DOS, Hybrids and DEN calculations

Total energy, geometry optimization, DFT+U, spin....

Moderator: bguster

Posts: 9
Joined: Wed Nov 07, 2012 11:40 am

DOS, Hybrids and DEN calculations

Post by lorenzo.sponza » Fri Mar 26, 2021 6:33 pm

I'm reporting a strange thing that I don't really explain. I wonder if it is a bug.

I'm calculating the DOS of Silicon using LDA and HSE (ixc1=1 ; ixc2=-428). I made two kind of calculations.

Calculations 1: DOS from DEN (DfD). In the DfD set of calculations, firs I launch a self-consistent run, then I launch a DOS run which reads the DEN file produced by the previous calculation. This is what one is supposed to do.

Calculations 2: DOS on the fly (Dof). In the Dof set of calculations, I launch a self-consistent run with the additional flag prtden 2 and nband 8. This is not what one is supposed to do, but still I have a result.

The thing I can't explain is that whereas the LDA results are identical in the the calculation schemes (including the Fermi energy), the hybrids differ both in the Fermi energy and the DOS. In particular, the conduction states are much higher in the Dof calculation.

In attachment you can find the input files of the two calculations and a screenshot of the comparison of the hybrid DOS (0 aligned at the top valence).

I can't really explain why this happens. Can you ?
Which shall I trust ?
Is this a bug, or there is something fundamental I am missing ?
(1.83 KiB) Downloaded 145 times
(1.73 KiB) Downloaded 139 times