[CPMD-list] Gaussian Core

[EG Kim]

Dear Prof. Hutter and Dr. Kohlmeyer,

Thank you very much for the replies.

In fact, I was running VASP (with PAW) to get the core charge in order to
complement my CPMD calculations---and then I started thinking it might be even
possible with CPMD with a few tweaks....

I already tried Henkelman's code on DENSITY but the resulting atomic charges
don't seem physical; my system has C, S, O, and H (BTW, what I meant by 'new' is
a new algorithm that does the 'old' Bader analysis a lot faster, as the authors
claim!).

Best regards,

EG Kim

On Tue, 16 Oct 2007, EG Kim wrote:

EGK> Dear List Subscribers,
EGK>
EGK> I wonder if there's a simple way to write the Gaussian core electron
density
EGK> into DENSITY along with the usual valence electron density (without
modifying
EGK> the code!).  I am trying to use a new grid-based Bader analysis program

the short answer to this is... no.

EGK> (developed by G. Henkelman at UT Austin).  For the program to work
correctly, I
EGK> need the charge density maxima (whether real or not) at the atomic centers.
 I
EGK> am using the latest version of CPMD (3.11).  Thank you very much for
reading.

depending on what property you are looking for and what
accuracy you expect, you can probably still can get some
useful results with the bader code (it is not _that_ new, btw).

other than that, you'll have to add the core density. in fact,
if you want it accurately, you'll have to record that during the
generation of the pseudopotentials, i.e. you'll have to build
your own pseudopotentials and potentially hack the atomic code, too.

an easier solution would be to use codes that implement some
version of the PAW method or something equivalent.


cheers,
  axel.
EGK>
EGK> Regards,
EGK>
EGK> EG Kim
EGK>
EGK> _______________________________________________
EGK> CPMD-list mailing list
EGK> CPMD-list at cpmd.org
EGK> http://cpmd.org/mailman/listinfo/cpmd-list
EGK>

Hi

there is no built in option to achieve what you want.
However, it would only require a few lines of code to
do it. The problem I see is, that using the standard
Gaussian core charges (with opposite sign) would
considerably change the valence density.
I would therefore suggest to take Gaussians with
a larger exponents. That causes the problem that your
grid might not be accurate enough and you might get
charge wiggels.

regards

Juerg Hutter

----------------------------------------------------------
Juerg Hutter                   Phone : ++41 44 635 4491
Physical Chemistry Institute   FAX   : ++41 44 635 6838
University of Zurich           E-mail: hutter at pci.uzh.ch
Winterthurerstrasse 190
CH-8057 Zurich, Switzerland
----------------------------------------------------------