How to obtain real part of phonon self-energy?

General discussion around the EPW software

Moderator: stiwari

Post Reply
hellolori
Posts: 26
Joined: Mon Dec 18, 2017 5:19 am
Affiliation:

How to obtain real part of phonon self-energy?

Post by hellolori »

Dear all,

The imaginary part of phonon self-energy is related to phonon linewidth, which are written in the file of "linewidth.phself", as well as the standard output file. My question is how to obtain the real part of the self-energy from EPW?

Bests!

hlee
Posts: 415
Joined: Thu Aug 03, 2017 12:24 pm
Affiliation: The University of Texas at Austin

Re: How to obtain real part of phonon self-energy?

Post by hlee »

Dear hellolori:

Currently EPW doesn't print out the real part of phonon self-energy.
If you need the real part of phonon self-energy, you need to modify the subroutine of selfen_phon_q in selfen.f90 to write the real part of phonon self-energy; all ingredients are there.

Sincerely,

H. Lee

tpsdxy
Posts: 1
Joined: Sun Feb 05, 2023 2:31 am
Affiliation: CQU

Re: How to obtain real part of phonon self-energy?

Post by tpsdxy »

Dear all,
now, i want to obtain the real part of phonon self-energy. Then, i add the code to output the real part. As displayed in Fig1,
weight_real: the factor of the real part of phonon self-energy, weight_imag : the factor of the imaginary part of phonon self-energy.

question 1:
As shown in Fig.2, the outputed imaginary part of phonon self-energy by using weight_imag is not consistent with that of weight (phonon linewidth). I want to know the reason. Is that something wrong with the polynomial of weight_imag: AIMAG (cone/(ekq-ekk-wq(imode) -ci * degaussw0)) ? , this formula is used with blue lines in Fig.1.
Attachments
Fig2.jpg
Fig2.jpg (180.32 KiB) Viewed 15895 times
Fig1.png
Fig1.png (45.55 KiB) Viewed 15895 times

hmori
Posts: 17
Joined: Fri Mar 17, 2023 4:28 pm
Affiliation: Binghamton University

Re: How to obtain real part of phonon self-energy?

Post by hmori »

Hi,

Several approximations exist for the delta function. From equations (1) and (2) (see the attached image), we use the fact that the imaginary part of 1/(x-iε) can be regarded as approximately Gaussian with infinitesimal ε and σ. Since both ε and σ are infinitesimals but different quantities, you need to check their convergence separately. I think you should change the value of degaussw0 for 1/(x-iε) instead of using that of Gaussian.

Best,
Hitoshi
Attachments
delta_functions.png
delta_functions.png (40.04 KiB) Viewed 15289 times

Post Reply