That is exactly what I observe.sponce wrote: You should not get exactly the same eigenvalues as nscf as they have been interpolated. However they should be very close (or else the interpolation is not good).
So that technically means, that for given K and q point, I take fix all three indexes (ibnd, jbnd and mode index) and, for example, loop through phonon mode indexes. In case, if frequency is the same for two states with different mode indexes, I average matrix elements for this two cases and so on. I do the same for E(ibnd, ikk) and E(jbnd,ikq), where ikk is a k index and ikq is a k+q index. Correct me if I am misinterpreting.sponce wrote: You also need to do an average over degenerate eigenenergies, phonon modes etc.
Does what you say goes more to the case, when fine grid is given by nqf1, nqf2, nqf3 ?sponce wrote: However you should either use all the equivalent q=A (depending on sym it could be 6 or so) or use the proper weighting factor. You have to do an average over the equivalent q=A if they are not exactly the same.
As I use filqf file, set in such way, that every qpoint enters only once I think I am fine with just setting q=A.
I was not accurate enough describing what I wanted to achieve, what I actually needed was |g_q| (q=A), and thus sqrt((abs( epf17(val_bnd, val_bnd, 11, 1) )**two)/(two*wq)). I wanted the result to agree with data on Figure 1, bottom panel, of http://epw.org.uk/Documentation/GaN , where |g_q| at q=A is around 93-95 meV (this is the source of the reference value of 100 meV), which I think I finally achieved. However, I didn't check against quantum espresso results yet.sponce wrote: You just need to be consistent with what is computed in elphon inside QE/PH. I do not remember exactly but you might have to do an average there as well.
I do not know how you got the 100 meV reference one. Is it printed by QE? In my case I directly printed them from the code both from QE and EPW for comparison so I'm not sure how you got the 100 meV?
I have one more question. Let's say I want to consider deformation potential, which is a D in <k+q|dV|k>=D*q. My question is what are the units of <k+q|dV|k> and hence sqrt(abs( epf17(val_bnd, val_bnd, 11, 1) )**two) ( sqrt(|g|)**2 without the 1/(two*wq) coefficient)? I believe it should be Energy/length (Ry/a.u.l, Ev/Ang) units, however I can't obtain such dimension. If I assume |g_q|, as well as phonon frequency to be in Ry, which is a default for QE and I believe EPW, I end up with <k+q|dV|k> having a strange units of Ry**3/2 .
Best,
Andrei