bug when running TDHF with explicit k-point

[awang95]

Hi,

I've been trying to run a TDHF calculation with an explicit k-point (0.333333 0.333333 0.000000 for a hexagonal cell). The initial static calculation and the one after with LOPTICS = True both worked, but in the last step with ALGO = TDHF I got the following error:

Code: Select all

internal ERROR in RE_READ_KPOINTS: the total number of non zero k-points in the
  full Brillouine zone has changed           2           6
 trying to continue but this does not look good at all
 responsefunction array rank=   50700
 LDA part: xc-table for Pade appr. of Perdew
 -----------------------------------------------------------------------------
|                     _     ____    _    _    _____     _                     |
|                    | |   |  _ \  | |  | |  / ____|   | |                    |
|                    | |   | |_) | | |  | | | |  __    | |                    |
|                    |_|   |  _ <  | |  | | | | |_ |   |_|                    |
|                     _    | |_) | | |__| | | |__| |    _                     |
|                    (_)   |____/   \____/   \_____|   (_)                    |
|                                                                             |
|     internal error in: screened_2e.F  at line: 102                          |
|                                                                             |
|     internal error in SETUP_IRZ_MAP: WDES%NKPTS needs to be identical       |
|     to KPOINTS_FULL_ORIG%NKPTS                                              |
 -----------------------------------------------------------------------------
|                     _     ____    _    _    _____     _                     |
|                    | |   |  _ \  | |  | |  / ____|   | |                    |
|                    | |   | |_) | | |  | | | |  __    | |                    |
|                    |_|   |  _ <  | |  | | | | |_ |   |_|                    |
|                     _    | |_) | | |__| | | |__| |    _                     |
|                    (_)   |____/   \____/   \_____|   (_)                    |
|                                                                             |
|     internal error in: screened_2e.F  at line: 102                          |
|                                                                             |
|     internal error in SETUP_IRZ_MAP: WDES%NKPTS needs to be identical       |
|     to KPOINTS_FULL_ORIG%NKPTS                                              |
 -----------------------------------------------------------------------------
|                     _     ____    _    _    _____     _                     |
|                    | |   |  _ \  | |  | |  / ____|   | |                    |
|                    | |   | |_) | | |  | | | |  __    | |                    |
|                    |_|   |  _ <  | |  | | | | |_ |   |_|                    |
|                     _    | |_) | | |__| | | |__| |    _                     |
|                    (_)   |____/   \____/   \_____|   (_)                    |
|                                                                             |
|     internal error in: screened_2e.F  at line: 102                          |
|                                                                             |
|     internal error in SETUP_IRZ_MAP: WDES%NKPTS needs to be identical       |
|     to KPOINTS_FULL_ORIG%NKPTS                                              |
|      usually this requires to regenerate the k-point grid without           |
|     symmetry                                                                |
|      WDES%NKPTS, KPOINTS_FULL_ORIG%NKPTS= 2 6                               |
|                                                                             |
|     If you are not a developer, you should not encounter this problem.      |
|     Please submit a bug report.                                             |
|                                                                             |
 -----------------------------------------------------------------------------

Would you mind helping me with this?

Attached are my files.

Thanks!

[alexey.tal]

Dear awang95,

BSE calculations in VASP require all k-points in the Brillouin zone and this problem occurs when we try to restore the full BZ from the irreducible wedge using the symmetry of the structure.
If you would like to use explicit k-points you should switch off the symmetry, and provide the full list of k-points.

[awang95]

ok thanks that works!

when you mean "restore the full BZ from the irreducible wedge" do you mean that, even if we only provide the gamma point in the KPOINTS file, it still tries to estimate the transitions from other areas in the BZ?

[alexey.tal]

In VASP, the calculation of the BSE requires wavefunctions to be computed at all k-points of the grid, not just the irreducible ones. The complete grid is restored from the irreducible wedge through symmetry operations. When symmetry is disabled, the irreducible and full grids become equivalent. Notably, at the gamma point, this equivalence remains even when symmetry is employed.