Band Structure

[Yunfeng]

Dear All,

Is it possible to calculate GW corrections for the whole band structure, like Marini, Onida and Del Sole Phys. Rev. Lett. 88, 016403 (2002)?

Thanks
Sincerely, Yunfeng

[andrea marini]

Is it possible to calculate GW corrections for the whole band structure, like Marini, Onida and Del Sole Phys. Rev. Lett. 88, 016403 (2002)?

Dear Yungfeng, your question is tricky. I managed to get the full band structure of Cu by using a complicate, ad-hoc interpolation of the GW corrections calculated on a set of high-symmetry points.

You can try to do the same. First calculate the GW corrections on as much high-symmetry points as you can. Then the method I used was to gradually deform the LDA bands in order to match the GW bands calculated on these high-symmetry points. Be careful to impose the GW, interpolated bands to approach the high-symmetry points with zero derivative (as a function of the momentum).

Hope I was hepful

[sdwang]

Is it possible to calculate GW corrections for the whole band structure, like Marini, Onida and Del Sole Phys. Rev. Lett. 88, 016403 (2002)?

Dear Yungfeng, your question is tricky. I managed to get the full band structure of Cu by using a complicate, ad-hoc interpolation of the GW corrections calculated on a set of high-symmetry points.

You can try to do the same. First calculate the GW corrections on as much high-symmetry points as you can. Then the method I used was to gradually deform the LDA bands in order to match the GW bands calculated on these high-symmetry points. Be careful to impose the GW, interpolated bands to approach the high-symmetry points with zero derivative (as a function of the momentum).

Hope I was hepful

Dear andrea marini,
You said 'calculate the GW corrections on as much high-symmetry points as you can', but in GW calculation, the k-ponit are from the pw scf and nscf file, and this is uniform grids. How can I chose 'as much high-symmetry points as I can'?
S. D. Wang

[claudio]

Dear Yungfeng

converge your GW calculation on a regular grid and the shift it on the points where you need yo calculate GW corrections, as it is explained here:

http://www.yambo-code.org/doc/inputs/ypp_kk.php

Claudio

[sdwang]

Dear developers:
I get some QP corrections at some high-symmetry k-poins, but how can I get the 'cond. stretching | val. stretching' as following input file?

QP corrections using a scissor operator and valence/conduction stretching.

scissor | cond. stretching | val. stretching

[claudio]

Dear Yungfeng

in order to get 'cond. stretching | val. stretching' you should fit your GW corrections for
the points you calculated as function of the energy, with a function as:

let's call:
Vs = val. stretching
Cs = cond. stretching
E_homo = top valence bands energy
E_lumo = bottom energy of conduction bands

the scissor operator with stretching is defined as:

for valence bands: E_gw = E_lda + (Vs - 1) * (E_homo - E_lda)
and for conductions: E_gw = E_lda + E_scissor + (Cs - 1) * (E_lumo - E_lda)

check also the file src/qp_ctl/QP_apply_global_stretch.F
to see how it works

Claudio

[sdwang]

Dear Yungfeng

for valence bands: E_gw = E_lda + (Vs - 1) * (E_homo - E_lda)
and for conductions: E_gw = E_lda + E_scissor + (Cs - 1) * (E_lumo - E_lda)

Claudio

I think it should be (E_lda-E_lumo) . Because the E_lumo is the lowest energy in the band, so E_lda larger than it.Am I right?
I noticed in http://www.yambo-code.org/tutorials/Rea ... /index.php,
# K-point Band Eo Eqp E-Eo LDA HF Sc(Eo) Sc(Eqp) Sc`(Eo) Z
#
1.00000 3.00000 0.00000 -2.52528 -2.52528 -22.71821 -30.22404 4.46001 4.98054 -0.20612 0.82910
1.00000 4.00000 0.00000 -2.52208 -2.52208 -22.71822 -30.22392 4.46344 4.98361 -0.20624 0.82902
1.00000 5.00000 9.00799 10.95764 1.94965 -11.07205 -5.98616 -2.89262 -3.13624 -0.12495 0.88893
1.00000 6.00000 19.90535 22.23744 2.33210 -8.85649 -2.74308 -3.40730 -3.78131 -0.16034 0.86180
#
By plotting some of the o-PPA.qp columns it is possible to discuss some physical properties of the LiF QPs. Using columns 3 and 4 we can deduce the band gap renormalization and the stretching of the conduction/valence bands

In the plot, the band stretching is the slope of the fitting lines?