Some question about ndb.dipole in yambo

[baiyunfei]

I want to use ndb.dipole generated by yambo to do futher analysis (calculate the charge center of a wavepocket composed of series of Bloch waves in a supercell), However I have a few question about how this database was produced.
Yambo use length gauge to calculate dipoles by default, the out put dipole is also in the unit of length. However, I have an impression that length operater is not well defined, how does yambo solve this problem? In other words , Is it safe to use <n1k1|r|n2k2> in ndb.dipoles to calculate the charge center?
I noticed yambo provides another option for dipole calculation (ie velocity gauge), so I can use modern polar modern polarization theory to calculate charge center. However , dipole in velocity gauge needs derivation with respect to wavevector , typically requiring a dense mesh. However, In a typical GW-BSE calculation , we can't afford a too dense mesh. Is it reliable to use velocity gauge ?
Best wishes !

[Daniele Varsano]

Dear Yunfei,
you can read in the latest Yambo paper (Section 3.1) how dipoles are calculated in Yambo with the different available options.
at:
https://iopscience.iop.org/article/10.1 ... b15d0/meta

Best,
Daniele

[baiyunfei]

Thanks for your reply, I will look into it.