From: Vincent Kraeutler (vincent_at_kraeutler.net)
Date: Wed Aug 29 2007 - 07:46:48 CDT
this is only partially correct. if you use a nonbonded method like PME,
your molecule is in fact guaranteed to see (the electrostatic aspect of)
the nearest image of itself and in fact all periodic copies.
your box size should therefore aim to satisfy the following properties:
a) as guillaume mentioned, in many (but not all) cases, you want to
avoid LJ-interactions between the molecule and its nearest images. your
minimum box diameter should therefore be larger than the longest
atom-atom distance (at any time during the simulation) in your molecule
plus the cutoff. it's not always easy to predict how your molecule will
behave, so just make a good guess.
b) regarding the electrostatics: the interactions between nearest images
should be small. if you employ criterion a), this is typically the case if:
-- your molecule has no net charge and the dipole moment of your
molecule is not huge.
-- your molecule has a net charge and you have a significant number of
ions, such that the overall charge of your system is zero.
if you want to read up on this, i suggest:
Guillaume LETELLIER wrote:
> hi Philip
> The answer depends on your molecule flexibility and simulation
> You must prevent your moldecule from seeing images of itself generated
> by periodic conditions.
> Therefore, your molecule should never approach the box edge beyond a
> margin of half your cutoff distance.
> Philip Peartree a écrit :
>> I am running a simulation of a protein, and I wish to solvate in TIP3
>> What would be the recommended distance from the protein to the edge
>> of the
>> periodic box that I need to solvate? Is 12 angstroms enough or would you
>> recommend more?
>> Philip Peartree
This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:45:10 CST