[gmx-developers] Re: Editing the GROMACS energy function

[Szilárd Páll]

On Fri, Aug 31, 2012 at 5:24 PM, alex.bjorling <alex.bjorling at gmail.com>wrote:

> You certainly have, thanks.
>
> The function can indeed be written as an independent module, it just needs
> the coordinates. So I guess that means that if I need to operate on all
> coordinates I will have to run without domain decomposition?
>

Or parallelize your algorithm straight away so that it work with DD
straight away ;)

Otherwise, 4.6 will support OpenMP, so if you use that code-base at least
you won't have to run single-threaded.

--
Szilárd


>
> Alex
>
>
> Francesco Oteri wrote
> >
> > You can do simply writing your own function and executing it after (or
> > before, it depends by the details of your modification)
> > do_force (in md.c) simply modifying the f vector and printing the output
> > after print_ebin (the function that actually prints results).
> > Of course this approach is right just if your modification can be
> > developed
> > as an independent module.
> >
> > Regarding coordinates, because of domain decomposition, each node can
> > access only to local atom data.
> >
> > ...
> >
> > I hope I've helped you :)
> >
> > Francesco
> >
> >
> > 2012/8/31 alex.bjorling &lt;alex.bjorling@&gt;
> >
> >> I'm now in the exact same position as Radhakrishna describes here. After
> >> posting the following, I did some more research and now have ways to get
> >> both energy and force from a complicated operation over all coordinates.
> >>
> >>
> >>
> http://gromacs.5086.n6.nabble.com/Adding-an-artificial-energy-term-for-structure-refinement-tp5000240.html
> >>
> >> Has anyone been successful in making these modifications?
> >>
> >> Best,
> >> Alex
> >
>
>
>
>
> --
> View this message in context:
> http://gromacs.5086.n6.nabble.com/Editing-the-GROMACS-energy-function-tp4669473p5000724.html
> Sent from the GROMACS Developers Forum mailing list archive at Nabble.com.
> --
> gmx-developers mailing list
> gmx-developers at gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-developers
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-developers-request at gromacs.org.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-developers/attachments/20120831/0db367e3/attachment.html>