Page 1 of 1

band index and k points of anisotropic Eliashberg output

Posted: Tue Feb 21, 2017 3:38 am
by jiachen
Dear All,

I am studying the example of MgB2 anisotropic Eliashberg equation. When the calculation was done, I had a file of quasi-particle energy, renormalization and gap function as a function of imaginary frequencies in the file: MgB2.imag_aniso_XX. For the example given, I counted 146 rows for each frequency. From the output file, I can see, the fine k grid is 20*20*20, 8000 points, with 484 irreducible k-point on the uniform grid. There are 2 bands and 120 irreducible k-point on the uniform grid within the Fermi shell. So, it seems to me 120*2 should be the reasonable number of points at each frequency. I am confused why I got 146 in the output file. I guess the general question is how band index and k points are arranged in output file: prefix.imag_aniso_XX . Thank you!

Re: band index and k points of anisotropic Eliashberg output

Posted: Tue Feb 21, 2017 3:19 pm
by roxana
Hi,

The prefilx.imag_aniso_XX file contains the following information

WRITE(iufilgap,'(5a20)') '# w [eV]', 'Enk-Ef [eV]', 'Znorm(w) [eV]', 'Delta(w) [eV]', 'NZnorm(w) [eV]'
DO iw = 1, nsiw(itemp) ! loop over omega
DO ik = 1, nkfs
DO ibnd = 1, nbndfs
IF ( abs( ekfs(ibnd,ik) - ef0 ) .lt. fsthick ) THEN
WRITE(iufilgap,'(5ES20.10)') wsi(iw), ekfs(ibnd,ik)-ef0, AZnormi(ibnd,ik,iw), ADeltai(ibnd,ik,iw), NAZnormi(ibnd,ik,iw)
ENDIF
ENDDO ! ibnd
ENDDO ! ik
ENDDO ! iw

The first loop is over frequency, the second over the irreducible k-points within the Fermi shell, the third of the bands with energy within the Fermi shell.

Best,
Roxana

Re: band index and k points of anisotropic Eliashberg output

Posted: Tue Feb 21, 2017 9:37 pm
by jiachen
Dear Roxana,

Thank you very much for the explanation. But, I still don't fully understand this, according to the output
"Nr. of irreducible k-points on the uniform grid: 484"
"Nr irreducible k-points within the Fermi shell = 120 out of 484"
What happens here is that some k points within the "Fermi shell" do NOT satisfy "abs( ekfs(ibnd,ik) - ef0 ) .lt. fsthick"
If fsthick is not the criterion for "Fermi shell", Can you explain to me what makes the Fermi shell, and how the number of k points is reduced from 484 to 120? I really appreciate your patience!

Cheers

Re: band index and k points of anisotropic Eliashberg output

Posted: Tue Feb 21, 2017 9:56 pm
by roxana
Hi,

fsthick is the criterion for defining the Fermi shell and as a result the nr of irreducible points is reduced from 484 to 120 ("nkfs=120")

All the k-points within the Fermi shell have at least one band that satisfies the condition "abs( ekfs(ibnd,ik) - ef0 ) .lt. fsthick".
This means that for each k-point within the Fermi shell not all "nbndfs" are printed to file but only those bands that satisfy the condition "abs( ekfs(ibnd,ik) - ef0 ) .lt. fsthick".
So in your case if "nbndfs=2", some of the nkfs points will have 1 band and others 2 bands within the Fermi shell.

Best,
Roxana

Re: band index and k points of anisotropic Eliashberg output

Posted: Tue Feb 21, 2017 10:46 pm
by jiachen
Dear Roxana,

Thank you very much!