BSE on hybrid k mesh

[xiaoming_wang]

Hello,

I'm trying to calculate BSE on a hybrid k mesh as in PRB 78, 085103 (2008), i.e., coarser at outer BZ but denser at inner BZ. However, I got error like

Code: Select all

 -----------------------------------------------------------------------------
|                     _     ____    _    _    _____     _                     |
|                    | |   |  _ \  | |  | |  / ____|   | |                    |
|                    | |   | |_) | | |  | | | |  __    | |                    |
|                    |_|   |  _ <  | |  | | | | |_ |   |_|                    |
|                     _    | |_) | | |__| | | |__| |    _                     |
|                    (_)   |____/   \____/   \_____|   (_)                    |
|                                                                             |
|     internal error in: mkpoints_full.F  at line: 1271                       |
|                                                                             |
|     internal error in KPOINT_IN_FULL_GRID: can not find 0 -0.25 0           |
|                                                                             |
 -----------------------------------------------------------------------------

and my KPOINT is

Code: Select all

Automatically generated mesh
       33
Reciprocal lattice
   -0.37500000000000    -0.37500000000000     0.00000000000000     0.06250000000000
   -0.37500000000000    -0.12500000000000     0.00000000000000     0.06250000000000
   -0.37500000000000     0.12500000000000     0.00000000000000     0.06250000000000
   -0.37500000000000     0.37500000000000     0.00000000000000     0.06250000000000
   -0.12500000000000    -0.37500000000000     0.00000000000000     0.06250000000000
   -0.12500000000000     0.37500000000000     0.00000000000000     0.06250000000000
    0.12500000000000    -0.37500000000000     0.00000000000000     0.06250000000000
    0.12500000000000     0.37500000000000     0.00000000000000     0.06250000000000
    0.37500000000000    -0.37500000000000     0.00000000000000     0.06250000000000
    0.37500000000000    -0.12500000000000     0.00000000000000     0.06250000000000
    0.37500000000000     0.12500000000000     0.00000000000000     0.06250000000000
    0.37500000000000     0.37500000000000     0.00000000000000     0.06250000000000
   -0.18750000000000    -0.18750000000000     0.00000000000000     0.01562500000000
   -0.18750000000000    -0.06250000000000     0.00000000000000     0.01562500000000
   -0.18750000000000     0.06250000000000     0.00000000000000     0.01562500000000
   -0.18750000000000     0.18750000000000     0.00000000000000     0.01562500000000
   -0.06250000000000    -0.18750000000000     0.00000000000000     0.01562500000000
   -0.06250000000000     0.18750000000000     0.00000000000000     0.01562500000000
    0.06250000000000    -0.18750000000000     0.00000000000000     0.01562500000000
    0.06250000000000     0.18750000000000     0.00000000000000     0.01562500000000
    0.18750000000000    -0.18750000000000     0.00000000000000     0.01562500000000
    0.18750000000000    -0.06250000000000     0.00000000000000     0.01562500000000
    0.18750000000000     0.06250000000000     0.00000000000000     0.01562500000000
    0.18750000000000     0.18750000000000     0.00000000000000     0.01562500000000
   -0.08333333333333    -0.08333333333333     0.00000000000000     0.00694444444444
   -0.08333333333333     0.00000000000000     0.00000000000000     0.00694444444444
   -0.08333333333333     0.08333333333333     0.00000000000000     0.00694444444444
    0.00000000000000    -0.08333333333333     0.00000000000000     0.00694444444444
    0.00000000000000     0.00000000000000     0.00000000000000     0.00694444444444
    0.00000000000000     0.08333333333333     0.00000000000000     0.00694444444444
    0.08333333333333    -0.08333333333333     0.00000000000000     0.00694444444444
    0.08333333333333     0.00000000000000     0.00000000000000     0.00694444444444
    0.08333333333333     0.08333333333333     0.00000000000000     0.00694444444444

The DFT part is fine but the (model)BSE part doesn't work. Any remedy?

Best,
Xiaoming

[ferenc_karsai]

Have you tried the tutorial on model BSE's first?
Here it is:
https://www.vasp.at/wiki/index.php/Impr ... c_function

[xiaoming_wang]

Yes, I'm using the model BSE. I also want to use the hybrid k mesh to accelerate the convergence.

[ferenc_karsai]

Can you please upload your calculations according to the forum guidelines?

[xiaoming_wang]

Thanks. I uploaded a small example to show the bug.

[xiaoming_wang]

The previous attachment has some problem. I attached a new one here.

[ferenc_karsai]

Unfortunately non-uniform k-meshes are not supported in the BSE routines in VASP.

If you have a look at the tutorials:
wiki/index.php/Improving_the_dielectric_function

There the dielectric function at different k-meshes with shifted origins were calculated.
This gives also a huge improvement and is potentially equally (or better) performing.

[xiaoming_wang]

Thanks. It's good to know that. Btw, in the bse.F file there are many places where the function KPOINT_IN_FULL_GRID is called. If all the calculations (DFT and model BSE) are performed with ISYM=-1, do I really still need that function?

[ferenc_karsai]

You definitely need the full Brillouin zone to make the kk' pairs for the BSE Hamiltonian.

[xiaoming_wang]

Thanks. I finally turn to UNIFORM shifted grid. However, I encountered similar bug even for the UNIFORM grid. Please see the attached files.

[ferenc_karsai]

Don't use the command KPOINT_BSE in your INCAR file, then it should work.