[CP2K:2109] Re: Inconsistency in molecular dipole calculation

Fawzi Mohamed fa... at gmx.ch
Fri Jun 5 13:30:41 UTC 2009

I had tested the total dipole, but mostly with orthorombic cells.

I never found something wrong with it.
I did not test the molecular dipoles, but as Teo said the basic backend is the same.

One thing that the periodic dipole has is that it is defined only up to a jump of an electron (or two) by a cell length.
The total dipole should always choose the things so that the dipole is as close to 0 as possible.

The molecular dipole on the other hand tries to calculate the dipole wrt. to the center of the molecule. Maybe in the definition of that it can jump (if the cell is smaller than 2 timese the molecule it can guess the "wrong" center.
If you express the dipole in au. (or reduced unit) then the jump should be readily corrected.
Better ways to define the reference point for the molecule are obviously welcome, or maybe you really found a bug, can you explain a little more?

One thing that I saw for ionic thing the direct method (which in general is much faster) needs to be converged to *very* high accuracy to avoid jumps in the localization during an md.

Teodoro Laino wrote:
> The way it is computed is similar to TOTAL_DIPOLE. This means that you 
> should get big problems as well
> with the TOTAL_DIPOLE..
> There could be a bug somewhere both in TOTAL_DIPOLE and 
> MOLECULAR_DIPOLES for non-orthorombic cells..
> Is this the case?
> Daniele Passerone wrote:
>> Dear Teodoro, is it possible that this postprocessing part does not
>> work with non-orthorombic cells?
>> We tried with our cell and got again big problems...
>> daniele
> >

Dr. Fawzi Mohamed,                      Office: 3'332
Humboldt-Universitaet zu Berlin, Institut fuer Chemie
Post:               Unter den Linden 6, 10099, Berlin
Besucher/Pakete:    Brook-Taylor-Str. 2, 12489 Berlin
Tel: +49 30 20 93 7140          Fax: +49 30 2093 7136

GMX FreeDSL mit DSL 6.000 Flatrate und Telefonanschluss nur 17,95 Euro/mtl.!

More information about the CP2K-user mailing list