EPW Forum

Dear all,

I want to calculate temperature dependent el. linewidths at certain k-points. To efficiently do so, it would be useful to just vary the "Fermi-temperature" and keep the matrix-elements in the memory. It seemed to be possible with eptemp(:) being an array of temperatures to be looped over. However this feature was uncommented in epw_readin.f90 in the latest release. What was the specific reason to get rid of that? Is there a smart workaround?

Any input is appreciated,

thanks Nicki
CAMd, Lyngby

Dear Nicki,

The main reason was that the way it was implemented could make the memory explode for some system (with a lot of temperatures).
In addition it was an extra outer loop in a couple routine that add a bit of mess.

I found (at least for my need) that you usually only want a few temperature and it was not an issue to restart.

It should be rather easy for you to re-add this feature by adding an outer loop in selfen_elec.f90, selfen_phon.f90, spectral_func.f90 etc.

Best,

Samuel