Post processing of SOC

Anything regarding the post-processing utility (e.g. excitonic wavefunction analysis) is dealt with in this forum.

Moderators: Davide Sangalli, andrea marini, Daniele Varsano

Post Reply
Quxiao
Posts: 96
Joined: Fri Mar 26, 2021 11:27 am

Post processing of SOC

Post by Quxiao » Fri May 05, 2023 5:17 pm

Dear all:
I used QE to calculate the DFT calculation considering SOC, and then I used the scissors operator to calculate BSE normally on top of it. However, after completing the calculation, I attempted to output the electron density of states using ypp - s s, but what puzzled me was that two files were output here, spin up and spin down, respectively.
Then I tried using - soc, and a parameter SOC_ DBs appeared in the ypp.in, how to use this?
If I use ypp, I will be prompted with [ERROR] SOC database NOT FOUND @ SOC_ DBS, please help me understand that.
Best wishes!
Quxiao
Quxiao in BIT,calculate the exciton

User avatar
Davide Sangalli
Posts: 610
Joined: Tue May 29, 2012 4:49 pm
Location: Via Salaria Km 29.3, CP 10, 00016, Monterotondo Stazione, Italy
Contact:

Re: Post processing of SOC

Post by Davide Sangalli » Mon May 08, 2023 8:38 am

Dear Quxiao,

Point 1: KS density of states
If you do

Code: Select all

ypp -s s
ypp will compute the KS DOS, including the total, the spin-up and the spin-down component.
The total is defined in the standard way.
DOS(w) ~ sum_n delta(w-eps_n)
Instead the up/down component are define as follow
DOS_sigma(w) ~ sum_n delta(w-eps_n) \int d3x |psi_n(x,sigma)|^2
where eps are the KS energies and psi_n are the KS spinors

Point 2: excitonic JDOS
The Ks DOS has nothing to do with the BSE simulation.
If you are looking for the excitonic JDOS, you need to set

Code: Select all

BSEprop="abs jdos"
in your bse input file. It will work with some of the solvers (i.e. diago and slepc)

Point 3: perturbative soc mapping
This step is meant for doing a BSE calculation with perturbative SOC.
You need two SAVE folders of the same material, one with SOC and one without SOC. All other parameters should be equivalent
Let's say that you have the two in "With_SOC/SAVE" and "Without_SOC/SAVE"
Then you can do

Code: Select all

cd Without_SOC
ypp -soc -F soc_mapping.in
then you edit the soc_mapping.in file setting

Code: Select all

SOC_DBs= "../With_SOC/"                      # Path to the folder containing the DB with SOC
Finally you can perform the mapping procedure and the perturbative BSE.

Code: Select all

ypp -F soc_mapping.in -J SOC_mapping_DBs
yambo -F bse.in -J "BSE_pert_soc,SOC_mapping_DBs" -C BSE_pert_soc
Best,
D.
Davide Sangalli, PhD
CNR-ISM, Division of Ultrafast Processes in Materials (FLASHit) and MaX Centre
https://sites.google.com/view/davidesangalli
http://www.max-centre.eu/

Quxiao
Posts: 96
Joined: Fri Mar 26, 2021 11:27 am

Re: Post processing of SOC

Post by Quxiao » Tue May 16, 2023 6:39 am

Dear Davide:
Firstly,thanks for your help!With your help,I know the dos defined in yambo,and I think that I could get the dos which consider the SOC by adding the spin up and the spin down component,is that true?
Secondly,in my opinion,if I consider the SOC in the BSE,I just do the SOC in the DFT level in QE,the SAVE contain the SOC effect,is that true?
Or I must do the thing as you mentioned:"You need two SAVE folders of the same material, one with SOC and one without SOC."Only follow that you mentioned,I could get the BSE which considered the SOC ,is that true?
Best wishes!
Quxiao
Quxiao in BIT,calculate the exciton

User avatar
Davide Sangalli
Posts: 610
Joined: Tue May 29, 2012 4:49 pm
Location: Via Salaria Km 29.3, CP 10, 00016, Monterotondo Stazione, Italy
Contact:

Re: Post processing of SOC

Post by Davide Sangalli » Tue May 16, 2023 9:26 am

With your help,I know the dos defined in yambo,and I think that I could get the dos which consider the SOC by adding the spin up and the spin down component,is that true?
Not sure I understand what you mean here.
The some of the up and down DOS, as defined below, gives, by definition, the full DOS.
Secondly,in my opinion,if I consider the SOC in the BSE,I just do the SOC in the DFT level in QE,the SAVE contain the SOC effect,is that true?
Correct
Or I must do the thing as you mentioned:"You need two SAVE folders of the same material, one with SOC and one without SOC."Only follow that you mentioned,I could get the BSE which considered the SOC ,is that true?
This is needed only if you want for perturbative SOC

Best wishes!
Quxiao
Davide Sangalli, PhD
CNR-ISM, Division of Ultrafast Processes in Materials (FLASHit) and MaX Centre
https://sites.google.com/view/davidesangalli
http://www.max-centre.eu/

Quxiao
Posts: 96
Joined: Fri Mar 26, 2021 11:27 am

Re: Post processing of SOC

Post by Quxiao » Fri May 19, 2023 5:20 pm

Dear Davide:
Thanks a lot for your reply!
The thing I want to confirm is that if we calculate a system with SOC in its DFT ground state, we only need to consider SOC in the QE calculation and then calculate BSE normally.
But in the case of considering SOC perturbation as you mentioned, preparing two folders actually means that on the basis of considering SOC in DFT ground state calculations, then in BSE calculations, when constructing BS kernels, also considering new SOC terms?Is that true?
Then, based on the ground state of DFT considering the SOC term, two BSE equations are solved, one considering the SOC perturbation term in the BSE calculation and the other not considering the SOC perturbation term in the BSE calculation. Is that the case?
Best wishes!
Quxiao
Quxiao in BIT,calculate the exciton

User avatar
Davide Sangalli
Posts: 610
Joined: Tue May 29, 2012 4:49 pm
Location: Via Salaria Km 29.3, CP 10, 00016, Monterotondo Stazione, Italy
Contact:

Re: Post processing of SOC

Post by Davide Sangalli » Sat May 20, 2023 1:08 pm

In the case of "perturbative SOC" BSE is solved in the space of scalar KS states (i.e. without SOC), but the excitonic eigenvalues are then corrected to first order using the SOC splitting.

In the case with SOC, BSE is done in the space of Spinorial KS states. This means the BSE is a factor 4 bigger compared to the previous case.

See this paper for details:
https://arxiv.org/abs/2103.02266

Best,
D.
Davide Sangalli, PhD
CNR-ISM, Division of Ultrafast Processes in Materials (FLASHit) and MaX Centre
https://sites.google.com/view/davidesangalli
http://www.max-centre.eu/

Quxiao
Posts: 96
Joined: Fri Mar 26, 2021 11:27 am

Re: Post processing of SOC

Post by Quxiao » Mon May 29, 2023 6:15 am

Dear Davide:
Thanks a lot for your help again!I have known that are two case.
I have understood these two situations. The first method is to use SOC as perturbation on the KS basis, where the KS basis vector does not consider spin and SOC is used as perturbation expansion on this basis? Or does the KS basis vector consider spin, such as independent spin down and spin up bases, and then unfold SOC on them?
The second type is originally unfolding in the so-called spin space, but at this point, the basis vectors caused by SOC may not be simply seen as separate spin up and spin down. They should be four mixed basis vectors, on which BSE can be solved. Only when SOC is not very strong can the basis vectors composed of spin up and spin down be approximately disassembled. Is this understanding correct?
I will continue to review the literature you provided and thank you very much for your help!
Best wishes!
Quxiao
Quxiao in BIT,calculate the exciton

User avatar
Davide Sangalli
Posts: 610
Joined: Tue May 29, 2012 4:49 pm
Location: Via Salaria Km 29.3, CP 10, 00016, Monterotondo Stazione, Italy
Contact:

Re: Post processing of SOC

Post by Davide Sangalli » Mon May 29, 2023 8:57 am

The first method is to use SOC as perturbation on the KS basis, where the KS basis vector does not consider spin and SOC is used as perturbation expansion on this basis? Or does the KS basis vector consider spin, such as independent spin down and spin up bases, and then unfold SOC on them?
The KS basis considers spin as independent spin down and spin up bases. If the QE run is done with n_spin=1, this is achieved by doubling the KS basis during the mapping procedure between states with and without SOC. The BSE is then solved in the spin singlets channel.
The second type is originally unfolding in the so-called spin space, but at this point, the basis vectors caused by SOC may not be simply seen as separate spin up and spin down. They should be four mixed basis vectors, on which BSE can be solved. Only when SOC is not very strong can the basis vectors composed of spin up and spin down be approximately disassembled. Is this understanding correct?
Correct
Davide Sangalli, PhD
CNR-ISM, Division of Ultrafast Processes in Materials (FLASHit) and MaX Centre
https://sites.google.com/view/davidesangalli
http://www.max-centre.eu/

Post Reply