molecular dipoles of charged molecules

[Axel]

On Oct 19, 11:26 am, Laino Teodoro <teodor... at gmail.com> wrote:
> Ciao Axel,

hi teo,

there is/was only one entry for "molecular dipoles".

> I checked carefully the part of the code connected with the
> calculation of the dipoles (both total and for each molecule)
> when using the wannier centers, i.e.:
>
> DFT%PRINT%LOCALIZATION%MOLECULAR_DIPOLES
> DFT%PRINT%LOCALIZATION%TOTAL_DIPOLE


hmmmm.... the organization of this part of the input is a bit
confusing.

i would have either expected that you move them to the
CP2K_INPUT / FORCE_EVAL / DFT / PRINT / LOCALIZATION / LOCALIZE /
PRINT
block where the other wannier center related output is found or
- which makes even more sense to me - move those subsections
directly into LOCALIZATION and merge LOCALIZATION%LOCALIZE
into LOCALIZATION.

> While for MOLECULAR_DIPOLES they are always computed w.r.t the
> position of the first atom in the molecule,
> the TOTAL_DIPOLE was (as you said) totally position dependent.

> Moreover, as I mentioned, it was absolutely not
> related to the other possible way of computing dipoles: DFT%PRINT%
> MOMENTS

the documentation there is very terse. i understand this section
as computing properties of the "total" system, right?

> Therefore:
> -) MOLECULAR_DIPOLES and TOTAL_DIPOLE are now print_key sections
> inside localization.
> -) both sections have the same possible parameters as MOMENTS, i.e.:
> PERIODIC (if the dipole should be
>     computed with the berry phase or not), REFERENCE and
> REFERENCE_POINT.
> -) While these keywords are fully implemented for the TOTAL_DIPOLE
> section, they are not for MOLECULAR_DIPOLES
>     since I think the convention used (i.e. to compute the dipole
> w.r.t. the first atom in the molecule) is already quite good.

well, that makes the resulting dipole depend on the shape of the
molecule.
it is not a big deal for most molecules, but it is for the kind of
molecule
that i am currently looking into, which is a small organic cation that
is
mostly flat and picks up a dipole moment from bending and being
polarized
from solvate molecules.

>     In case somebody will really need it, the implementation is quite
> straightforward..

i'll look into it. i don't really need that kind of output, since i
have
an equivalent implementation with sufficient flexibility already
integrated
into VMD (Extensions->Visualization->Dipole Moment Watcher), but i was
a
bit surprised by the discrepancies and wanted to check with the folks
here
whether this was intentional or accidental.


cheers,
     axel.
> Hope this solves your issue.
> Teo
>
> On 19 Oct 2008, at 03:32, Axel wrote:
>
>
>
> > hi,
>
> > this is a bit of a badly conditioned problem, but i noticed
> > that the molecular dipole value computed for charged molecules
> > does not take out any net charge and thus makes the resulting
> > dipole strongly position dependent, and thus practically useless.
>
> > i've met the same problem while writing a visualization tool for
> > VMD and ultimately decided to remove the total charge at the
> > center of the molecule to retain position independent results
> > (and visualizations).
>
> > are there any opinions on this matter?
> > or even examples for where the current
> > implementation is useful.
>
> > thanks,
> >     axel.