Inconsistent k-point grid description between DFT codes
The handling of k-points is inconsistent between different DFT codes. I will describe two examples here:
- calculation 'https://nomad-lab.eu/prod/rae/gui/entry/id/_ijarO1hQCqsTpOyb62WXw/tpEO3vMCaKGl6CKRw0lxbbponc-j', code: VASP
In the input file 'vasprun.xml', under ,a k-point grid with the divisions 13x13x13 is reported. In the Archive, under 'section_method/0/k_mesh_points' there are only 1099 k-points available for the calculation. My suspicion is, that in the Archive the irreducible k-points are being shown for VASP, while in the input file reducible ones are shown.
- calculation 'https://nomad-lab.eu/prod/rae/gui/entry/id/MF3kN7zlT-q942wC1G33Jw/07mqk3ovUAz43fXJl-2p-KkfFQUJ', code = FHI-aims
In the file 'output.out', under item 'k_points' a value of 13x13x13 is reported. The same value is in the Archive entry under 'section_root/0/section_method/0/x_fhi_aims_controlInOut_k_grid'. These are in my opinion (based on the documentation of FHI-aims), the reducible k-points, which were used for the ground-state calculation.
However, inspecting the documentation of FHI-aims, one finds, that the DOS is computed using a more dense grid. The k-point grid used for the DOS is constructed by multiplying values of 'k_points' by the corresponding factors 'dos_kgrid_factors' found in 'output.out', which are in this case, 3x3x3. This means, that the DOS is computed on a k-point grid of 39x39x39. I was unable to find this in the Archive, if it is indeed not avaiable, this would be important to make the distinction, that the DOS is computed on a different k-grid than the groud-state, in the best case scenario, one would include both of the k-point grids in the Archive.
In short: The Archive user does not report, whether the reducible or the irreducible k-points are parsed. For two DFT codes one or the other is reported. This is quite misleading, unless the user digs through the documentation of each code and through the input files of the calculations.