k-point mesh for HF type calculations and the IBZKPT_HF subroutine
Posted: Sun Mar 06, 2011 1:05 pm
Hello,
I'm starting to do hybrid calculations by using the HSE06 functional implemented in VASP 5.2.8 07. My unit cell is a 40 atom structure BCC lattice, which I define in the POSCAR file (just as in the normal GGA-PBE runs to generate the respective WAVECAR file).
Accordingly to your manual on how to do less expensive HF calculations one can use a manual k-point settings in the BZ: such as the BCC or FCC -like lattice defined in the KPOINT file. In order to perform a DOS calculation, and using the tetrahedron method with Blöchl corrections smearing, I need at least 4 k-points in the BZ. So, in order to follow the given suggestion I need to use a fcc-like lattice in the BZ (because the bcc-like configuration results only in 2 k-points).
When doing so, I (obviously) get the following error message
VERY BAD NEWS! internal error in subroutine IBZKPT:
Reciprocal lattice and k-lattice belong to different class of lattices. Often results are still useful... 4
Should I be worried about this message - or may I ignore it as just a warning about the different lattice settings - bcc for the reciprocal space and fcc for the BZ? Using this procedure it really speeds up the calculation time...
When performing the same calculations on a conventional cell, of the same material, using a 80 atom cubic structure, the warning disappears. I think its because the fcc lattice belongs to the cubic structure and VASP hence doesn't complain?
Another thing that I've noticed - when using an automatic Monkhorst-Pack or Gamma centered mesh the k-points generated in the 1st BZ by a GGA-PBE run is different than with a hybrid run returned by the IBZKPT_HF subroutine (generally it gives more k-points for the HSE06 run). Can anyone explain me why?
Thank-you
EL !alien
I'm starting to do hybrid calculations by using the HSE06 functional implemented in VASP 5.2.8 07. My unit cell is a 40 atom structure BCC lattice, which I define in the POSCAR file (just as in the normal GGA-PBE runs to generate the respective WAVECAR file).
Accordingly to your manual on how to do less expensive HF calculations one can use a manual k-point settings in the BZ: such as the BCC or FCC -like lattice defined in the KPOINT file. In order to perform a DOS calculation, and using the tetrahedron method with Blöchl corrections smearing, I need at least 4 k-points in the BZ. So, in order to follow the given suggestion I need to use a fcc-like lattice in the BZ (because the bcc-like configuration results only in 2 k-points).
When doing so, I (obviously) get the following error message
VERY BAD NEWS! internal error in subroutine IBZKPT:
Reciprocal lattice and k-lattice belong to different class of lattices. Often results are still useful... 4
Should I be worried about this message - or may I ignore it as just a warning about the different lattice settings - bcc for the reciprocal space and fcc for the BZ? Using this procedure it really speeds up the calculation time...
When performing the same calculations on a conventional cell, of the same material, using a 80 atom cubic structure, the warning disappears. I think its because the fcc lattice belongs to the cubic structure and VASP hence doesn't complain?
Another thing that I've noticed - when using an automatic Monkhorst-Pack or Gamma centered mesh the k-points generated in the 1st BZ by a GGA-PBE run is different than with a hybrid run returned by the IBZKPT_HF subroutine (generally it gives more k-points for the HSE06 run). Can anyone explain me why?
Thank-you
EL !alien