[gmx-developers] Small hack regarding dihedrals

[MURAT OZTURK]

Thanks for all the replies.

pdihs() appears to be the main functon that also calls dopdihs().

Is it OK to intorduce the conditional into pdihs(), and bypass even calling
dopdihs() if r_kj  ( as reported by calling dih_angle() ) is less then
threshold?

I will go for this as it makes sense to me, if it is not directly colliding
with something that I don't know about.

I would truly appreciate if someone with intimate knowledge of the code
comment on the acceptability of this approach. I am not entirely
comfortable acting on such limited knowledge.

I am already familiar with the -noddcheck option, thanks to your mentioning
it previously, Dr. Hess, but thank you very much.

Regards

Murat


On Wed, Sep 25, 2013 at 3:53 PM, Berk Hess <hess at kth.se> wrote:

> On 09/25/2013 02:31 PM, Erik Lindahl wrote:
>
>> Type 1 is a "proper dihedral", which means "do_pdihs()" ;-)
>>
>> However, beware if you are running in parallel with domain decomposition
>> - even if _you_ won't evaluate the interaction when it is beyond the
>> cutoff, Gromacs will complain if it cannot communicate the coordinates for
>> you.
>>
> You can turn off that check by running mdrun -noddcheck
>
> Cheers,
>
> Berk
>
>  Cheers,
>>
>> Erik
>>
>> On Sep 25, 2013, at 2:25 PM, Mark Abraham <mark.j.abraham at gmail.com>
>> wrote:
>>
>>  I forget which, but you will anyway need a test case that shows you
>>> that your modification is working correctly. Make a small but relevant
>>> system and use it with mdrun -rerun to compute single-point energies
>>> and forces. Then hack functions one by one by forcing them to output
>>> zero until you see an effect with mdrun -rerun. Now you know what
>>> function to work on, and have your test machinery ready to use!
>>>
>>> Mark
>>>
>>> On Wed, Sep 25, 2013 at 2:15 PM, MURAT OZTURK <murozturk at ku.edu.tr>
>>> wrote:
>>>
>>>> Hey guys,
>>>> I am trying to introduce a small hack that will  turn off dihedrals if
>>>> the
>>>> middle bond is longer than a threshold.
>>>>
>>>> It should be as simple as
>>>>
>>>> midDist= iprod(r_jk,r_jk)
>>>> midDist *= gmx_insqrt(midDist)
>>>>
>>>> if ( midDist > threshold)  {don't even bother calculating anything for
>>>> this
>>>> dihedral}
>>>> else {continue as usual}
>>>>
>>>> But there are so many functions in bondfree.c. I need to locate that is
>>>> responsible for regular, periodical dihedrals (type 1). Which one is
>>>> this?
>>>> My suspects are..
>>>>
>>>> dopdihs, dopdihs_XXXX (noener etc.) , pdihs, pdihs_XXXX (noener etc.)
>>>>
>>>> Is there any documentation I can look at to figure our what these do?
>>>> There
>>>> are not many comments within the code.
>>>>
>>>> I appreciate the input. This is all for a fancy coarse grained system.
>>>>
>>>> Murat
>>>>
>>>> --
>>>> gmx-developers mailing list
>>>> gmx-developers at gromacs.org
>>>> http://lists.gromacs.org/**mailman/listinfo/gmx-**developers<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@**gromacs.org<gmx-developers-request at gromacs.org>
>>>> .
>>>>
>>> --
>>> gmx-developers mailing list
>>> gmx-developers at gromacs.org
>>> http://lists.gromacs.org/**mailman/listinfo/gmx-**developers<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@**gromacs.org<gmx-developers-request at gromacs.org>
>>> .
>>>
>>
> --
> gmx-developers mailing list
> gmx-developers at gromacs.org
> http://lists.gromacs.org/**mailman/listinfo/gmx-**developers<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@**gromacs.org<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/20130925/c58fea40/attachment.html>