[Wannier] SCDM-k method and centering of WFs

[Giovanni Pizzi]

Dear Marco,
unfortunately it's not so easy to combine SCDM with symmetries. This is a topic of research, and we are working for instance on other approaches that do not use SCDM to get automatically WFs that are instead atom-centred with symmetries.

If you are working on specific systems, the only suggestion is to try to look better into the systems and get WFs in the "usual" way, specifying initial projections close to what you want to get.
You can also use the SLWF (selectively-localised WFs) and the SAWF (symmetry-adapted) routines, implemented in W90, to see if they help for your case (look also at the tutorials http://www.wannier.org/support/, e.g. #21, #22 and #26).

For the delocalised orbitals, it's hard to say without knowing more details on the system (and even there, it requires careful testing in multiple configurations). But one general comment (not sure it applies to you) is that rather than removing the problematic ones, you might want to check if adding a few more bands helps, or if "playing" with the disentanglement windows helps (also, to check: is the disentanglement converged? Are the WFs real-values if you plot them in real space?)

Hope this helps.
Best,
Giovanni

--
Giovanni Pizzi
Theory and Simulation of Materials and MARVEL, EPFL
http://people.epfl.ch/giovanni.pizzi
http://nccr-marvel.ch/en/people/profile/giovanni-pizzi

On 7 Nov 2022, at 11:23, Marco Marino <marco.marino1 at unimi.it<mailto:marco.marino1 at unimi.it>> wrote:

Dear Wannier90 group,

I am working with NiO and Cr2O3 bulk systems, in their antiferromagnetic phase. Using the SCDM-k method and considering only the valence bands (from a DFT collinear calculation), I obtain WFs sufficiently localized (apart from a few WFs totally delocalized (3 with respect to 138 wannier functions), which reproduce quite well the electronic band structure (apart from those few bands associated with the totally delocalized WFs).

The problem is that these WFs are not completely centered on the atoms. How can I center my WFs? How can I decompose my WFs into angular components, in order to be able to do a wannier calculation with the projectors so obtained? Moreover, how can I solve the problem of the few delocalized WFs (removing them makes the results worse)?

thank you for your willingness,
best regards,
marco marino
phd student
University of Milan


<signature.jpg>


La Statale per il futuro
Salute, transizione digitale, sostenibilità
Il tuo 5xmille ai nuovi progetti di ricerca dell’Università degli Studi di Milano
Codice fiscale: 80012650158
_______________________________________________
Wannier mailing list
Wannier at lists.quantum-espresso.org<mailto: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/20221115/553a1164/attachment.html>