Running wannierization outside of workflow

[Jake Burner]

Hi there,

I would like to run the wannierization outside of the koopmans workflow (using the SCDM functionality in VASP 6). Is it relatively straightforward to run the remainder of the workflow using the results of the wannierization if I choose to do this?

Thanks in advance for any advice!

Jake

[Edward Linscott]

Hi Jake,

Great question -- automated Wannierisation is certainly super-useful! In principle, yes, you could do this by moving the output of your SCDM calculation into your calculation directory and then running the calculation with from_scratch = False. This would be a bit fiddly, however, because you would need to make sure that the directory structure follows precisely what koopmans expects to find.

As it turns out, we are in the process of integrating a newer (and we believe better) automated Wannierisation algorithm into the koopmans package (see 10.1038/s41524-023-01146-w and 10.1038/s41524-023-01147-9). This should be publicly available in the not-too-distant future. (We will be looking for guinea pigs soon if you are interested!)

We are also working on making the workflows more modular so this sort of thing is easier in general, but that work is more long-term in nature... 

Best wishes,

Edward

[Jake Burner]

Hi Edward,

Thanks for the info and papers! I will keep an eye out for the implementation in koopmans.

Best,

Jake

[Jake Burner]

Edward,

Sorry, I had one other question that came to mind! Will there be any issues with using the wannier90 outputs that I obtain from VASP in the workflow? Is there any reason I should specifically be using QE to generate the wannier functions? I am really a true beginner when it comes to wannier90, so I am not exactly sure what is going on with postprocessing, etc.

Thanks!

Jake

[Edward Linscott]

Dear Jake,

Ah, sorry I overlooked the fact you said you wanted to do the SCDM with VASP -- I was focused on whether or not you could do SCDM at all.

Yes, this will be an issue, because the Koopmans calculations that follow the Wannierization are performed in QE, and they will need to read in information from the preceding DFT calculations such as the wavefunctions (in QE format) in addition to the Wannier90 files. You do not want to mix-and-match Wannier90 files generated from a VASP calculation with wavefunctions from QE, because applying the U matrices from VASP+W90 to the KS states of QE will not give you MLWFs.

I am not familiar with W90 + VASP, but performing SCDM with W90 + QE should not be too different.

Best wishes,

Edward

--
You received this message because you are subscribed to the Google Groups "koopmans-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to koopmans-user...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/koopmans-users/3f36d8f8-00df-4cbc-8310-62934b60b3a5n%40googlegroups.com.

[Jake Burner]

Thanks very much for the detailed response!

Have a nice day,

Jake