[Wannier] Hamiltonian in MLWF basis

[Zhu, Jianxin]

Dear Jonathan and Respectful Wannier Users,


I have recently been using wien2wannier (thanks to Jan Kunes, Philipp
Wissgott, and Elias Assmann) together with wannier90 for several physical
systems.

Noticeably, for some systems (e.g., tetragonal LaOFeAs etc.), the
Hamiltonian matrix elements in the MLWF basis are diagonal in the WF
orbitals on the same atomic site.
However, for some other systems (e.g., hexagonal YCo5 etc.), the
Hamiltonian matrix elements in the MLWF basis are not diagonal in the WF
orbitals on the same atomic site.
The latter case is surprising to me because the conventional tight-binding
approximated Hamiltonian (based on a local atomic orbitals) are always
diagonal on the same atomic site.

I check back to the paper by Mostofi et al., Computer Physics
Communications 178. 685 (2008), and see the Hamiltonian is derived from
rotated Bloch states through U^k and U^dis(k) (cf. Eqs. (19-20) in the
above mentioned paper).
Due to these U^k and U^dis(k), it seems to me that there is no guarantee
that the Hamiltonian matrix elements must be diagonal in the basis of MLWF
on the same site. 
Do I understand correctly?

Have you encountered this kind of situations in applying MLWF to practical
systems?

Your suggestions/thoughts are appreciated.

Best,

Jianxin

P.S.: Elias, I cc this message to you in case you also have suggestions.



#################################
Jian-Xin Zhu, Ph.D
Theoretical Division, MS B262
Los Alamos National Laboratory
Los Alamos, New Mexico 87545
Phone: (505) 667 2363 (T-4);
(505) 667 6602 (CINT)
Fax: (505) 665 4063
Email (main): jxzhu at lanl.gov
Email (backup): physjxzhu at gmail.com
URL: http://theory.lanl.gov
#################################