Problematic isotropic Eliashberg function for monolayer MgB2

[CHILLL]

Dear users,

I have encountered problems running the EPW code for monolayer MgB2. Seeing that the gap spectrum does not look healthy at all in my case I started to check all the possible output. The first thing that caught my eye is the isotropic Eliashberg function, which is clearly not correct (https://ibb.co/2YLmwxr). Such an unhealthy behavior of the function at the low frequencies resulted in staggering total coupling of 293. I suppose that if even at this stage the results are so dire then there is no point to try to make any sense out of the anisotropic output.

Electronic structure, including Wannierisation, seems well converged and looks correct. Now, the phonons have a bit of a problem. (https://ibb.co/7K5hn0w)
Around the Gamma point one of the acoustic modes plunged below zero. I tried increasing accuracy of the calculations, but there was not much success. As I recall for monolayer LiC6 there are also slightly negative phonon values and it does not harm the results.

So my question are:
exactly how harmful negative phonons can be?
How does the code treat them?
Is there anything else I should check except for phonons?

If you need some of my input/output I will send it at once.
Thank you!
Best regards, Mikhail

[CHILLL]

I have a small update on the problem. To check the correctness of my input I calculated the isotropic Eliasherg function in Quantum Espresso (matdyn.x), and it turned out healthy and quite accurate. (https://ibb.co/cLGHFRM)
Maybe this will help to clarify things somehow.
As far as I know in EPW the isotropic Eliashberg function is calculated automatically in the beginning of the main run, so there is hardly a place to make a mistake.

Thank you!

[roxana]

EPW doesn't have a special treatment for 2D materials and the behavior of the e-ph coupling may be problematic. Please search the forum for the discussions regarding issues in 2D systems.

Roxana

[CHILLL]

Dear users,

Eventually, I switched from LDA pseudopotentials to PBE and did the calculations again.
Now everything works like a charm. Unfortunately, I cant really say what conclusion can be drawn from this story.

Best, Mikhail