GW band structure with SOC

Deals with issues related to computation of optical spectra, solving the Bethe-Salpeter equation.

Moderators: Davide Sangalli, andrea.ferretti, myrta gruning, andrea marini, Daniele Varsano

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

Re: GW band structure with SOC

Post by Daniele Varsano » Mon May 09, 2022 8:05 am

Dear Mitra,
can you please post the report file of the GW calculations? report with SOC is missing.
Anyway, it would be useful to compare calculations done at the same degree of convergence. Moreover I can see that the k sampling of the two calculations is also different.

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/

8813204602
Posts: 58
Joined: Tue Jun 15, 2021 5:53 pm

Re: GW band structure with SOC

Post by 8813204602 » Mon May 09, 2022 4:49 pm

Dear Daniele,

Input,output and r-setup files related to GW calculations with converged parameters with and without SOC are attached. Both calculations have been performed for k-grid 6*6*1 and there is the problem mentioned in previous posts (for k-grid 21*21*1) about qp energies.
Moreover I can see that the k sampling of the two calculations is also different.
As seen in the r-setup,without the SOC, the number of k-points in the IBZ is considered for calculations, which is less than the number in the BZ, while with the SOC, the number of k-points in both of IBZ and BZ is the equal. Therefore, although both GW calculations have been performed for k-grid 6*6*1, the number of k-points in calculations without SOC is 20 and with SOC is 36:
(without SOC) IBZ Q-points : 20 , BZ Q-points : 36
(with SOC) IBZ Q-points : 36 , BZ Q-points : 36

Thanks for your help

Best,
Mitra
Last edited by 8813204602 on Wed May 11, 2022 6:05 pm, edited 1 time in total.
Mitra Helmi,Ph.D student
Ferdowsi university of Mashhad
Mashhad,Iran

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

Re: GW band structure with SOC

Post by Daniele Varsano » Tue May 10, 2022 4:03 pm

Dear Mitra,
I had a look at your report file and I could not find anything strange at first glance.
I have inspected to kpt for the two cases "gamma" and "0.500000 -0.250000 0.000000".

SOC effect indeed seems very small, KS energies, for the considered k points in the two cases differs up to 50meV while QP energies differ by about 0.2 eV.
eksqp.png



Please note that convergence parameters are different for the two cases:

Code: Select all

< # Sc/G bands                                      :    1  160
---
> # Sc/G bands                                      :    1  120
The bad shape you obtain in the SOC case can be due to a failure of the interpolation procedure.

Best,
Daniele
You do not have the required permissions to view the files attached to this post.
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/

8813204602
Posts: 58
Joined: Tue Jun 15, 2021 5:53 pm

Re: GW band structure with SOC

Post by 8813204602 » Wed May 11, 2022 1:09 pm

Dear Daniele,

Thank you very much for your help and reply and for taking the time to inspect the report file.
The bad shape you obtain in the SOC case can be due to a failure of the interpolation procedure.
Does it mean that the performed GW calculations are correct and the obtained ndb.QP database can be used for BSE calculations and only the interpolation problem related to ypp should be solved?
Please note that convergence parameters are different for the two cases:

Code: Select all

< # Sc/G bands : 1 160
---
> # Sc/G bands : 1 120
I performed the convergence test for the GbndRnge parameter for both with and without SOC. Without the SOC, I had 17 full bands for up spin and 15 full bands for down spin, so I started the convergence study from Gbnd = 20, but with the SOC, I had 32 full bands and started the convergence study from Gbnd = 40. According to the plots shown in the figure, I considered the convergent value of GbndRnge without SOC equal to 120 and with SOC equal to 160. Is this convergence process correct or must the values of GbndRnge be considered equal for both cases with and without SOC?

Thanks and Regards,
Mitra
You do not have the required permissions to view the files attached to this post.
Mitra Helmi,Ph.D student
Ferdowsi university of Mashhad
Mashhad,Iran

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

Re: GW band structure with SOC

Post by Daniele Varsano » Wed May 11, 2022 1:38 pm

Dear Mitra,
Does it mean that the performed GW calculations are correct and the obtained ndb.QP database can be used for BSE calculations and only the interpolation problem related to ypp should be solved?
That's what seems to me, even if a deeper inspection is recommendable.
Is this convergence process correct or must the values of GbndRnge be considered equal for both cases with and without SOC?
The gap seems "nearly converged", even if differences between the two calculations can still be due to a lack of a complete convergence. In principle SOC calculations needs a number of bands that is the double respect the collinear case as you have a band for each spinors.

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