dielectric function with transferred momenta using GW+BSE method

Questions and doubts about features of non linear optic in Yambo (yamb_nl)

Moderators: Davide Sangalli, claudio, myrta gruning

Post Reply
Liang
Posts: 3
Joined: Mon May 01, 2023 1:13 pm

dielectric function with transferred momenta using GW+BSE method

Post by Liang » Mon May 01, 2023 2:57 pm

Dear all,

I am planning to calculate the dielectric function of a given material with transferred momenta (i.e., q is not equal to 0) using the GW+BSE method. While reading the "BSE hBN Yambo Virtual 2021 version" (I really appreciate your hard work and selfless and detailed guidance), I noticed that all the BSE solvers seem to be performed under the condition of q approaching zero, rather than arbitrary transferred momenta.

I think the transferred momenta could be added by setting the corresponding parameter in the input files, but I don't know the specific operation. Should I set the 'BSEQptR' parameter in the BSE Hamiltonian or make corresponding settings in the input file of the BSE solver? By the way, can someone please tell me the format of BSEQptR, as I cannot find it in https://www.yambo-code.eu/wiki/index.ph ... s#QPkrange? For example, if I need to calculate N q-points, is the following format correct?

Code: Select all

% BSEQptR
  1 |  N |                             # [BSK] Transferred momenta range
Also, are the q points generated directly from the k points that I have set in the input file of the SCF and NSCF calculations?

----------------------------------------
Liang
University of Science and Technology of China
G.Y. Liang
University of Science and Technology of China

User avatar
Davide Sangalli
Posts: 605
Joined: Tue May 29, 2012 4:49 pm
Location: Via Salaria Km 29.3, CP 10, 00016, Monterotondo Stazione, Italy
Contact:

Re: dielectric function with transferred momenta using GW+BSE method

Post by Davide Sangalli » Tue May 02, 2023 8:06 am

Dear Liang,
the format is correct.

Code: Select all

% BSEQptR
  1 |  N |                             # [BSK] Transferred momenta range
%
and it is needed both for the construction of the BSE kernel and for the solver step.

Please notice that, if you have a system with SOC, you need to use a specific branch:
https://github.com/yambo-code/yambo/tre ... nite-q-dir

Best,
D.

P.S.: please add the signature in the settings of your user.
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/

Liang
Posts: 3
Joined: Mon May 01, 2023 1:13 pm

Re: dielectric function with transferred momenta using GW+BSE method

Post by Liang » Tue May 02, 2023 9:32 am

Davide Sangalli wrote: Tue May 02, 2023 8:06 am Dear Liang,
the format is correct.

Code: Select all

% BSEQptR
  1 |  N |                             # [BSK] Transferred momenta range
%
and it is needed both for the construction of the BSE kernel and for the solver step.

Please notice that, if you have a system with SOC, you need to use a specific branch:
https://github.com/yambo-code/yambo/tre ... nite-q-dir

Best,
D.

P.S.: please add the signature in the settings of your user.
Dear Mr Sangalli:

Thank you for your guidance! I have already calculated the dielectric function with different transferred momenta.And may I ask another question?

I have come across a theoretical issue when performing GW+BSE calculations. In solving the BS kernel, which refers to the eigenvalues and eigenvectors of the two-body Hamiltonian(see below), many examples directly use the uncorrected band structure obtained from scf+nscf without introducing self-energy corrections.

Code: Select all

yambo -o b -k sex -F input.in -J workdir
Image

Self-energy corrections, such as GW corrections or scissor operators, appear to only manifest in the energy correction of Eλ when solving for the optical absorption spectrum (i.e., dielectric function). The eigenvalues (A^{ \lambda}_{c v k} ) of the two-body Hamiltonian used in this process are still based on the uncorrected band structure. Is it an approxiamtion method? Or is this approximation acceptable?

Image
G.Y. Liang
University of Science and Technology of China

User avatar
Davide Sangalli
Posts: 605
Joined: Tue May 29, 2012 4:49 pm
Location: Via Salaria Km 29.3, CP 10, 00016, Monterotondo Stazione, Italy
Contact:

Re: dielectric function with transferred momenta using GW+BSE method

Post by Davide Sangalli » Tue May 02, 2023 9:38 am

Dear Lian,
the DFT band structure underestimates the band gap. If you use it, also the energy position of the excitonic peak will likely be underestimated.

There are different options to address this issue. You can for example:

1) Add a scissor operator correction (end eventually a stretching) to the DFT band structure.
Relevant input variable here for a scissor of 2.4 eV and a stretching of 10% both in valence (c) and conduction (v)

Code: Select all

% KfnQP_E
 2.400000 | 1.100000 | 1.100000 |        # [EXTQP BSK BSS] E parameters  (c/v) eV|adim|adim
%
2) Load the QP correction from a GW calculation done with yambo

Code: Select all

KfnQPdb=" E < 03_QP/ndb.QP"
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/

Liang
Posts: 3
Joined: Mon May 01, 2023 1:13 pm

Re: dielectric function with transferred momenta using GW+BSE method

Post by Liang » Tue May 02, 2023 11:25 am

Davide Sangalli wrote: Tue May 02, 2023 9:38 am Dear Lian,
the DFT band structure underestimates the band gap. If you use it, also the energy position of the excitonic peak will likely be underestimated.

There are different options to address this issue. You can for example:

1) Add a scissor operator correction (end eventually a stretching) to the DFT band structure.
Relevant input variable here for a scissor of 2.4 eV and a stretching of 10% both in valence (c) and conduction (v)

Code: Select all

% KfnQP_E
 2.400000 | 1.100000 | 1.100000 |        # [EXTQP BSK BSS] E parameters  (c/v) eV|adim|adim
%
2) Load the QP correction from a GW calculation done with yambo

Code: Select all

KfnQPdb=" E < 03_QP/ndb.QP"
Best,
D.
Dear Sangalli:

Thanks a lot! It really helps me.

Best wishes
G.Y. Liang
University of Science and Technology of China

Post Reply