[gmx-users] Oh topology .. go away ...

[Marc Baaden]

Sorry .. that one did not go to the list, so again ...

Thanks Berk,

>>> "Berk Hess" said:
>> There are two ways of doing this:
 >> 1) Add the 1-3 interactions in the [ pairs ] section in the topology.
 >>    You have to make sure that the pair LJ parameters are set
 >>    correctly. You can also add the parameters to the pairs in the
 >>    topology.

But if I do this, will it also create electrostatic interactions ? Or
only LJ (which I would guess).

 >> 2) Use a harmonic potential (bonds type 6) instead of a normal bond
 >>    for the bonds with the iron. For this bond type grompp does not
 >>    generate exclusions. Any 1-4 exclusions that you do want can be
 >>    added in the [ exclusions ] section.

Perfect. That's what I need.


BTW there is something strange about the non-bonded interactions.
Using either of the lines

   FE3   55.84700       3.000       A   0.0          0.0
or
   FE3   55.84700       3.000       A   9.661081e-01 4.287123e-03

in the [ atomtypes ] section, doesn't make a difference (in the
energy) - whereas it should, as c6/c12 are chosen exceptionnally large
in order to make it visible.

BUT, if I include in either [ nonbond_params ] or [ pairtypes ]
a line like eg
      FE3       C  1

Then I get an effect. The weird thing is whether nonbond_params or
pairtypes gives exactly the same energy, and changes are ONLY in
the LJ (SR) energy component. (I guess that is because I chose bond
type 6 and so there are no 1,4-interactions for the Fe .. but I
was surprised that the pairtypes then also switches on the normal
VDW between those atomtypes)

So my interpretation/guess is that LJ interaction between 2 types
are only calculated if the type-pair combination is explicitly
given in either nonbond_params or pairtypes. If the pair combination
is not explicitly mentioned, the c6/c12 values defined by atomtypes
will just not be used at all.

Is that correct ? If so, then maybe it should be clarified in the manual,
because I think it is a bit confusing.

Marc


-- 
 Dr. Marc Baaden - Laboratory of Molecular Biophysics, Oxford University
 mailto:baaden at smplinux.de  - ICQ# 11466242 -  http://www.marc-baaden.de
 FAX/Voice +49 697912 39550  -  Tel: +44 1865 275380  or  +33 609 843217