search

Chengcheng-Xiao / VASP2WAN90_v2_fix

An updated version of the VASP2WANNIER90v2 interface
86 stars 37 forks source link

write UNK.NC file? #6

Closed Chengcheng-Xiao closed 4 years ago

Chengcheng-Xiao commented 4 years ago

This seems to be easier than the .spn file.

So, do this first.

WVFN(IND)=WCOLLECT(ISTRIP)%CR(NX+(NC-1)*WP%WDES%GRID%RL%NROW)

By the looks of it, we only need to halve the CR array and output it to two WVFN arrays and then print it out.

Chengcheng-Xiao commented 4 years ago

The UNKxxxxx.NC file can now be written with non-collinear calculations.

Also, I've added a flag LUNK_FMTED to the INCAR file to control the formate of the UNK file ( similar to wvfn_formatted in pw2wannier90)

However, I'm not able to reproduce the exactly same result as the QE version. Specifically, for the "spin-up" spinor wavefunction (e.g. one of the doubly degenerated band, like WF=[1,0]) I'm seeing erratic behavior in the second spinor channel (using wannier_plot_spinor_mode = down).

Not sure if this is expected. Need more investigation.

Chengcheng-Xiao commented 4 years ago

So, changing spin quantization axis and use projection only scheme gives me much more smooth plots (for both spinor components) and I think the erratic behavior of the z axis plot is just a natural artifact of ZERO WF (essentially we are plotting 0 WF which should in theory be zero).

Chengcheng-Xiao commented 4 years ago

Done in https://github.com/Chengcheng-Xiao/VASP2WAN90_v2_fix/commit/b923aace861177f112be10038a4fd6bfbb93d056. Issue closed.