[CPMD-list] Pseudopotentials and diferences in results

[Juerg Hutter]

Dear Jim

the problem is related to your input using MT pseudos
You probably want to use the Kleinman-Bylander form of
these pseudopotentials. However, not specifying anything
defaults in the current version to a Gauss-Hermite
integration using 0 points -> neglect of non-local part.

I suggest you use
*C_MT_BLYP_2 Kleinman-Bylander
 LMAX=P

for all elements with LMAX > S.
To get good numerical results you should also increase
the cutoff. 70-85 Ry for TM pseudos and 80-120 Ry
for the Goedecker pseudos.

regards

Juerg Hutter


----------------------------------------------------------
Juerg Hutter                   Phone : ++41 1 635 4491
Physical Chemistry Institute   FAX   : ++41 1 635 6838
University of Zurich           E-mail: hutter at pci.unizh.ch
Winterthurerstrasse 190
CH-8057 Zurich, Switzerland
----------------------------------------------------------


On Mon, 5 Apr 2004, Jim Kress wrote:

> I'm using CPMD v3.7.2 on a small Linux cluster and am a novice in the
> application of the CPMD code.   I am a Computational Chemist by education
> and experience with quite a bit of experience working with pcGAMESS and
> similar Quantum Chemistry codes.
>
> To familiarize myself with the CPMD code and its applications I have been
> performing some single molecule calculations.  I recently tried some
> calculations where I used different pseudopotentials.  I got some rather (to
> me) strange looking results, i.e., for my molecule of interest (using the
> Goedecker pseudopotentials distributed with CPMD) I calculated a total
> energy of -54.265019 au.  However, using exactly the same input file but
> specifying the Trouiller-Martins pseudopotentials (also distributed with
> CPMD) I calculate a total energy of -85.349239 au.
>
> Is this to be expected or do I have a whopper of an error in my input?
>
> I've attached in input and output files for inspection.
>
> Any comments, suggestions, or explanations of why these results are observed
> would be greatly appreciated.
>
>
> Jim
>