Silicon GW correction

[jbae]

I calculated quasi-particle energy of bulk silicon within the GW approximation using PPA (same as tutorials)
Obtained quasi-particle correction is very similar to that of tutorials. Below is some part of o.qp file.

# K-point Band Eo E-Eo Sc(Eo)
....
1.00000 1.00000 -11.93283 -2.62226 4.12997
1.000000 2.000000 0.000000 0.321444 1.536214
1.000000 3.000000 0.000000 0.321384 1.536104
1.000000 4.000000 0.000000 0.323863 1.538630 : valance band maximum (VBM)
1.000000 5.000000 2.557771 1.538283 -2.709212 : conduction band minimum (CBM)
1.000000 6.000000 2.557771 1.539280 -2.708224
....
Is it reasonable that qp energy correction(E-Eo) of both VBM and CBM are positive?
As I know, qp energy correction of VBM is negative and that of CBM is positvite.

[Daniele Varsano]

Dear Jaehyun Bae,

As I know, qp energy correction of VBM is negative and that of CBM is positvite.

this is what very often happen, but it is not a rule.
In your case, I think that the results you get are consistent with other similar calculations:
see for instances these papers:
Godby,Schluter,Sham, PRB 37, 10159, 1988, available here:
http://www-users.york.ac.uk/~rwg3/Paper ... 20PRB).pdf
or:
http://www.jim.or.jp/journal/e/pdf3/51/12/2150.pdf

Cheers,

Daniele

[Sheleon]

Dear Jaehyun Bae and Daniele
I think the reason for both of the VBM and CBM are positive is the calculation is not converged. When i learn the Si case of the tutorials, at first, VBM and CBM are both positive indeedly, but with rasing the parameter NGsBlkXp, the VBM will turn into negative and CBM is still positive with a relatviely small value.
And The fig.1 from the second reference which Daniele listed, they also obtained a negative correction to the VBM and a positive correction to CBM.
Since i am also a new beginer of Yambo, if there someting wrong, please let me know~

Cheers!

Sheleon

[Daniele Varsano]

Dear Sheleon,
nothing wrong in what you write.
As you argued, it is important to stress that the tutorial examples are usually out-of-convergence.
We often use this examples in the Yambo schools, and they are thought to be run in simple PC in a reasonable time,
and the parameter they should *not* be taken for production runs.
About the references I listed, you are right that the VBM has a general negative corrections, even if
at the gamma point LDA and GW are practically the same, with a very small positive correction, at
least looking at table 1. Next it looks that the sign corrections changes changing methodology, pseudopotentials...etc.

Cheers,

Daniele