[gmx-users] Shift and Switch are obsolete!?

[K.A. Feenstra]

Alan Wilter Sousa da Silva wrote:
> 
> Hi, pardon if I go straightforward to the question.
>         I'm trying to set coulomtype to Shift (or Switch).  This is my
> file.mdp:
> constraints              = h-bonds
> constraint-algorithm     = shake
> nstenergy                = 0
> ns_type                  = simple
> coulombtype              = Switch
> rlist                    = 1.3
> rcoulomb                 = 1.2
> rcoulomb_switch          = 1.1
> vdwtype                  = Switch
> rvdw                     = 1.2
> rvdw_switch              = 1.1
> 
> GMX manual recommends rlist to be 0.1 to 0.3 nm larger than rcoulomb.  But
> I got it from grompp:
> 
> ERROR: rcoulomb must be >= rlist
> ERROR: rvdw must be >= rlist
> 

Yes, the manual says:

Shift
	The Coulomb potential is decreased over the whole range and 
	the forces decay smoothly to zero between rcoulomb switch and 
	rcoulomb. The neighbor search cut-off rlist should be 0.1 to 
	0.3 nm larger than rcoulomb to accommodate for the size of charge
	groups and diffusion between neighbor list updates.

*But* that is *not* correct. As grompp says, you must have rcoulomb>=rlist.
I think this issue of charge group size and diffusion is now already taken
care of by the neighbor searching algorithm directly (see manual p76).

[ David and/or Erik, please correct me if I am wrong... 
  And, someone should fix this in the manual, too... ]

> Please, one question more.  If I use pure cut-off,  I still do not follow
> why rlist must be larger or equal to rcoulomb.  How does GMX couple with
> an atom at the exactly distance of cut-off, and atoms beyond?  One cannot
> have discontinuity!  So, is a shift function implicit with cut-off?

Well, actually it should be the other way around (see above), i.e.
rcoulomb >= rlist. The reasons are explained on page 76 of the manual.
The important points are, that neighbor searching is done per charge
group, i.e. a charge group as a whole is placed in the neighbor list
or not. That means that some of the actual atoms could be inside the
cut-off radius but not included in the neighborlist because their
chargegroup is not. Also, neighborlists are updated only every nstlist
steps (typically every 10-20 fs), which means that in the meantime a
charge group can have moved into the cut-off radius. Although, of
course, in 20 fs one cannot move a (group of) atom(s) very far.

With a cut-off you always have a discontinuity. This *does* introduce
errors (see numerous discussions on cut-offs etc. in this list), but
these are usually not critical. So, no, shift is not implicit in cut-off.


-- 
Groetjes,

Anton
 ________ ___________________________________________________________
|        | Anton Feenstra                                            |
| .      | Dept. of Pharmacochemistry - Vrije Universiteit Amsterdam |
| |----  | De Boelelaan 1083 - 1081 HV Amsterdam - The Netherlands   |
| |----  | Tel: +31 20 44 47608 - Fax: +31 20 44 47610               |
| ' __   | Feenstra at chem.vu.nl - http://www.chem.vu.nl/afdelingen/FAR|
|  /  \  |-----------------------------------------------------------|
| (    ) | Dept. of Biophysical Chemistry - University of Groningen  |
|  \__/  | Nijenborgh 4 - 9747 AG Groningen - The Netherlands        |
|   __   | Tel +31 50 363 4327 - Fax +31 50 363 4800                 |
|  /  \  | K.A.Feenstra at chem.rug.nl - http://md.chem.rug.nl/~anton   |
| (    ) |-----------------------------------------------------------|
|  \__/  | "If You See Me Getting High, Knock Me Down"               |
|        | (Red Hot Chili Peppers)                                   |
|________|___________________________________________________________|