plotting GW wavefunctions

[alsaidi]

Hello-

Is there a way to output the GW wavefunctions to visualize them.

Thanks.

Wissam Al-Saidi,
Univ. of Pitt.

[Daniele Varsano]

Dear Wissam,
in order to have GW wavefunctions you should calculate
quasi-particle equations self-consistently. Actually the GPL version
of Yambo does permit calculation G^oW^o which does not update
wavefunctions, under the assumption that GW wfs are "not so different"
from the Kohn-Sham wfs (which is very often the case). Why do you think
your GW wfs will be substantially different from the KS? What kind of system
are you studying?

Cheers,

Daniele

[alsaidi]

Dear Daniele,

Thank you very much for the prompt reply.
So I guess the generally big change in the KS eigenvalues does not necessarily mean that the quasi-particle wavefunctions are different from the zeroth order ones.
I understand that with a full self consistent calculations, one could get the eigenvalues and wavefunctions. But also with a "one-shot" G0W0 calculation, one can get the first iteration eigenvalues and eigenfunctions, right?
I guess the code only calculates the eigenvalues since this is what is needed in the G0W0 approach? Isn't it straightforward to calculate also the first order wavefunctions?
Is there any planning to have self-consistency implemented? I could help if needed.
By the way, my systems of interest are carbon nanotubes. I am not sure whether there are significant changes in the KS wavefunctions.
Thanks again.

Wissam

[Daniele Varsano]

Dear Wissam,
let me ask you first to put your affiliation in all your post, this is a rule of the forum.
You can put a signature in your forum profile.

I guess the code only calculates the eigenvalues since this is what is needed in the G0W0 approach? Isn't it straightforward to calculate also the first order wavefunctions?

The calculation is done as explained in the reference manual that you can find here
and even it should be not a difficult task to calculate the first order wfs, could be quite heavy to calculate the several off-diagonal element of the self-energy needed to calculate the 1st order wfs, I don't know how much it is useful.

Is there any planning to have self-consistency implemented? I could help if needed.

Yes there is, and this is a feature that has been implemented in the developer version of the code, but as it is a quite new part of the code
or the moment it is not released in the GPL version. We hope to include this part soon in the GPL release but at this stage I cannot say to
you exactly when it will happen.

By the way, my systems of interest are carbon nanotubes. I am not sure whether there are significant changes in the KS wavefunctions.

I'm not an expert in carbon nanotubes, anyway looking at many body perturbation theory calculations on these systems present in the
literature:

(a) Chang, E.; Busi, G.; Ruini, A.; Molinari, E. Phys. ReV. Lett. 2004, 92, 196401.
(b) Spataru, C. D.; Ismail-Beigi, S.; Benedict, L. X.; Louie, S. G. Phys. ReV. Lett. 2004, 92, 077402.
(c) Gruning, A. Marini, and X. Gonze, Nano Lett. 9, 2820 (2009)

the G0W0 note approximation looks to work nicely.

Hope this helps,

Daniele

[alsaidi]

Thanks again.
In the ypp postprocessing utility, one can plot an average of the wavefunctions.
I guess these are the KS wavefunctions, right?
Is there any documentation for ypp?

Another subject which I have not seen discussed in the manual is the optimum way for performing a GW convergence study. There are several parameters which one should check. I believe the list includes only FFTGvecs, EXXRLvcs, BndsRnX, NGsBlkXp, and GbndRnge. Should one check the convergence on the first independent of the others, for example and so on. (please let me know if I need to start a new thread for this topic).

Best, Wissam
Univ. of Pitt.

[Daniele Varsano]

Using ypp you can plot both KS wavefunction (ypp -s w) as well as
excitonic wave-function after having solved the Bethe-Salpeter equation
by diagonalization (ypp -e w). Typing ypp -H you will have a little help on the options.
In this part of the tutorial
you can find some documentation.

About the convergences test, yes you have to check the convergences of all the variable you mentioned,
and usually you can treat them as independent.

Best,

Daniele

[alsaidi]

Thanks for the info.

Wissam
Univ. of Pitt.