[gmx-users] [pairs] vs [exclusions]

Mark Abraham mark.j.abraham at gmail.com
Wed Jul 13 17:25:11 CEST 2016 

Hi,

Maybe I missed the point earlier, but as e.g. 5.4.4 Exclusions section of
PDF reference manual says:

"Extra exclusions within a molecule can be added manually in a [ exclusions
] section. Each
line should start with one atom index, followed by one or more atom
indices. All *non-bonded*
interactions between the first atom and the other atoms will be excluded."

Non-bonded interactions don't exist until after you've measured the
distance between the particles and found that they're within the cutoff.
Adding an exclusion (howsoever generated) says "look, it doesn't matter,
these particles don't have a non-bonded interaction ever."

The preceding section talks about *intramolecular* [pairs], which are
organized as one of the kinds of bonded interaction. Those just get
computed - whatever [exclusions] says doesn't matter.

If so, then you can show that it works that way by taking a molecule (e.g.
capped alanine) where nrexcl is low enough that the automatic generation of
bond-based exclusions doesn't apply for some pair of atoms, and computing a
potential energy. Add an [exclusions] between two atoms to observe that the
nonbonded energy changes. Then add [pairs] as well on the same atoms, to
show the total energy restored, but now in a different component.

Mark

On Wed, Jul 13, 2016 at 4:04 PM Peter Kroon <p.c.kroon at rug.nl> wrote:

> Hi Xavier (and list),
>
> I think that's a good way of rephrasing my question ;)
> Anyway, I made a mock system with just 2 particles (included below).
> gmx dump'ing the two tprs (with and without exclusions) I can't tell the
> difference, apart from the number of exclusions generated:
>
> >>> diff dump_no_excl dump_with_excl
> 1c1
> < no_excl.tpr:
> ---
> > with_excl.tpr:
> 12c12
> <    ld-seed                        = 4213247321
> ---
> >    ld-seed                        = 2241521802
> 221,223c221,223
> <          nra=2
> <          excls[0][0..0]={0}
> <          excls[1][1..1]={1}
> ---
> >          nra=4
> >          excls[0][0..1]={0, 1}
> >          excls[1][2..3]={0, 1}
>
> I ran this mock-system with a simple energy minimization. Since the
> pairs directive gives both particles an identical charge I expect them
> to drift apart if pairs > exclusions.
> And indeed this is what vmd shows, so it seems that exclusions only work
> on "normal" LJ (and coulomb?) interactions, and not specific pairwise
> interactions.
> I'm hoping someone with intimate gromacs knowledge can still
> confirm/deny this though, I'd rather not dig through the code.
>
> And my mock-topology:
> [ moleculetype ]
> blab 0
>
> [ atoms ]
> 1   Qd  1   VAL A   1   0;
> 2   P5  1   VAL B   2   0;
>
> [pairs]
> 1 2 2 0.1 -0.4 -0.4 0 1.0e-06
>
> #ifdef EXCL
> [exclusions]
> 1 2
> #endif
>
> [ system ]
> blub
>
> [ molecules ]
> blab 1
>
>
> Peter
>
>
> On 13/07/16 14:03, XAvier Periole wrote:
> > Hi Peter,
> >
> > Wouldn’t the exclusion act on the regular LJ and pairs add a different
> LJ(14)?
> >
> > With exclusions, do you have the regular LJ present?
> >
> > XAvier.
> >
> >> On 13 Jul 2016, at 13:17, Peter Kroon <p.c.kroon at rug.nl> wrote:
> >>
> >> Hi Mark (and list),
> >>
> >> thanks for the ideas.
> >> `gmx check -s1 with_excl.tpr -s2 no_excl.tpr` didn't tell me much,
> >> unfortunately:
> >> ...
> >> comparing blocka excls
> >> excls.nra (1944 - 1920)
> >> comparing groups
> >> ...
> >>
> >> I think this means that with_excl.tpr contains 22 more exclusions.
> >>
> >> Running gmx dump on both files shows that both list the relevant `LJC-14
> >> q` entries. However, with_excl.tpr also produces the appropriate
> exclusions.
> >>
> >> So which takes precedence? From these results I'm tempted to think the
> >> exclusions take precedence, otherwise they wouldn't end up in the tpr.
> >> Or does mdrun do something intelligent here?
> >>
> >> Peter
> >>
> >>
> >> On 13/07/16 12:18, Mark Abraham wrote:
> >>> Hi,
> >>>
> >>> I don't know which acts first, but making a few such .tpr files and
> >>> (hopefully) gmx check on them should make clear what happens. Or worst
> case
> >>> gmx dump the .tpr and do a textual diff. You can probably make your
> life
> >>> easier by playing around with atoms with high indices.
> >>>
> >>> Mark
> >>>
> >>> On Wed, Jul 13, 2016 at 10:46 AM Peter Kroon <p.c.kroon at rug.nl> wrote:
> >>>
> >>>> Dear list,
> >>>>
> >>>> I a have a short question on topologies, and the manual is somewhat
> >>>> unclear.
> >>>> If I have a topology which defines both an exclusion between an atom
> >>>> pair and a specific interaction between them using the [pairs]
> directive
> >>>> (function type 2), will they then have a non-bonded interaction, or
> not?
> >>>>
> >>>> Thanks in advance,
> >>>> Peter Kroon
> >>>>
> >>>> PS.
> >>>> short example topology:
> >>>>
> >>>> [ moleculetype ]
> >>>> my_molecule 1
> >>>>
> >>>> [ atoms ]
> >>>> ...
> >>>>
> >>>> [ pairs ]
> >>>> 1   4   2   .....
> >>>>
> >>>> [ exclusions ]
> >>>> 1   4
> >>>>
> >>>> [ bonds ]
> >>>> no bond between 1 and 4
> >>>>
> >>>> ...
> >>>>
> >>>> --
> >>>> Gromacs Users mailing list
> >>>>
> >>>> * Please search the archive at
> >>>> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> >>>> posting!
> >>>>
> >>>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> >>>>
> >>>> * For (un)subscribe requests visit
> >>>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> >>>> send a mail to gmx-users-request at gromacs.org.
> >>
> >> --
> >> Gromacs Users mailing list
> >>
> >> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
> >>
> >> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> >>
> >> * For (un)subscribe requests visit
> >> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.
>
>
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.

More information about the gromacs.org_gmx-users mailing list