[QE-users] Problem with tetrahedra method: from tetra_init : error # 26, cannot remap grid on k-point list

[Pietro Davide Delugas]

Dear Pacome

6.3 was released in July 2018 and in fact the new variable to dos.x was 
introduced only later in august

You can download 6.4 and use that one.
Pietro


On 03/17/2019 07:07 AM, Pacome NGUIMEYA wrote:
> Dear Pietro,
>
> I ran the scf calculation follows by the nscf calculation with
> occupation='smearing', smearing and degauss (with QE-6.3.0).
> I added  bz_sum = 'tetrahedra_opt' in the dos.x input file and I ran
> it but, it crashed and I got the following error message
>
>       task #         0
>       from dos : error #        19
>       reading dos namelist
>
> Apparently, the tetrahedron method is used if
>
>           - "the input data file has been produced by pw.x using the
> option
>              occupations='tetrahedra', AND"
>           - "a value for degauss is not given as input to namelist &dos"
>
> read here https://gitlab.com/QEF/q-e/blob/qe-6.3/PP/Doc/INPUT_DOS.txt
>
> But pw.x (nscf) does not work in my case
>
> Further research on the internet led me to this website
> https://eamonnmurray.gitlab.io/modelling_materials/lab04/ where I could
> read that
>   
> "Essentially this corresponds to doing a three dimensional linear
> interpolation from a regular grid of values."..."It’s important to note
> that in a real measurement of the density of states of a system the
> there is an implicit broadening that comes from
>
> 1. Electron-phonon coupling: the states are not simply at a fixed
> energy, but will have some distribution as the atoms vibrate.
>
> 2. Any measurement probe will have a finite energy width associated
> with it, which will limit how finely it can resolve density of states
> features.
>
> So while tetrahedron may seem the more accurate approach, you shouldn’t
> necessarily think of it as a more correct representation of a real
> system."
>
> I think I am just going to use the common way, that is, running the
> nscf calculation with smearing and degaus. Then, run this code
>
> &DOS
> 		prefix = '...',
> 		outdir = '...',
> 		fildos = '....dos.dat',
>   /
> and plot the DOS.
>
> Regards - Pacome