EPW Forum

Dear EPW developers and EPW experts,
I notice that in EPW, there are two different ways implemented to calculate "a2f" function.
One is to set "eliashberg = .true." and "liso = .false. & laniso = .false.". The subroutine evaluate_a2f_lambda generates a "PREFIX.a2f" file.
The other is to set "a2f = .true.". The subroutine selfen_phon_q calculates all \lambda_qv and then subroutine eliashberg_a2f generates a "PREFIX.a2f.01" file.
Are these two "a2f" files supposed to be identical? From my simple test, they look different (probably something is wrong in my calculations).
More importantly, I implemented band-resolved a2f functions following the two methods (by fixing ibnd and jbnd). The band-resolved a2f functions calculated from the above two methods are very different (for the same ibnd and jbnd).
Which method shall I use to implement band-resolved a2f file? Or the two methods should be identical and there are some bugs in my implementation?
Thank you very much for your help.

Dr. Hanghui Chen
Department of Physics
New York University Shanghai and New York University

Hi,

If you set

delta_approx=.true.
a2f=.true.

you should get the same results as for

eliashberg=.true.

If you want to understand the difference between the two methods used to evaluate the Eliashberg spectral function with or without the delta_approx you can take a look at the following discussion:

viewtopic.php?f=3&t=741&p=2441&hilit=delta_approx#p2441

Another thing to keep in my is that eliashberg=.true. only works with uniform meshes why a2f=.true. can work with both uniform and random meshes.

The code can calculate a2fij(ik,iq,ibnd,jbnd,iwph) if you call the subroutine evaluate_a2fij found in superconductivity_aniso.f90


Best,
Roxana

Dear Roxana,
Thank you very much indeed.
I am sorry that I have been dwelling on this issue. I know the subroutine to calculate a2fij(ik,iq,ibnd,jbnd,iwph). It is a power subroutine, but it contains too much information (k- and q- dependence). Therefore I decided to implement band-resolved "a2f" calculation myself and also follow exactly the same format of "a2f" to output band-resolved "a2f_ij".
Anyway, thank you very much for your help.

Dr. Hanghui Chen
Department of Physics
New York University Shanghai and New York University