[gmx-users] constant term in dihedral potential function

Mark Abraham Mark.Abraham at anu.edu.au
Fri May 29 08:52:50 CEST 2009

Kukol, Andreas wrote:
> Many thanks for your help.
> As for the 1-4 interactions, I am using the Gromos53a6 forcefield for everything apart from this particular dihedral.
> I did not exclude 1-4 interactions previously with the multiple dihedral terms. Does this mean, if I use Ryckaert-Bellemans now (yielding the same potential function), I don't need to make any changes to 1-4 interactions ?
> Note that the previous method has been validated (with Gromacs 3.3), the problem is that it doesn't work with Gromacs 4 anymore. So I am trying to achieve the same result with Ryckaert-Bellemans instead of multiple periodic dihedrals.

My earlier side point was that you can use multiple periodic dihedrals 
of type 9 if those dihedrals follow each other directly. Such dihedrals 
are effectively of type 1 (see code fragment below).

You still won't get one of multiplicity 0, but such a function doesn't 
influence the dynamics (dV/dr = 0) and would always cancel out of any 
energy comparisons you might do. So I don't think it serves a purpose.

>> I don't believe this works in the way you might expect. Page 109 of the
>> manual contradicts me, but (undocumented) dihedral type 9 was added
>> recently to allow multiple strictly-consecutive dihedrals of type "1" to
>> be accumulated on the same atoms. Anyway, this can readily be checked by
>> comparing 0-step EM runs, before and after commenting out the last of
>> the multiple dihedrals pre-grompp. Also, you can look at a gmxdump of
>> the .tpr.
> This works fine.

I was looking at (4.0.5) src/kernel/toppush.c today for something else, 
and there's a comment in support of my statement above:

	if(ft == 9)
		/* Previously, we have always overwritten parameters if e.g. a torsion
		 with the same atomtypes occurs on multiple lines. However, CHARMM and
		 some other force fields specify multiple dihedrals over some bonds,
		 including cosines with multiplicity 6 and somethimes even higher.
		 Thus, they cannot be represented with Ryckaert-Bellemans terms.
		 To add support for these force fields, Dihedral type 4 is identical to
		 normal proper dihedrals, but repeated entries are allowed.
		bAllowRepeat = TRUE;
		ft = 1;
		bAllowRepeat = FALSE;

The comment should refer to type 9 iso type 4, of course (and again in 
the warning text in push_bondtype() in the same source file). Further, 
push_bondtype() produces a grompp warning about overriding parameters in 
the other cases. So I think Berk is wrong here. Andreas should have a 
look at his grompp warnings, perhaps.


More information about the gromacs.org_gmx-users mailing list