[gmx-users] pme/cutoff in membrane proteins

[Erik Lindahl]

On Fri, 2002-03-08 at 12:50, Periole Xavier wrote:
> 
> Hi all,
> 
> interesting works have been done in McCammon lab. about
> problems using PME for globular proteins ...
> 
> P.H. Hunenberger et al., Biophysical Chemistry 78 (1999) 69-88
> W Weber et al., J. Phys. Chem. 104 (2000) 3668-3675
> Xavier
> 

Just to clarify things: It's not really correct to speak of PME or
lattice sum artifacts as Phil Hunenberger (and other people) usually do.

The artifacts are introduced by the periodic boundary conditions that we
introduce to compensate for finite size of the system. Once you have
decided that this is the geometry (and system size) you want to
simulate, lattice sums will give you the most accurate forces and
potentials (given the geometry and periodicity) - period. 

What Phil and others have done is to show that the *additional*
artifacts introduced by cutoffs might compensate for some of the
artifacts caused by PBC in small systems (while other properties could
be worse off). This might be a good compromise in some cases, but it is
definitely not theoretically more correct since you are making two
approximations (PBC + cutoff) instead of one (PBC).

Cheers,

Erik