The ABINIT Users Mailing List ( CLOSED )

[TORRENT Marc <marc.torrent@cea.fr>]

Dear Pio,

Your problem isn't related to PAW... but to the use of the prtdos input 
keyword.
If you choose prtdos 1 (this is the case here), and if your number of 
k-points, connected to the searing temperature (tsmear) is too small, 
you get a picture as yours.
You have several solutions:
- continue to use prtdos 1 and increase tsmear (poor efficiency)
- use prtdos 2 : the tetrahedron method (recommended)
- use prtdos 3 : projected DOS + tetrahedron; With PAW (and abinit 
v5.5.4, not 5.4.4), you can also use pawprtdos 2 to speed up the 
computation.
 Be careful, prtdos 3 is not parallelized with PAW; so you have to 
1-converge the SCF cycle (in parallel) ; 2-plot the DOS (sequential 
run), using iscf=-3.
 It will be parallelized in the future 5.6 version of Abinit.

I enclose ehat I get with prtdos 3, pawprtdos 2, using exactly your 
input files and PAW dataset.
As you can see, the agreement is perfect with LAPW !

Last warning : the Fermi energy written in the DOS file is wrong if you 
use iscf=-3; you have to use the value in the .out file. This has been 
corrected in the 5.6 version,

Regards,

Marc

Pio Bättig a écrit :
> Dear Abinit-users,
>
> I am tryin to generate a PAW for In using atompaw 2.2.
> I managed to get the logarithmic derivatives to coincide nicely (other 
> than l=3, but In does not have f-electrons), as well as the 
> wavefunctions (other than wfn3 and wfn5, cf. pdf-file).
>
> However, if I run a test-calculation, there seems to be a big problem 
> with a heavily oscillating DOS, which is at the right energies 
> (compared to a LAPW-calculation).
>
> I tried to change the pseudopotential scheme and use different rc_i 
> for the partial waves (according to the atompaw usersguide, (Step 2.), 
> but without success.
>
> I would be grateful for any hints.
>
> Best regards,
>
> Pio Baettig