[gmx-users] Treating nonbondend interactions in GROMACS

Justin A. Lemkul jalemkul at vt.edu
Sun Jul 11 04:05:12 CEST 2010

jojo J wrote:
> Hello Justin,
> Thanks for your comments and also I really liked your reasoning for 
> getting no answer from people that are not awake :)))) You are right ;)
> Actually, I have no charged particles in my system. All I have is carbon 
> and hydrogen. I read through the manual meticulously about nonbonded 

I have yet to see a force field that treats even hydrocarbon groups (like amino 
acid side chains) as completely uncharged.  There is a slight electronegativity 
imbalance between C and H, so the bonds are slightly polar, even if the 
molecule, overall, is nonpolar.  It is up to you to justify your model.

Depending on how you treat your system may make the following points entirely 
irrelevant.  If you indeed have 0 charge on all the atoms in your system, how 
you treat electrostatics is unimportant - there are no electrostatic 
interactions in your system at all.

> interaction terms. I would like you to correct me if I am wrong. See the 
> summary below (my understanding) please:
> Treating nonbonded interactions for a hydrocarbon system:
> 1-Short range:
> vdw: shift
> electrostatics: shift
> so the settings would be :
> rvdw=1
> rcoulomb=1
> rlist=1.1
> rcoulomb-switch=0
> (when this is zero means shift function is added to actual function for 
> all r?)

Yes, you're shifting over the entire range of 0-1.1 nm.

> I have a question here: in manual I read setting this last parameter 
> nonzero means it acts as a switch function. I thought if it is nonzero 
> shift acts just at a nonzero-r between zero and r_c (rcoulomb or rvdw)

No idea, sorry.

> 2-Long range
> vdw: Dispcorr:Enerpres
> electrostatics: PME
> There are 2 issues from last post (quotes) I am unsure about.
> "in md.mdp file I have only coulombtype=PME"
> "This contradicts your statement above.  You said you had only uncharged 
> particles, but clearly this is not the case."
> What you mean is Basically that I dont need to consider PME for 
> hydrocarbon system?

Or any method for electrostatics at all.

> and if you mean I am using wrong setting:
> I do not know how else I can treat LR electrostatics in my system. I 
> read in the manual that PME almost for all cases performs better than 
> Ewald and also PPPM cant do P coupling so I picked PME :)
> Also you said:
> "A method like PME is significantly more accurate, but you can also 
> achieve reasonable results using switch, shift, etc.  Read about the 
> relevant mechanics of the force field you're using for what might be 
> appropriate."
>  From what your statement above I realize that PME and shift or switch 
> func's are different alternatives. To be honest I am a little confused 
> about the methods for treating nonbonded interactions. Thats why I wrote 
> that little summary above so that you can correct me. This could be 
> useful for other beginners as well. My understanding from reading the 
> manual is that PME is for LR and switch and shift functions for SR. How 
> could they be alternatives?

You don't pick a separate short-range and long-range method for either vdW or 
Coulombic interactions.  Dispersion correction is applied for systems using 
plain vdW cutoffs to account for truncation artifacts.  There are several basic 
textbooks that discuss this.

For PME, the short-range interactions are treated by a modified switch function. 
  See the manual, section 4.1.6 for the description.  The remaining 
electrostatics (beyond rcoulomb) are calculated in reciprocal space.  So by 
setting "coulombtype = PME" you treat short-range and long-range effects 



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


More information about the gromacs.org_gmx-users mailing list