs, however both terms need to be implemented
> and understood (I am not sure if I do) in order to deal
> with systems with a permanent net-charge and/or dipole moment.
> ... you also might want to read:
> S. Boresch and O. Steinhauser
> Presumed versus real artifacts of the Ewald summation
> technique: The importance of dielectric boundary conditions.
> Ber. Buns. Ges. 101:1019, 1997.

All the appropriate correction terms are included in the Gromacs PME  
version. However, the problem with the ligand seeing periodic copies  
other isn't any artifact - it SHOULD see the periodic copies since we  
applied PBC+PME, and e.g. two parallel dipoles in the z-direction  
will feel a slight repulsion in the x-y plane.

The dipole correction term only alters the force in the same  
dimension as the net dipole, i.e. z in this case, so there is still a  
net interaction.

The problem in the decoupled state is that there is nothing screening  
this interaction, and it is thus likely to be overestimated (since  
the periodicity is an approximation to start with). We have no idea  
how severe it is, though.


Cheers,

Erik

-----------------------------------------------------------
Erik Lindahl  <lindahl at sbc.su.se>     Backup address:  
<erik.lindahl at gmail.com>
Assistant Professor, Computational Structural Biology
Stockholm Bioinformatics Center, Stockholm University, SE 106 91  
Stockholm
Phone: +46 8 5537 8564     Fax: +46 8 5537 8214