[Wien] Weird wannier functions coming out from wien2wannier

[Pablo Villar Arribi]

Dear Wien2k/Wien2wannier users,

I am trying to plot (with XCrysDen) the Wannier projections of the Fe-d 
orbitals in some iron-based superconductors with Wien2wannier, but the 
resulting orbitals do not resemble at all the typical d-like shape that 
one should expect for them, however the agreement with the DFT bands is 
perfect.

The spreads of the wannier orbitals are very reasonable compared with 
what can be found in the literature, all the hoppings in the real-space 
hamiltonian coming out of Wannier90 are real and the centers are exactly 
located in the Fe atoms. I have also tried exactly the same procedure 
with Quantum Espresso and I was able to obtain very d-shaped orbitals, 
almost identical in the case of the dz2 and a bit distorted towards the 
ligands in the case of the dxz and dyz (as it should be in these 
compounds), with almost indentical spreads and centers, and both band 
structures coming out of the Wannier projection fit perfectly one on top 
of each other. In addition, I have been able to set a proper mesh with 
"write_inwplot.x" and I have checked visually that they are located in 
the Fe atoms (but with the weird shape I mentioned before).

Any ideas on why this different behaviour between Wien2k and Quantum 
Espresso? Could it be due to the intrinsic diferences between an 
all-electron code and a pseudopotential code in terms of their Bloch 
states, which are used as a starting point for Wannier90?

Thank you very much in advance for your answers.

Pablo

-------------------
Pablo Villar Arribi
PhD student at ESRF
Grenoble, France