[Wannier] inconsistency in Hamiltonian elements (_hr.dat) for spinor-valued WF, when adding "wannier_plot_spinor_mode=true" & "wannier_plot_spinor_phase=up/down"

[H. Lee]

Dear Shi:

It is very strange if we look at the subroutine of plot_main in plot.F90;
the call to plot_wannier is after the call to hamiltonian_write_hr and more
importantly, both of them are after Wannierization; all stuffs related to
the plotting of Wannier functions (WFs) are included in the subroutine
of plot_wannier.

So, in general the plotting of WFs should not affect the Wannierization
process and if something affects the Wannierization process, it is natural
for you to have different H(R). I think you need to know which one affects
the Wannierization process (I think the plotting of WFs doesn't).

I am assuming that you performed two W90 runs from the exactly same input
files to W90 such as amn, mmn, etc.

Sincerely,

H. Lee

On Mon, Dec 7, 2020 at 10:36 PM Shima Sharifi <shima.sharifi at tyndall.ie>
wrote:

> Deal all,
> I am new user of wannier90. I have extracted wannier functions from a DFT
> calculation (plane wave basis, PWSCF) including spin orbit coupling (SOC).
> Then I turned on spinors=T and double the num_bands in my .win file. My
> systems have no spin polarization.
>
> I am using quantum Espresso version 6.4.1 and the version of wanneir90
> library code there, is 3.0.0.
> According to the paper for new features of Wannier90 (J. Phys.: Condens.
> Matter 32 (2020) 165902 (25pp)), there are several possibilities for
> plotting MLWFs. I’d like to one plot norm of spinor WF and in another case
> I wanted to include phase information for both spin up and spin down
> components. So for the first case I just turned on these items:
>  wannier_plot = .true.
>  wannier_plot_supercell = 3
>
> and for the second case, I turned on these two variables as well (here for
> spin up):
>  wannier_plot_spinor_mode = up
>  wannier_plot_spinor_phase = .true.
>
> Meanwhile, I needed to print out metric elements, so I added relevant
> variables in my wannier90 input file:
> write_hr  = .ture.
>
> Then, I ran these two calculations separately using wannier90.x. I
> compared my _hr.dat files from those two calculations and I obviously see
> that there are some differences, which I wonder why. (I mean I expected
> that “wannier_plot_spinor_mode” and “wannier_plot_spinor_phase” make some
> changes just in plotting Wfs as discussed in the above paper, however it
> seems that it changes metric elements, as well!)
> In addition, it effects wannierisation process and even final WF centers.
> In first case, it gets converged after 625 iterations, in second one after
> 1256 iterations.
>
> Here’s my .pw2wan file:
> &inputpp
>   outdir     =  './tmp'
>   prefix     =  'r-3m_Bi2Te3'
>   seedname   =  'r-3m_bi2te3_LDA'
>   write_amn  =  .true.
>   write_mmn  =  .true.
>   write_unk  =  .true.
>   reduce_unk =  .true.
>   spin_component =  'none'
>   write_spn  =  .true.
> /
>
> and in my Non-scf calculation, I have turned off the symmetries to include
> the whole BZ.
>
> I’d appreciate if one can guide me how to interpret those different
> _hr.dat outputs.
>
> Thank you in advance.
>
> Best,
> Shi
> _______________________________________________
> Wannier mailing list
> Wannier at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/wannier
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/wannier/attachments/20201208/cbf9331f/attachment.html>