[gmx-users] System size and minimum image convention

Justin Lemkul jalemkul at vt.edu
Fri Feb 22 01:17:36 CET 2013

On 2/21/13 6:30 PM, Elisabeth wrote:
> Thanks Justin. Can you please elaborate on why for a binary
> (solute+solvent) the size should be larger than 2 Rc? I thought minimum
> image convention works for all atoms (solute ans solvent) and this makes
> sure forces are not double-counted. What if the solute is a polymer, I mean
> how can one calculate diameter of a long chain?

One can easily obtain an end-to-end distance in visualization software or with 
g_dist.  Caveat to my previous post: I am assuming that r_c is standard 
nomenclature, i.e. the longest nonbonded cutoff.  If, for instance, your force 
field requires a 1.0 nm cutoff for some term, and your solute is 5.0 nm across, 
it obviously wouldn't make sense that you can define a box that is 2.0 nm 
across.  The solute can potentially rotate in a spherically symmetric way, thus 
requiring that the minimum distance between any atoms in each periodic image is 
larger than r_c, thus for practical purposes, you would set a box with a size of 
5.0 + (2 * 1.0) = 7.0 nm.  Such a box is not strictly necessary, as you could 
theoretically avoid minimum image violations by adding 0.5 * r_c to each 
dimension, but there are potentially solvent layering effects and if using NPT, 
the box size will change and you can thus have sporadic violations.  In 
practice, the risk of junk data outweighs the relatively minor system size increase.



Justin A. Lemkul, Ph.D.
Research Scientist
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080


More information about the gromacs.org_gmx-users mailing list