KPOINT_IN_FULL_GRID (Absorption Spectra Calculation for hBN)

[idan_haritan]

Dear VASP Team,
I have encountered a bug while calculating the absorption spectra for hexagonal BN.
The system is based on a 6x6x1 supercell (POSCAR attached), on top of which I am calculating a k-point mesh grid.

The issue:
• I am using the HSE06 functional.
• I am performing a TDDFT calculation using the WAVECAR and WAVEDER files from a previous VASP job (that calculates the unoccupied states of the system) - INCAR attached.
• The previous job, which calculates the unoccupied states, runs successfully.
• The calculation fails when I use a 2x2x1 Monkhorst-Pack grid in the KPOINTS file.
• However, when I run the same calculation with a 3x3x1 Monkhorst-Pack grid, the error does not occur.
The calculation fails with the following error message:

vasp_bug.png

VASP Version and Environment:
• I am using VASP version 6.4.3 on Rocky Linux.
• My setup includes a Xeon 6338 node with 128 threads and 512 GB RAM
• For this specific job I am using 23 threads and a RAM of 224280M.

Files for Reproducibility:
Please find attached a minimal set of input files and relevant outputs.

[merzuk.kaltak]

Dear idhan_haritan,

I was able to reproduce the error. Also, the error message is misguiding and slightly confusing,
because the actual error/warning is printed to stdout before the Bug message reading

Code: Select all

|     Your generating k-point grid is not commensurate with the symmetry      |
|     of the lattice. This can cause slow convergence with respect to         |
|     k-points for HF-type calculations.                                      |
|     Suggested SOLUTIONS:                                                    |
|      ) If not already the case, use automatic k-point generation,           |
|      ) and/or shift your grid to Gamma (G) (e.g. required for               |
|        hexagonal, fcc, and fcc-orthorhombic lattices).                      |
|      ) and/or set KBLOWUP = .FALSE.                                         |
|                                                                             |

As proposed, I have redone the calculation successfully with a gamma centred k-point mesh.
The warning as well as the Bug-message are then gone.
So, please use a gamma centred k-point mesh.

[alexey.tal]

Dear idhan_haritan,

I would like to add one comment to Merzuk's answer.

If you want to use a shifted k-mesh, you can do it by disabling all the symmetries, i.e., ISYM=-1 in both ground state and BSE/TDHF calculations.