[gmx-users] using two different combination rules for the LJ potential

Mark Abraham Mark.Abraham at anu.edu.au
Thu Jul 23 04:12:46 CEST 2009

Ana West wrote:
> Hello,
> I am trying to simulate a system composed of two different molecules.  I 
> made the topology files for each of the two molecules and I did some 
> tests to validate the force-field parameters for my system. I do have 
> all the correct parameters for my system but it turns out that for the 
> two types of molecules I must use two different combination rules of the 
> LJ potential (the combination rule 1 and the combination rule 3).

That suggests you're mixing force fields, which is a bad idea. In any 
case, you can convert between these parameter forms easily (see manual 
section 5.3.3). Obviously, compare outputs between the old combination 
rule parameters and the converted ones to check your conversion.

> I did some tests using Gromacs and it seems that I can specify the 
> directive
>  [ defaults ] only one time. The manual seems to indicate that I might 
> be able to do something about the combination rules by including a [ 
> nonbond_params ] section in an *.itp file. However, I don;t seem to 
> understand the effect of the  [ nonbond_params ] directive onto the [ 
> defaults ] directive normally found in the *.itp file. It seems to me 
> that there is no combination rule specified in the under the [ 
> nonbond_params ] section. Could someone please explain to me how this 
> works?

There is one combination rule available, specified in [ defaults ] and 
used everywhere required. This is sound methodology.

> In addition/alternatively, does anyone have any suggestions on how I 
> could go about creating a successful set of topology files using two 
> different combination rules for the LJ potential?

Depending on the underlying reason, probably don't do it. For a start, 
when an atom whose parameters are of combination type 1 interacts with 
an atom whose parameters are of combination type 3, you would need to 
pre-compute their interaction and describe that in [ nonbond_params ]. 
That's rapidly tedious as the number of atoms grows.


More information about the gromacs.org_gmx-users mailing list