Page 1 of 1

Cumulant expansion

Posted: Wed Jun 14, 2023 8:29 pm
by mushroom
Dear all,

I have been trying to use the cumulant expansion for some time now, and have not been able to obtain any results so far:

1/ The diamond example in EPW/examples/diamond/epw provided with the code does not seem to work. If I use it, the spectrum from the cumulant expansion is just zero. Changing meshes, orbitals (bnd_cum), all possible things, introduces either nothing new or a very few finite numbers at specific k-points but also lots of Infinities and NaNs.

2/ The input in test-suite/epw_base/epw9.in from which I adapted some input for my own project does not seem to work either.

3/ At the EPW School 2023 it was mentioned that one can find something in the tutorials of the EPW School 2018, but I was not able to find anything (neither in other schools).

4/ In my own project I was eventually able to get finite results, but then there are negative spectral regions. Already the quasi-particle spectrum calculated before the cumulant expansion contains negative regions. At the moment I am not sure if Eq. (53) in http://dx.doi.org/10.1103/PhysRevB.94.035103 might be indefinite (at first glance it might become negative), and when it becomes negative, if the whole theory is not applicable at all. As already the quasi-particle spectral input for the cumulant expansion may become negative, the problem should not be related to the negative spectral regions mentioned in https://doi.org/10.1103/PhysRevB.105.224304.

5/ Changing the smearing from 'fermi-dirac' to 'methfessel-paxton' in scf calculations did not help.

In any case, the diamond example should work, and I would kindly like to ask for an updated input specifically and only for the cumulant expansion for diamond. Thanks a lot in advance for your kind consideration!