[gmx-users] Water molecule can not be settled

[Anton Feenstra]

From: Ester Chiessi <ester.chiessi at uniroma2.it>
> 
> But couldn't  this anomalous behavior of a water molecule suggest some heavy mistake
> in the model?
> Which controls may I performe to avoid a similar event in another point of my
> trajectory?

Not necessarily. There is always a finite chance of some molecule (or part,
e.g. -OH, -NH3 group) building up a lot of velocity. This is a physical
phenomenon which, however, is not well described by the MD algorithms. If
I remember explanations by Herman Berendsen correctly, the situation gets
out of hand due to the lack of, or loss of, energy exchange between this
fast moving group and the rest of your system; in other words once it starts
to start, it'll never stop. That is not any longer a physical phenomenon.

Unfortunately, there is no straightforward preventive method for this.
We've thought about imposing a 'speed limit' on atoms, but that would be
very ad-hoc and artificial, and furthermore requires a velocity chech for
every atom at every step; that might be costly. My own working solution,
which is satisfactory and reasonably well justifyable, is the one I
outlined in my previous message.

> I verified that the 'incriminated' water is 4.11 angstrom far from the nearest atom of
> the solute ( a quite hydrophilic polymer ), in the pdb of the last completed step.
> This water is in the proximity of a face of the box.

As David suggested, you might have this water spinning in a 'vacuum pocket'
possibly due to a too low density of your system. Did you check that?
I'm pretty sure all of the PBC bugs we had in the past, have been thorougly
removed. There shouldn't be any new ones either... ;-) You *are* using
a recent version, I assume?

-- 
Groetjes,

Anton
  _____________ _______________________________________________________
|             |                                                       |
|  _   _  ___,| K. Anton Feenstra                                     |
| / \ / \'| | | Dept. of Pharmacochem. - Vrije Universiteit Amsterdam |
|(   |   )| | | De Boelelaan 1083 - 1081 HV Amsterdam - Netherlands   |
| \_/ \_/ | | | Tel: +31 20 44 47608 - Fax: +31 20 44 47610           |
|             | Feenstra at chem.vu.nl - www.chem.vu.nl/~feenstra/       |
|             | "If You See Me Getting High, Knock Me Down"           |
|             | (Red Hot Chili Peppers)                               |
|_____________|_______________________________________________________|