[QE-users] symmetry traces in sym_band.f90

[Hongyi Zhao]

On Tue, Mar 8, 2022 at 6:08 PM Gerson J. Ferreira
<gersonjferreira at ufu.br> wrote:
>
> Yes, you are right. The first examples I gave above were 2x2 single group matrices, but for the numerical case I only had the spinful double group ready to show, but I guess it was clear enough for the discussion.
>
> Yes, I know what a non-symmorphic symmetry is. My python code calculates the matrix representations of generic symmetry operations {S,tau}, where S is the point group rotation and tau is a possible fractional translation. But it is still under development, we are testing it for different materials to make sure it works, and that's why it is not public on github yet. We'll make it public once we have a first version and a draft of the paper ready.
>
> The problem with my python code is that it is too simple. It does the calculation using QE plane waves and does not carry the PAW or US pseudo-potential corrections. For PAW pseudos we the matrix elements must be corrected by the partial-waves, pseudo-partial-waves and projectors, while for US pseudos we have to account for the overlap matrix S. I'm not account for these, and I have the impression that for symmetry operators these corrections are not necessary, since the plane waves should obey the symmetry constraints with and without the corrections. But I'm not 100% sure yet.
>
> In any case, the plan is to drop this python code and implement the same thing directly in QE's sym_band.f90 (or a nearly equal new routine). That's why I'm currently studying it. If I cannot make it work with sym_band.f90, then I'll use my current python code.
>
> For this purpose, I need to understand where exactly lies the limitation of sym_band.f90 regarding non-symmorphic symmetries. I'm studying the code, but if some could give me some hints, I would appreciate it!

I think you should start by studying the following code:

https://github.com/dalcorso/thermo_pw/blob/master/src/sym_band_sub.f90

My point of view is based on the fact that this code is an updated
version of QE's sym_band.f90 and was developed specifically to solve
non-symmorphic symmetries, as indicated by the following comment at
the beginning of it:

!
! Copyright (C) 2006-2007 Quantum ESPRESSO group
! Copyright (C) 2016-2017 Andrea Dal Corso
!    (added symmetry analysis at zone border point in nonsymmorphic space
!     groups)


Yours,
Hongyi