The ABINIT Users Mailing List ( CLOSED )

[Pio Baettig <baettigp@gmx.net>]

Dear Professor Torrent,

thank you very much for you help. The main difference stemmed from the
fact, that I was using Gamma-point sampling with HiLAPW and auto with
abinit.. I repeated the calculations and now the results agree much better.

Now the results agree very well with HiLAPW.

I also compared results with HiLAPW for BiAlO3 in 3 different
structures. I used the attached PAW for Bi and the Al- and O- PAWs from
N. Holzwarths site.

In the attached pdf, the results of the analysis can be found.

-On the first and second page, the logarithmic derivatives are plotted.
(black all electron, red PAW) for l=0 there is a small discontinuity at
about 4.5 Ry which does not seem to influence the results.

-On pages 3-5 are the wavefunctions and projectors.

-On page 6 are the densities of state for metallic Bi calculated with
abinit and HiLAPW. The features agree very well, but with 6*6*6
k-points, the calculations are not fully converged.

-On page 7 are the DOS for R3c BiAlO3 calculated with HiLAPW and abinit.
The agreement between the two codes is really well.

-Page 8 is the summary of the convergence tests on the following pages.
Ground state energies have been calculated for three different
structures of BiAlO3. The results 'PAW US' are the ones obtained with
the attached PAW. The energy-differences between two structures for
calculations with abinit and HiLAPW agree within 30 meV

-On the last page are convergence results for metallic Bi. it can be
seen, that the energies converge rather fast with respect to the cutoff
energy. for 20-25 Ha, the energy-difference is already smaller than 1mHa.

Best regards,

Pio Baettig


P.S. The PAW will follow in the next mail


Marc Torrent wrote:
> Hi,
>
> Just to be sure...
> Did you perform all calculations with the tetrahedron method (i.e. 
> prtdos=2 for Abinit, ismear=-5 for vasp or lbroad=-1/0 for hilapw) ?
> Are you sure that the k-points sets are the same for the three codes ? 
> (note that Abinit uses by default a shifted grid...see shiftk keyword).
>
>
> Marc Torrent

Attachment: Bi_analysis_PAW.pdf
Description: Adobe PDF document