Isotropic Eliashberg calculation not converging

[mdogan]

Hello,

I'm new to EPW, and want to compute the isotropic Eliashberg function for a solid hydrogen phase. I'm following the steps from Exercise 1 of the "Eliashberg function and superconducting properties" tutorial (Fri.4 at https://docs.epw-code.org/doc/School2018.html - PDF attached). I run an epw calculation based on the input file on the page 3 of the PDF, but I get the following error:

Code: Select all

     Error in routine sum_eliashberg_iso_iaxis (1):
     increase nsiter or reduce conv_thr_iaxis

Basically the isotropic Eliashberg calculation is not converging. What can I do to improve the convergence of this calculation? All the relevant files (hopefully) can be found at
[https://www.dropbox.com/sh/163gj9ampwzp ... RDlja?dl=0]. Any suggestions will be greatly appreciated.

Best,
Mehmet Dogan

[hpaudya1]

Hi Mehmet,

I see the convergence has not achieved for the temperature T = 250 K, which is much larger than Allen-Dynes Tc (~57 K). You can also see a WARNING about this.

WARNING WARNING WARNING

The code will crash for tempsmax much larger than Allen-Dynes Tc

This is because the superconducting gap vanishes (~zero) at the superconducting transition temperature (Tc).

On the other hand, the convergence is achieved at 50 K, which looks fine. You can try few more temperatures below 60 K and plot gap vs Tc (as shown in page 8 of Fri.4.Margine.pdf) and see what would be your Tc.

Finally, I would like to suggest you to use the recent official version of EPW v5.3 (https://gitlab.com/QEF/q-e/-/releases) or the recent development version of EPW (https://gitlab.com/QEF/q-e).

I hope this helps.

Best,
Hari Paudyal

[mdogan]

Dear Hari,

Thank you for the explanation! That totally makes sense. Is there a way to run the code up to the point where it finds the estimate of the Allen–Dynes Tc (in this case ~57 K) so that I can set the temperature grid accordingly for a second epw calculation?

Best,
Mehmet

[hpaudya1]

Hi Mehmet,

Yes, there is a way to run the code up to the point where it estimates the Allen–Dynes Tc by de-activating the following flags;

! liso = .true.
! limag = .true.
! lpade = .true.
! lacon = .true.

If you want to run iso/aniso calculation later, you can re-start by reading saved files from 'prefix.ephmat' directory 'selecq.fmt', 'and 'crystal.fmt' with the following set-ups;

ep_coupling = .false.
elph = .false.

ephwrite = .false.

liso = .true.
limag = .true.
lpade = .true.
lacon = .true.

I hope this helps. Happy EPWing

Best,
Hari Paudyal