Currently I met a problem when I calculate the surface state of a topologically non-trivial thin film system. I did a DFT calculation and fit the result with wannier function. The fitting looks pretty good around the band gap region. Then I use tight binding to calculate the surface state. I found that the band structure along -X -> G -> X didn't preserve the time reversal symmetry (E(k) is not equal to E(-k) and it become a trivial edge state. For a system without magnetisation and external B-field, the time reversal symmetry can be preserved. So the relation of E(k) = E(-k) should be always true? I can't find any reason for my result. Is there any people who did similar things before and have some suggestion ?

For the band gap region: enter image description here

The surface state was shown in below: enter image description here For the notation of R in the surface state plot, it is -X. So it is the plot from -X -> G -> X. and it didn't contain the time reversal symmetry.

  • 1
    $\begingroup$ +1 perhaps you could share a plot to show what you are seeing, which may help identify the problem? $\endgroup$
    – ProfM
    Feb 19, 2021 at 6:13
  • $\begingroup$ @ProfM I have already uploaded a plot for a 5atomic layers system. For other atomics layers system, the surface state are still very messy which provided no information for the material. This material was provide to have a non-trivial edge state in bulk. I have already verify the result. But I found that the result of the edge state depends on the fitting of the wannier function. For my wannier function even though the spread is a bit large, as you can see the fitted result is still very good. But it give a non-sense surface state result. $\endgroup$
    – JensenPang
    Feb 19, 2021 at 9:02

1 Answer 1


Wannier90 might not be good at preserving the symmetry. But they probably include a few new methods to enforce symmetry in Wannier90.v.3.1.0. Maybe you can check this. http://www.wannier.org/features/

Also, WannierTools can symmetrize the hr.dat, but from my personal experience it sometimes gives you worse results than the original hr.dat. http://www.wanniertools.com/symmetrization.html

  • $\begingroup$ Thank you for your answer. May I ask a few more follow up question. Like you said, the wannier90 may not good at preserving the symmetry. But from the result I got, did I do anything wrong which may cause this mistake? Because by the direct inspection, the fitting is very good, even though each Wannier function is a bit larger than the Lattice constant. Is this an acceptable result ? $\endgroup$
    – JensenPang
    Feb 20, 2021 at 0:29
  • $\begingroup$ fitting is definitely NOT good to check symmetry... Maybe spin texture is better $\endgroup$
    – Bo Peng
    Feb 20, 2021 at 2:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.