[Wannier] Abinit interface to Wannier90 code for spinor wave functions

[Giovanni Pizzi]

Hi,

I cannot comment on the specifics of the ABINIT interface. But note that the code, when run in library mode, ignores the value in the input (it should also print it in one of the outputs):

https://github.com/wannier-developers/wannier90/blob/a75344b646227c9d7f12d90af4a35a902e940dca/src/parameters.F90#L750

One needs to ensure that the library instead passes the spinors_loc variable from the interface fortran call.

Let me also mention that we merged last month a major refactor of the whole Wannier90 code, to make the library much more flexible. In the future it might be useful to ping the developers of the interface codes you use (if you want) so that they are aware of the new library mode and consider releasing a new interface with the new library (that will be released in the next few months with a new version of Wannier90). The main contact person for support on the library mode (who is one of those who did most of the work!) is Jerome Jackson at STFC.

Best,
Giovanni


On 9 Mar 2024, at 14:44, Shima Sharifi (Tyndall) <shima.sharifi at tyndall.ie> wrote:

Dear all,
I am currently working on generating MLWFs for spinor-values wave functions using Wannier90 interfaced with ABINIT ver9.10.5. The version of Wannier90 utilized as a library during the installation of Abinit is 2.0.1.1, which is compatible with spinor wave functions.

However, upon running the calculation, despite setting the variable "nspinor=2" in the Abinit input file and specifying "spinors=.true." in the Wannier90 input file, I encountered an error indicating that "param_get_projections: spin is defined but spinors=.false."

It appears that the interface program is disregarding the specified value of "spinors" in the Wannier90 input and defaulting to a case without spin-orbit coupling calculation. On the other hand, it recognized that the spin is defined by reading the projection section of Wannier90 input file.

Does this imply that the Abinit interface to Wannier90 code currently only supports cases without SOC?

Best regards,
Shima

_______________________________________________
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/20240319/1228d124/attachment.html>