[Wannier] Help: WF centering and projected band dos calculations

[Stepan Tsirkin]

Dear Abhiyan,

Let me answer your question 1.

If there is a symemtry that constrains the WFs at certain positions - 
they will stay there. And in your case there is. However, symmetries are 
not kept exactly in wanier90.

The maximal localization procedure may deviate from initial projections, 
e.g. by constructing hybrid orbitals, if it leads to more localized WFs. 
To prevent this, one needs to use Symmetry-sdapted WFs (SAWF), which 
forces the WF to retain the same symmetry properties as the initial 
projections. However, in the current implementation this procedure does 
not work with VASP, and does not allow use of frozen window. (at least 
in version 3.1, not sure if any of that is planned to be fixed in the 
next release)I recently implemented SAWF in wannierberri, which does 
work also with VASP and frozen window, you may have a look here: 
https://tutorial.wannier-berri.org/tutorials/6_wannierisation/wannierise.html 


Other option - NOT to use maximal localization (projections-only, or 
disentanglement-only), i.e. set num_iter=0.

However, the bigger problem  that I see in your wout file is not the 
centers, but the spreads - they are too large. As a rule of thumb, if I 
see spreads of some WFs that are larger than 10 A^2, I consider the 
wannierisation as unsuccessful. But typically a good wannierization has 
spreads ~< 5. This typically happens because of inconsistency of 
projections with the outer and frozen window that you have selected. 
Recently I have written a tool which helps to assure this compatibility, 
please have a look here: 
https://tutorial.wannier-berri.org/tutorials/7_find_projections/find_projections.html

Best,

Stepan.

On 4/23/25 23:01, Pandit, Abhiyan wrote:
>
> Dear Wannier Developer(s),
>
> I am using VASP(V-6.5.1)+Wannier90 interface to wannierise the atomic 
> orbitals. _Referring to my sample files_ (INCAR, wannier90.win, and 
> wannier90.wout for bcc Rb) shared in the link: 
> https://drive.google.com/file/d/19qJwwRKMPEuFVzx_oRowaJae5PCCg06F/view?usp=sharing, 
> I have following queries (which I could not resolve (or not working) 
> through the general overview of the Wannier90 User Guide + 
> Email/Discussion archive).
>
>  1. How can I constrain (tags to be modified) the Wannier functions
>     (WFs) to completely localize at the desired/finite positions (at
>     defined projection sites) that I want?
>  2. Would the output WFs (*.xsf) be in the same order as the defined
>     projections in the input? How to make this happen if it is not
>     present in my current input file?
>  3. Although I input dos_project and band_project tags, I just get
>     seedname-dos.dat with just two columns, should be total dos; and
>     seedname_band.dat. What other tags should I input and to which
>     files and format would these be written?
>  4. Any obvious comments or feedbacks from your side over my inputs
>     for improvements or better accuracy?
>
> P.S.: This is reposted to my initial post of 04/20/2025 (I changed 
> file attachment to the sharing option here as suggested by Prof. 
> Marzari).
>
> Thank you so much for your valuable support.
>
> Best,
>
> Abhiyan Pandit, PhD
>
> Postdoctoral Researcher,
>
> California State University, Northridge, CA, USA
>
>
> _______________________________________________
> 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/20250424/793d4b17/attachment.html>