I have tried to calculate for a system with spin-orbit coupling (SOC) using SIESTA and Wannier90, but an error

non-collinear spin not implemented yet

appears in a SIESTA output file. I would be very grateful if you have some time or knowledge to solve this error or suggest any clue for that.

Here is what I have tried so far.

Graphene is only used as an example to check the spin-orbit calculation. Fisrt, I calculated as the input files are written, then the calculation completed without any errors.

Next, I added Spin spin-orbit to the fdf file to include SOC. The calculation couldn't be completed and an error

getFileNameRoot: non-collinear spin not implemented yet

was displayed. Comparing the output file without SOC, the error is in the line where SIESTA reads nnkp file for Wannier90. It should have been

siesta2wannier90: Reading the graphene.nnkp file,

instead of the above "non-collinear" error which only occurs when SOC is enabled. If I calculate with SOC without Siesta2Wannier90 lines (see the last part of the fdf file), there is no error. So my conclusion so far is that the cause is in the Wannier90 or the combination of SIESTA and Wannier90, especially in the nnkp file.

Adding spinors = T to the win file and recreating nnkp file before starting the SIESTA calculation with SOC did not make any difference.

Is there anyone who can suggest where to correct or which command is needed to complete this calculation? I appreciate your help in advance.


1 Answer 1


As it states, there is no implementation of Wannier90 with spin-orbit.

It is in the works, so it will come.

  • $\begingroup$ Thank you nickpapior for your reply. According to Wannier90's tutorial, I thought that spin-orbit could be considered in Wannier90. For example, in #17: Iron (page 28) of the tutorial, titled "Spin-orbit-coupled bands and Fermi-surface contours", does this tutorial not include SOC? If I am misunderstanding this, I would be grateful if you would let me know. $\endgroup$
    – Yuma K
    Commented Jun 25 at 9:15
  • 1
    $\begingroup$ @YumaK it is because the spin-orbit in siesta requires some adaptation to fully work with Wannier90 spin-orbit. So both siesta and wannier90 can do those things, but currently siesta doesn't have the interface ready for doing wannier90 SOC. $\endgroup$
    – nickpapior
    Commented Jun 25 at 18:59
  • $\begingroup$ Thank you again @nickpapior for your reply. So the problem is in the interface between SIESTA and Wannier90, which is what I fear. For reference, if you know, could you give me some documents (a link or a title is enough) that show the interface like Siesta2Wannier90 does not support calculation with SOC? I really appreciate and do not doubt your information, but I just want to find the official source if possible, because I have tried to search such a description somewhat before posting this question and ended up not finding it. $\endgroup$
    – Yuma K
    Commented Jun 26 at 3:15
  • $\begingroup$ The official source is the output of siesta which says: non-collinear spin not implemented yet. Otherwise you have to look in the source code, I can't recall if when we announced it it was specifically noting this point... $\endgroup$
    – nickpapior
    Commented Jun 26 at 10:12
  • $\begingroup$ Thank you for your response. I understand now that the Siesta output was related to the interface with Wannier90. I may have made inappropriate comments because I wasn't clearly aware that you were one of the developers, and I apologize for that. If you happen to recall any official documentation that addresses the interface not ready for SOC, I would be grateful if you could let me know. Thank you again for your time and expertise. $\endgroup$
    – Yuma K
    Commented Jun 27 at 1:08

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .