search

wannier-developers / wannier90

Official repository of the Wannier90 code
http://www.wannier.org
GNU General Public License v2.0
236 stars 139 forks source link

Inconsistency of the rmn and AA_R data, is this a bug? #302

Open goodluck1982 opened 4 years ago

goodluck1982 commented 4 years ago

I want to reproduce the Berry curvature and optical conductivity of postw90.x by hand through reading seedname_hr.dat and seedname_r.dat file. I took monolayer MoS2 as example. However, I find the data I calculated manually do not coincide with the data calculated by postw90.x. Concretely, BCz (z component of Berry curvature) and Sxx (xx component of optical conductivity) are consistent with the results of postw90, but BCx, BCy, and Szz are different from the results of postw90 obviously. See the figures below. I find that the differences originate from the different rmn=<um|r|un> data used. The data written in seedname_r.dat are different from the AA_R data used in postw90 (subroutine get_AA_R in get_oper.F90). Then I output the AA_R data used by postw90. After recalculating using the AA_R data I got all data consistent with the results of postw90.

Is this a bug? And which results are more reasonable, the data calculated using seedname_r.dat or using the AA_R of postw90?

Berry curvature Omega: MoS2-curv

Optical conductivity sigma: MoS2-sigma

stepan-tsirkin commented 4 years ago

Dear goodluck!

What value of parameter use_ws_distance are you using?

goodluck1982 commented 4 years ago

To be simple for comparison, I used use_ws_distance=F. In fact, because I used k mesh up to 10x10x1 in DFT calculation, I don't think use_ws_distance=T will make noticeable difference.

giovannipizzi commented 4 years ago

Hi! Thanks for the report. In case you already resolved this issue offline, could you please report back? Otherwise, I'm pinging also @ivosouza0 for further comments.

goodluck1982 commented 4 years ago

I only found the issue. And I also found that the optical conductivity calculated using seedname_r.dat agrees better with the dielectric function from VASP. However I'm not able to resolve it.

Hi! Thanks for the report. In case you already resolved this issue offline, could you please report back? Otherwise, I'm pinging also @ivosouza0 for further comments.