Dear Mab!
Ingredients to obtain the Allen-Dynes Tc (electron-phonon coupling constant \lambda and the logarithmic frequency w_log) are both calculated from the Elishberg spectral function. You could compare the Eliashberg spectral function obtained using the two choices of fsthick, because this function is obtained by summing over the states inside the fsthick window. If this window is too small, most of the contributions are not collected and the result might be wrong.
Best,
Nina.
Unphysically Large Electron-Phonon Coupling Constant in EPW
Moderator: stiwari
-
Nina Girotto
- Posts: 13
- Joined: Wed Mar 12, 2025 9:39 am
- Affiliation: The Catholic University of Louvain
Re: Unphysically Large Electron-Phonon Coupling Constant in EPW
Hi Mab,
From your input1–3 and output1–3 files, it seems that the results do not change with the increasing number of Wannier bands. This makes sense, since only one band crosses the Fermi level as you mentioned.
What surprises me more is the large variation in Tc with the fsthick window. To better understand this, could you share your input and output files? Without them, it’s difficult to pinpoint what might be causing the issue.
When you varied fsthick, did you keep the rest of the input parameters unchanged? What specific modifications did you make? Also, if you varied both fsthick and degaussw, can you please try fixing degaussw and testing only with fsthick first?
Regards,
Shashi
From your input1–3 and output1–3 files, it seems that the results do not change with the increasing number of Wannier bands. This makes sense, since only one band crosses the Fermi level as you mentioned.
What surprises me more is the large variation in Tc with the fsthick window. To better understand this, could you share your input and output files? Without them, it’s difficult to pinpoint what might be causing the issue.
When you varied fsthick, did you keep the rest of the input parameters unchanged? What specific modifications did you make? Also, if you varied both fsthick and degaussw, can you please try fixing degaussw and testing only with fsthick first?
Regards,
Shashi