Dear All, I want to study two polymorphs of an iron containing compound. Each polymorph has 4 symmetrically equivalent Fe-sites (space group Pnma). One of the polymorphs have the Fe-sites at the inversion centers of the crystal and the other does not. Looking at the band structure I find 20 isolated low-energy bands of strong d-orbital character, with no hybridization with other states, and a small bandwidth. So I want to extract 20 wannier functions to construct a low energy model. In order for the model Hamiltonian to be symmetrical I would like to have the wannier functions well centered exactly at Fe atomic positions, and those WFs should be interchangeable between them by crystal symmetry operations.
When using QE with pw2wannier90 and wannier90, I am able to extract non-symmetry adapted WFs that represent the band structure perfectly, but the wannier centres are slightly misplaced from the Fe sites, and therefore I have 20 non inequivalent WF and slightly different Hamiltonian matrix elements between sites. However when I try to do the calculation with the ‘site_symmetry’ tag set to .true., I find that the wannier functios keep being misplaced and look exactly the same as the ones without symmetry. Looking further into the .sym and .dmn I find that the only operation appearing is 1.0000 0.0000 0.0000 0.0000 1.0000 0.0000 0.0000 0.0000 1.0000 0.0000 0.0000 0.0000 And the header of the .dmn file is 170 1 40 40 So, definitely the calculation has not detected the symmetry of the structure, because only one operation is applied and the amount of irreducible k-points (which is 18 in my scf calculation) is 40. I’m sure I am missing something, I think that the fact that I had to use explicit 40 k-points for the 'nscf' calculation to match the k-point list in the .win file Is contrary to the calculation considering any symmetry. So thats why the later pw2wannier90 run just does not consider any symmetry. Any suggestions for solving this would be much appreciated. Best regards, Oier.
_______________________________________________ Pw_forum mailing list [email protected] http://pwscf.org/mailman/listinfo/pw_forum
