GW bands get tangled up

Run-time issues concerning Yambo that are not covered in the above forums.

Moderators: myrta gruning, andrea marini, Daniele Varsano, Conor Hogan

YChen
Posts: 15
Joined: Wed Apr 14, 2021 2:55 am

Re: GW bands get tangled up

Post by YChen » Sun Sep 26, 2021 8:57 am

Dear Dr. Daniele,

Thank you for your reply! The quasiparticle band calculated by GW method could not reproduce the shape of LDA band, and the phenomenon of band crossing exists in the conduction band, as shown in the figures below. A similar phenomenon occurs in the valence bands of other materials I calculated. From your previous reply, I have known that the problem occurred in the process of GW calculating truncation instead of YPP interpolation, but I still failed to eliminate the band crossing phenomenon after many attempts.
band-GW.png
band-LDA.png
Best,
Ying Chen
You do not have the required permissions to view the files attached to this post.
Ying Chen
Academy for Engineering and Technology, Fudan University, Shanghai 200433, China

User avatar
Daniele Varsano
Posts: 3773
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: GW bands get tangled up

Post by Daniele Varsano » Sun Sep 26, 2021 10:04 am

Dear Chen,

thanks for reporting, I've not well understood if the problem is due to interpolation or you have crossing in GW results (which is possibile and not a big problem).

Could you plot your calculated (not interpolated) QP energies along the path, e.g. each band different color points?

Best,
Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/

YChen
Posts: 15
Joined: Wed Apr 14, 2021 2:55 am

Re: GW bands get tangled up

Post by YChen » Sun Sep 26, 2021 11:57 am

Dear Dr. Daniele,

Thank you for your reply! Can Yambo software directly output GW energy along a specific path? Or do I have to manually process the data stored in o-all_Bz.qp?

Best,
Ying Chen
Ying Chen
Academy for Engineering and Technology, Fudan University, Shanghai 200433, China

YChen
Posts: 15
Joined: Wed Apr 14, 2021 2:55 am

Re: GW bands get tangled up

Post by YChen » Mon Sep 27, 2021 3:52 am

Dear Dr. Daniele,

I have plotted the calculated (not interpolated) QP energies along the path of 'Γ-M' as shown in the figure below.
Graph1.png
Band crossing has appeared in GW results, how should I solve this problem?

Best,
Ying Chen
You do not have the required permissions to view the files attached to this post.
Ying Chen
Academy for Engineering and Technology, Fudan University, Shanghai 200433, China

User avatar
Daniele Varsano
Posts: 3773
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: GW bands get tangled up

Post by Daniele Varsano » Mon Sep 27, 2021 9:03 am

Dear Ying ,

it is possibile that GW correction are different along the path for each band (for instance when the character of the bands changes along the path) and you end up with crossed bands, but actually it is a matter of reindexing the bands, ie a GW bands continues with another index and you do not have a real crossing. If this is the case you can try to plot your bands with no lines, just points, and see if you have continuous bands.
This is possibile as looking at your DFT bands it seems you have several avoided crossing.
If this is the case anyway I imagine that the interpolator will suffer (btw did you try to use the BOLTZ algorithm?).
I imagine that you also checked all the convergence parameters.


Best,
Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/

YChen
Posts: 15
Joined: Wed Apr 14, 2021 2:55 am

Re: GW bands get tangled up

Post by YChen » Mon Sep 27, 2021 1:29 pm

Dear Dr. Daniele,

As you said, I drew the data points calculated by GW (without interpolation), and the conduction band in the Γ-M direction is shown in the figure below. I am not sure whether the band crossing phenomenon exists.
Graph4.png
If this is the case anyway I imagine that the interpolator will suffer (btw did you try to use the BOLTZ algorithm?).
The picture I showed is the interpolation result of the BOLTZ algorithm, as shown in the figure below. The band interpolated by the NN algorithm is jagged.
band-GW.png
I imagine that you also checked all the convergence parameters.
Yes, I have performed the convergence test of NGsBlkXp, BndsRnXp, GbndRnge, RandQpts and RandGvec on the energy at a single k point and band. Will these parameter values affect the shape of the band? Or only affect the energy value?


Best,
Ying Chen
You do not have the required permissions to view the files attached to this post.
Ying Chen
Academy for Engineering and Technology, Fudan University, Shanghai 200433, China

User avatar
Daniele Varsano
Posts: 3773
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: GW bands get tangled up

Post by Daniele Varsano » Mon Sep 27, 2021 4:05 pm

Dear Ying,
Will these parameter values affect the shape of the band? Or only affect the energy value?
The shape of the bands exactly reflects the QP energies.

Be sure they are well converged, next you can try the following solutions:

1) Interpolate the QP correction instead of the QP energy as explained in this post.
2) Use Wannier90 interface as explained in this post by Antimo Marrazzo.

Best,

Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/

YChen
Posts: 15
Joined: Wed Apr 14, 2021 2:55 am

Re: GW bands get tangled up

Post by YChen » Sat Oct 30, 2021 5:10 am

Dear Dr. Daniele,

Thank you very much for the information! I have tried the solutions. But the band structure obtained by interpolating the QP correction instead of the QP energy is not ideal, the band crossing still exists and the band seriously wiggles, as shown in the figure below.
shan.png
Best,
Ying Chen
You do not have the required permissions to view the files attached to this post.
Ying Chen
Academy for Engineering and Technology, Fudan University, Shanghai 200433, China

User avatar
Daniele Varsano
Posts: 3773
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: GW bands get tangled up

Post by Daniele Varsano » Sun Oct 31, 2021 12:40 pm

Dear Ying,
it seems there it something inconsistent comparing this plot with the one before, looking to the band that are not critical (e.g. valence bands and first conduction band).
In any case I suggest you to try to perform a Wannier interpolation.

Best,
Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/

Post Reply