Electron Velocity Matrix ELement

[yimaverickxia]

Dear All,

I am using EPW4.0 to calculate electronic transport lifetime, which is generally different from the quasi-particle lifetime with an additional velocity-related term. I am wondering is there a way to directly calculate it within EPW4.0? It seems to be possible according to the the recent paper introducing EPW4.0.

I also find that the original electronic velocity through Wannier interpolation is probably disabled (as stated in the document "not working"). But in the EPW4.0 paper, it is stated that the velocity terms can be extracted through (i) finite differences and (ii) analytic derivatives in the local approximation. Is there a way to directly access these velocity matrix element for each interpolated electronic state (only diagonal term will be good)?

Thanks in advance!

Yi Xia
MSE UCLA

[sponce]

Dear Yi Xia,

Indeed, the vme input variable is the correct velocity through Wannier interpolation but currently does not work (need to additional coding).

There are no input variables to get the velocity in the local approximation but it is included when you compute \lambda_tr for example.

If you are interested in them, you can find them in the selfen_phon.f90 routine for example:

vkk(:, ibnd ) = 2.0 * REAL (dmef (:, ibndmin-1+ibnd, ibndmin-1+ibnd, ikk ) )

Best,

Samuel

[yimaverickxia]

Dear Samuel,

That's very useful! It looks like I can directly add the related part to selfen_elec.f90 in order include this velocity-related term. Thank you

Best,
Yi

[mfzhu]

Dear Samuel,

I am using EPW4.0 with QE5.4. I am interested in the velocity term too.
I've compared

"dmef", 2*REAL (dmef (:, ibndmin-1+ibnd, ibndmin-1+ibnd, ikk ) )

and

"vmef", REAL (vmef (:, ibndmin-1+ibnd, ibndmin-1+ibnd, ikk )

.
They are different for the case of Pb (in the example directory in case of woSOC) and they don't seem to have any relation from the . dmef ~ 10^-3 and vmef ~ 1
According to our calculation, it seems that vmef has the right value if it is in Rydberg unit compared with the velocity roughly estimated from Pb's band structure.

My question is : what's the differences of these two matrix? or Why they are so different?

All the data we calculated are based on the input files in EPW/examples/pb/woSOC/ and outputed in file selfen_phon.f90.