The ABINIT Users Mailing List ( CLOSED )

["Anglade Pierre-Matthieu" <anglade@gmail.com>]

Hello,

For some materials we witness large discrepancies when using Abinitio
pseudopotentials. It is however possible to tune the pseudopotential
in order to reproduce better usual calculation (like cell parameter)
than full electron spectrum.

Also, your calculation seems quite correct. Yet I wonder:
Are you sure your Ecut is correct ? It looks small for a d-metal +
oxide ? What is the level of absolute convergence you impose for
k-point and wave function cutoff.  You mention using TM psps and I
would expect the cutoff to be higher.  One expect much higher cutoff
energies when using norm conserving pseudopotentials than PAW psps.

regards

PMA



On 9/27/07, steeve.chretien@videotron.ca <steeve.chretien@videotron.ca> wrote:
> Dear Abinit users,
>
> I am experienced VASP user trying to learn Abinit. I did all four tutorials
> but I still have some problems. I'll appreciate if someone could give me
> some hints.
>
> I am trying to reproduce the experimental structure of Ag2O and AgO. With
> VASP, the lattice parameters are within 1% of the experimental values using
> the PBE functional, PAW pseudopotential, and a 2x2x2 Monkhorst-Pack grid.
> With Abinit 5.3.5, the lattice parameters are over estimated by at least 5%
> using the PBE functional and the Troullier-Martin (FHI-Berlin) optimized for
> PBE. The problem is not with the automatic optimization of the cell
> parameters (ioptcell=1) because they are overestimated as well when I
> performed a volume scan. I looked at the rutile TiO2 structure to see if the
> problem was related to the oxide. With VASP the lattice parameters are
> within 0.5% of the experimental values while the errors are ~3.5% with
> Abinit. Could the difference between VASP and Abinit due to pseudopotential?
> How can improve my calculations?
>
> A typical input file for Ag2O looks like:
>
> ndtset 4
> prtkpt 1
> kptopt 1
> kptrlatt1  6  0  0    0  6  0    0  0  6
> kptrlatt2 -4  4  4    4 -4  4    4  4 -4
> kptrlatt3  8  0  0    0  8  0    0  0  8
> kptrlatt4 -5  5  5    5 -5  5    5  5 -5
> ACELL 3*4.72 Angstrom
> RPRIM
> 1.000  0.000  0.000
> 0.000  1.000  0.000
> 0.000  0.000  1.000
> NATOM       6
> NTYPAT      2
> TYPAT 4*1 2*2
> ZNUCL 47 8
> XRED
>   0.250  0.250  0.250
>   0.250  0.750  0.750
>   0.750  0.750  0.250
>   0.750  0.250  0.750
>   0.000  0.000  0.000
>   0.500  0.500  0.500
> IXC        11
> ECUT       15.0
> NBAND      35
> OCCOPT      7
> TSMEAR      0.002
> DIEMAC     12.0
> ISCF        7
> NPULAYIT   10
> NSTEP      30
> NSPPOL      1
> TOLDFE      0.000001
> ionmov   2
> optcell  1
>
> Thanks in advance,
>
> All the best,
>
> Steeve
>


-- 
Pierre-Matthieu Anglade