[Wannier] bcc Fe with VASP+Wannier90

[Marcus Ekholm]

Thank you, Stepan!

We'll try this!

Best regards,
Marcus
________________________________
From: Stepan Tsirkin <stepan.tsirkin at uzh.ch>
Sent: 16 October 2020 20:51
To: Marcus Ekholm <marcus.ekholm at liu.se>
Cc: wannier at lists.quantum-espresso.org <wannier at lists.quantum-espresso.org>
Subject: Re: [Wannier] bcc Fe with VASP+Wannier90

Dear Markus,

I did calculations for bcc Fe with spin-orbit (example18), and the spreads were quite equal (they differed in 5th digit), although I used a smaller grid of 8x8x8, and only 1000 iterations.

I think you have to  few bands in the calculation (14 per spin)
When I used 32 bands in total (16 per spin), I also had bad Wannier functions. With 64 bands it was much better. Try 32 bands per spin or more.

Probably you may play with disentanglement windows - narrow the frozen window and extend the outer window.

Regards,
Stepan.


------------------------------------
Stepan Tsirkin
University of Zurich
stepan.tsirkin at uzh.ch<mailto:stepan.tsirkin at uzh.ch>
[cid:B8278E93-B140-4764-8D01-D66183C3E242 at ggaweb.ch]
https:///wannier-berri.org

On 16 Oct 2020, at 14:31, Marcus Ekholm <marcus.ekholm at liu.se<mailto:marcus.ekholm at liu.se>> wrote:

Dear all,

We are trying to reproduce example 8: bcc-Fe from the tutorial (v3.1), but with VASP. However, we find that the spread of all 9 Wannier functions are quite different. In the the solutions booklet, the spread of all 5 d-orbitals are identical to the 6th decimal, but we get differences already in the 2nd or 3rd decimal. This seems to have some implications when we later use the Wannier Hamiltonian for other purposes.

How can we improve this? I use a Monkhorst-Pack 21x21x21 mesh for the VASP calculations and I have tried setting the number of iterations (both num_iter and dis_num_iter) to 100000 and wait for convergence. We tried different frozen windows, different numbers of steps, different k-meshes, different initial projections, but we have never been able to reproduce the Quantum Espresso results.

Has anyone tried bcc-Fe based on VASP?

Best,
Marcus Ekholm
Linköping University
<final_state_VASP_bccFe_up><INCAR.Fe><wannier90.up.win<http://wannier90.up.win>>_______________________________________________
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/20201019/2a225361/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: wannierberri.png
Type: image/png
Size: 7302 bytes
Desc: wannierberri.png
URL: <http://lists.quantum-espresso.org/pipermail/wannier/attachments/20201019/2a225361/attachment-0001.png>