[QE-users] The difference of two wavefunctions from two methods

[徐思源]

Hi all

We are using the wannier function in the QE, and we have noticed an interesting question, it seems like the nnkp blocks in nnkp file will largely influence the phase of the wave funciton (the zerophase judge in pw2wannier90.f90), more specifically, the g_kpb parameter in nnkp block. we want to know the role of this phase change play or its physical meaning. 
we have notice that even different places in the real space of a certain wave funcition will have different phase change, we really want to know the reasons beneath this setting, however those codes are too many to search and follow.
The context is that we have tried another calculation method to obtain the wave function, by calculating those wave functions at certain kpoints one by one. Under this situation, the g_kpd parameters are often 0 0 0 no matter which k point this wave function belongs to, and accordingly, the wave funtions will have a different phase factor from those wave functions calculated under normal process while their abs seem like equal. These two kinds of wavefunctions will have different results in our following calculations. It is out of our expectations since we didn't think this will influence our results.

Thanks
Xu Siyuan