From: Giacomo Fiorin (giacomo.fiorin_at_gmail.com)
Date: Fri Nov 24 2017 - 12:31:37 CST
Do the LP sites have zero mass? That could be the reason of the division
by zero -> nan.
As you mentioned, you can use the cartesian coordinates directly and deal
with this special case the way intend it.
Giacomo
On Fri, Nov 24, 2017 at 11:44 AM, Bogdan MAREKHA <marekha_at_mpip-mainz.mpg.de>
wrote:
> Dear NAMD community,
>
> I am running some *a posteriory* analysis of my MD trajectories generated
> with the latest CHARMM Drude polarizable force-field using the capabilities
> of the collective variable module.
>
> The problem I am facing is that whenever I call distance, distanceZ or
> distanceVec colvar component involving at least one LP-type site I get a
> *nan* as an output. It also happens with other geometrical colvars like
> angle or dihedral. Everything works fine for other types of Drude
> interaction sites like D***. This has been tested with the NAMD 2.12
> release version and with the latest night build (24.11.2017).
>
>
> Does anybody have an idea of possible reasons for that?
>
>
> Another interesting observation is that I do have access to the
> coordinates of the LP-sites via cartesian colvar component. This means
> that I can overcome those *nan*'s by constructing the corresponding
> colvars using linear combinations, custom functions or scripted
> variables,e.g.
>
> distanceVec {
>
> group1 { atomNumbers 1 }
>
> group2 { atomNumbers 2 }
>
> }
>
> could be equivalently computed as
>
> cartesian {
>
> componentCoeff -1.0
>
> atoms { atomNumbers 1}
>
> }
>
> cartesian {
>
> componentCoeff 1.0
>
> atoms { atomNumbers 2}
>
> }
> Cheers,
> Bogdan
>
-- Giacomo Fiorin Associate Professor of Research, Temple University, Philadelphia, PA Contractor, National Institutes of Health, Bethesda, MD http://goo.gl/Q3TBQU https://github.com/giacomofiorin
This archive was generated by hypermail 2.1.6 : Mon Dec 31 2018 - 23:20:43 CST