The ABINIT Users Mailing List ( CLOSED )

[Konstantin Rushchanskii <Konstantin.Rushchanskii@physik.uni-regensburg.de>]

Dear Bohdan,

> After the calculation with 'abinis' I would like to calculate optical
> conductivity of my crystal using the 'conducti'. But I do't know which set
> of k-point for 'abinis' will be correct: one gamma-point, or the set
> generated by 'kptopt=1' and 'ptrlen=35.0' for example.
>  

I suppose you should first obtain the ground-state (and corresponding 
_DEN-file) with reasonable set of k-points.
I think, one Gamma point is not sufficient. You need a set of 
calculations with different
Ecut's and k-point sets to check the convergence. Then you should obtain 
the wave-functions with larger k-points
(it is also issue of convergence test) to build the first-order 
wavefunctios perturbations...

For ground state use
kptopt 1
ngkpt 4 4 4 (it depends on anisotropy of lattice parameters)

for d/dk calculation use
kptopt 2
nqpt 1
qpt 0 0 0


> I use PSP of the type 'xy-z.LDA.fhi'.
>  
Try to use TM  *.pspnc pseudopotentials for your elements first.
It seems, that the Ecut of 30 Ha is good for your system for all PSP's 
of this type (but check it).
When you use FHI PSP's, use
optnlxccc 2
(see the latest discussion about EOS fit in ABINIT-forum, subject 
"Built-in lattice optimization failed",
and comments of Don Hamann).

And the last thing: if the convergence test will fail with PSP's you 
use, or calculation needs an unusually large Ecut's,
then OPIUM is perfect and very usefull package for engeneering of PSP's 
with reasonable Ecut,
especially for 2-nd row elements. You can also make a series of tests 
for the PSP's using this package.

Sincerely,

--
K.Rushchanskii, Dr.
Theoretische Physik, Universität Regensburg
93040 Regensburg Germany
tel: +49 (0) 941-943-2466  fax: +49 (0) 941-943-4382