search

irreducible-representations / irrep

GNU General Public License v3.0
61 stars 31 forks source link

"some eigenvalues are not unitary" Error #69

Open sumannandi121998 opened 9 months ago

sumannandi121998 commented 9 months ago

I want to identify the symmetry operations that may be protecting the crossings. I am using VASP. When I run the command "irrep -isymsep=6 -plotbands -code=vasp -spinor -Ecut=350 -IBstart=40 -IBend=41" it gives following error. Can you help me how to solve this issue? Screenshot 2024-01-29 120610

MIraola commented 9 months ago

Dear Sumannandi,

Assuming that the DFT calculation is right, I think that there might be 2 reasons for this error:

1) The bands 40 and 41 might be last bands in the DFT calculation. The convergence of the last bands is often poor. Please, check NBANDS in your INCAR and increase it if necessary.

2) The bands 40 and 41 are connected to higher (say 42, 43, ...) or lower (39, 38, ...) bands in some k-points. To calculate the eigenvalues, we need to be careful to avoid breaking a reducible representation. Let's say that bands 39,40,41 and 42 transform as an irrep of dimension 4 at some k-point, then we should take -IBstart=38 -IBend=41.

sumannandi121998 commented 9 months ago

Hi, As per your suggestion, I have changed the IBstart and IBend and it works fine without -isymsep tag. But when -isymsep tag is on it gives following error. Can you please help me to resolve this issue? Screenshot 2024-01-29 135752

stepan-tsirkin commented 9 months ago

@sumannandi121998 , isymsep works properly only with norm-conserving pseudopotentials, which are not available for VASP.

Why do you need it? If there is a crossing on a rotational/screw axis, just plot the bands color-coded by the eigenvalues of this rotation, and you will see that two lines with different eigenvalues are crossing.

Does that work for you?