The ABINIT Users Mailing List ( CLOSED )

[ธนูสิทธิ์ บุรินทร์ประโคน <thanusit@kku.ac.th>]

Dear Fabien

Thank you very much for your reply. Sorry if I didn't express my points
clearly. 
I hope to improve myself on this.

Your answers make me sure that I've got it right about (1) to keep the
KSS and SCR files constant for a GW calculation, and (2) to perform the
linear interpolation from the plot of E-E0 with respected to E0.

Concerning the input file, I did it that way because I though Abinit 
calculates GW corrections to the only two specified bands at a time, at
each k-point. Am I missing any big points here?

With the single data set you suggested below, I wonder if Abinit
calculates the 
E0, ..., E-E0, E of all bands from 1 to 16 or just returns those of band
1 and 16?
I'll try to fiugre this out myself, but any clue or advise before hand
would be
helpful. Many thanks.


Best Regards,

Thanusit Burinprakhon



> On October 3, 2007 3:55:06 PM ICT, "Fabien Bruneval"
<fabien.bruneval@polytechnique.fr> wrote:
> 
> Dear Thanusit,
> 
> Sorry, but I didn't get all your points, but it seems that your 
> calculations are perfectly sounded.
> 
> Keeping the KSS and SCR files constant is the "standard" GW procedure, 
> which is performed perturbatively with respect to LDA. You may want to 
> read some articles to convince yourself:
> Hybertsen & Louie PRB 34, 5390 (1986)
> Aulbur, Jonsson & Wilkins, Solid State Phys. 54, 1 (2000).
> 
> In the tutorial, the GW calculations are refered to as "cumbersome", but 
> in the meantime, the computers have improved a lot...
> 
> I have seen that your input file uses the datasets in a strange way. You 
> could instead calculate all the GW band structure in one single dataset.
> nkptgw  9
> kptgw
>     -0.12500 -0.12500  0.00000
>     -0.25000 -0.25000  0.00000
>     -0.12500  0.37500  0.00000
>     -0.12500  0.50000  0.12500
>     -0.25000 -0.37500  0.12500
>     -0.12500  0.12500  0.00000
>      0.50000  0.50000  0.00000  # X-equivalent?
>      0.50000  0.00000  0.00000  # L
>      0.00000  0.00000  0.00000  # (Gamma)
> bdgw
>     1   16
>     1   16
>     1   16
> etc...
> 
> And finally, for the linear interpolation, I advise you to interpolate 
> not directly E, but rather E-E0, which is quite k-point independent!
> 
> I hope I answered your worries.
> 
> Fabien
> 
>
Dr Thanusit Burinprakhon
Physics Department, Faculty of Science
Khon Kaen university
Khon Kaen, Thailand
40002

tel 00 66 4320-2222 ext. 2248
mobile  00 66 86-8620510