[gmx-users] one more pme

David van der Spoel spoel at xray.bmc.uu.se
Sat Mar 9 14:17:51 CET 2002

On Fri, 8 Mar 2002, Peter Tieleman wrote:

>Hunenberger's papers aren't the first to show lattice sum artifacts. I 
>think PME is the best algorithm in Gromacs but there definitely are 
>problems with it. Warshel has some interesting papers, some very old 
>already, about hydration free energies that become strongly size 
>dependent with lattice sums. References are in a review:

What everybody knows (or should know) is that you have to verify that the
approximations you make in a simulation do not influence the results. In
special cases you can get away with compensation of errors (e.g. for FEP
calculation mutating one neutral molecule to another) but compensation of
PBC effects by cut-offs sounds like a *bad idea*.

The obvious thing to do is use several box sizes and see whether your
results converge to a meaningful number.

Some hints:
- water-water interactions have a range of roughly 1.5 nm, boxes smaller
  than 3.0 nm will have artefacts.
- ion - ion or ion-water interactions have much longer range without PME
  you're simply introducing lots of errors.

Even though PME may enhace PBC artefacts, cut-offs are not a solution.

Groeten, David.
Dr. David van der Spoel, 	Biomedical center, Dept. of Biochemistry
Husargatan 3, Box 576,  	75123 Uppsala, Sweden
phone:	46 18 471 4205		fax: 46 18 511 755
spoel at xray.bmc.uu.se	spoel at gromacs.org   http://zorn.bmc.uu.se/~spoel

More information about the gromacs.org_gmx-users mailing list