[gmx-users] Theoretical questions about cut-offs and neighbor searching

Justin Lemkul jalemkul at vt.edu
Wed Sep 19 20:09:18 CEST 2012

On 9/19/12 12:54 PM, Lara Bunte wrote:
> Hello
> In my practice calculation I made wrong cut-offs in an energy minimization. After the help of Justin and Peter (thanks a lot) it works and now I have some theoretical question about this. I read about cut-offs in the manual but I don't understand it really.
> Could you please explain me in a physical meaning what cut-offs are and could you please explain what wrong cut-offs mean for the system? What is the correspondece to neighbor searching, because I always read about neighbor search and don't really understand what is this about.
> I would be thankful if you use not the same words as in the manual, because I sadly don't understood that :-( (but I will keep trying to understand and read the manual)

The specifics differ depending on the algorithms used, but in general, the 
cutoff is the distance beyond which atoms do not contribute to short-range 
interactions and participate in neighbor lists that are updated every step. 
Whether or not they contribute to longer-range interactions depends on the 
algorithms being used (switch and shift for van der Waals, switch, shift, PME, 
etc for electrostatics).  Several molecular simulation textbooks discuss these 
topics in depth; the manual will not explicitly provide all the necessary 
information, nor should it.  Everything that one needs for better comprehension 
is cited therein.

The main thing for practical purposes is that each force field is validated 
using a specific set of cutoffs and nonbonded calculation algorithms from which 
you should not deviate unless you (or others) have proven those settings to be 
superior to the original derivation.  Otherwise, you're using a hacked force 
field that has no guarantee of producing reliable results.



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