I’m repeating a question from some time ago that didn’t seem to have an answer. I can successfully do a GS calculation with a hybrid like HSE06, with occupied bands, but how does one compute a band structure this way? kptopt 0 and < 0 are forbidden for hybrids, it seems. Thanks–
Hi Joe,
Here is my partial answer. I hope @gmatteo can complement it.
Hybrid functionals require a regular k-point grid (because of the non-local exchange operator). Therefore, doing non-self-consistent calculations for the high-symmetry lines is impossible.
A working strategy to obtain “band structures” would be to calculate the hybrid functional eigenvalues on a \Gamma-center k-point grid and then do an interpolation on the desired lines as a post processing.
abipy has this functionality. I used it once some years back. But I don’t remember the keywords, that’s why @gmatteo would be helpful here.
Best,
Fabien
Oh perfect! Thanks much–
Additional question: I thought I had this figured out, but I can’t seem to the run the HSE06 calculation with empty bands. After the first few cycles of the Fock part I always get the error
-– !ERROR
src_file: m_pawcprj.F90
src_line: 994
mpi_rank: 0
message: |
Bug in pawcprj_lincom: n2 wrong sizes!
Is there a real issue with PAW here? Or am I missing something in the set up?