ASR violation in AlAs fermi smearing

Running Abinit Response calculations
1

Hello

I am having trouble running phonon calculations with metallic compounds and I went back to the tutorial and tried running AlAs as a metal (occopt 3) and ended up with huge differences in the DFPT calculations.

Am I missing something obvious? I am using the standard pseudopotentials from the abinit tutorial and all I changed was the occupation scheme between the two runs.

Thank You
AlAsOccopt3.in (725 Bytes)
AlAsOccopt1.out (55.8 KB)
AlAsOccopt3.out (67.3 KB)
AlAsOccopt1.in (727 Bytes)

2

Hello, I have since solved my issue.

The problem lies with the default value of nbdbuf. It automatically assumes that there are 2 or more unoccupied bands in the ground state but for small systems this is not true. For AlAs, the code was freezing an occupied band when doing RF calculations thus giving incorrect results.

The solution is to set nbdbuf to 0.

3

Thank you Konrad for reporting your solution!
Best wishes,
Eric