[salmon-user:00133] SOC and LDA+U

[Towfiq Ahmed]

Dear Professor Yabana and the Salmon Development Team, 

I was wondering if you have the following two versions of SALMON code? 

1) SALMON with spin-orbit coupling ? 

2) SALMON with LDA+U for strongly correlated Mott type insulators. 

If these methods are not currently implemented, can you please let me know (particularly for spin-orbit) what would be the potential challenges if someone wants to implement spin-orbit coupling ?  I believe, one probably can start with Paul-spinor and non-collinear spins, but might  also consider an appropriate treatment of psueopotentials ? But are there any other inherent issues that can make such implementation very difficult ? 

I will greatly appreciate your feedback. If you have any experimental version of SALMON that can handle SOC and (or) LDA+U, I would like to give them a try if possible. I want to calculate non linear response and HHG of some systems which either are 3d and 4d transition metal compounds or have high Z number. 

Best regards, 

Towfiq

####################################################

Towfiq Ahmed, PhD

Staff Scientist 

T-4, Theoretical Division, 

Los Alamos National Laboratory, Los Alamos, NM 87544, USA

Phone: +1 505 665 8413 (office)

           +1 206 370 9255 (cell)

Email: atowfiq@lanl.gov

Alt-email: ato...@gmail.com

####################################################

[Kazuhiro Yabana]

Dear Towfiq,

Thank you for your inquiry and interest in SALMON.

As for the spin-orbit coupling, we are currently working on it.
As written in our web site, we develop our code at github,
https://github.com/salmon-tddft
SALMON2 is the current branch of our developing code.
Although the spin-orbit function looks mostly implemented in the code,
there remain (perhaps many) problems. You can download and try it.
If you find bugs (that we believes there remain quite a few), we will be
grateful to have your reports.

We usually release a new version which is taken from the developing
branch when we have (at least one) publication so that we are
confident on the code (at least for the published calculation).
I am not sure how it takes, but I anticipate it will take a few (or several) months.

As for the LDA+U, it will be a bit future. Although a number of
subroutines are already prepared in the code, they have never
been checked.

Best regards,
Kazu

---
Kazuhiro Yabana
Center for Computational Sciences
University of Tsukuba