Z2 number for Odd No of Occupied Bands

[shahid-sattar]

As the subject: it is not an issue in code but mainly a discussion on the consequence of odd number of occupied bands I am studying the case of 2D single layer MnBi2Te4 (a magnetic TI in bulk but not a TI in single-layer limit) and want to calculate topological indices for the single layer only or odd no of layers (3L, 5L). The wannier-projected BS of 1L nicely match to DFT result with 29 occupied bands and 17 unoccupied bands (total 46 bands and projections). Having the FL at 0 eV, while the Chern No calculation gives zero regardless of the choice of Nk1, Nk2, the Z2 no (for K3:0/0.5) is changing for the change in Nk1, Nk2. I am wondering how one should understand this change owing to the odd no of occupied bands (29 in 1L, 87 in 3L and so on).

[quanshengwu]

Please check the No of occupied bands, it could be 28 instead of 29 since the indices shown in the Xmgrace sometimes starts from 0, sometimes starts from 1.