Trouble with hybrid calculation in ISMEAR=-2

[Kousei]

Hello Everyone,

I am performing a verification related to the following topic.
https://www.vasp.at/forum/viewtopic.php ... RWE#p21265
Specifically, we are using VASP 6.3.2 to perform hybrid calculations for BaCl2, and during the ISIF=-2 relaxation, the fixation of the occupancy rate is off.
I am aware that the version upgrade has fixed the bug related to ISMEAR=-2 that was reported in VASP 6.1.0. Is there something wrong with my configuration?
I would like to hear your opinion.

Best regards.

[Kousei]

I forgot to attach the file. Here it is.

[henrique_miranda]

I did not quite understand your question.
What do you mean by "the fixation of the occupancy rate is off"?
What would you expect to obtain and what did you obtain?
Do you mean that even after implementing the bugfix in the post you linked the occupations are still not fixed during the ionic relaxation loop?

[Kousei]

It's correct.

[henrique_miranda]

I changed the name of the topic. Before you wrote ISIF=-2 but I guess you mean ISMEAR=-2.

I managed to reproduce the issue.
It seems that the option to keep the occupancies fixed is broken in vasp 6.3.2.
The easiest way to fix it is by modifying the following lines in the DENSTA routine in dos.F

Code: Select all

      [...]
      ELSE IF ((KPOINTS%ISMEAR==-5).OR.(KPOINTS%ISMEAR==-15)) THEN
        CALL DENTET[...]
      ELSE
        CALL DENMP[....]
      ENDIF

to

Code: Select all

      [...]
      ELSE IF ((KPOINTS%ISMEAR==-5).OR.(KPOINTS%ISMEAR==-15)) THEN
        CALL DENTET[...]
      ELSE IF (KPOINTS%ISMEAR==-2) THEN
        ! do nothing here, otherwise, the occupations are recomputed
      ELSE
        CALL DENMP[....]
      ENDIF

Let me know if this solves the problem for you.
We will perform further tests and include this fix in a future release of VASP.
Thanks for reporting it

[Kousei]

The suggested modifications have allowed me to successfully perform structural relaxation while keeping the occupations fixed.

Thank you very much!