[Pw_forum] about elphon.f90

[Stefano de Gironcoli]

Antonio Sanna wrote:

> Dear Developers,
>
> I'm tryng to obtain the electron-phonon coupling constants  g_kq  from 
> the   subroutine "elphel" in  "elphon.f90",
> I'm keeping:
>
>     elphmat (jbnd, ibnd, ipert)           " 
> elphmat(j,i)=<psi_{k+q,j}|dvscf_q*psi_{k,i}>  "
>
> this, I think, is the coupling with respect to the phonon mode q (not 
> to cartesian displacement of the atoms )  Is it right?
>
NO.
it is the coupling corresponding to the displacement patterns used in 
the dynamical matrix generation which are
neither of the two.
In order to go to cartesian  coordinates or projecting on vibrational 
eigenvectors a transformation is needed.

> And also, in the subroutine that calculates the linewidths gamma:
> after a summation over the fermi surface and a symmetrization, gammas 
> are generate as
>
>       gamma = gamma + real (conjg (dyn (mu, nu) ) * el_ph_sum (mu, vu) &
>                     * dyn (vu, nu) )
>
> what this transformation (dyn* elphsum dyn)  means?

after the call to symdyn_munu the array el_ph_sum contains the el-ph 
matrix  with respect to cartesian displacements.
the array dyn contains instead the eigenvectors of the dynamical matrix 
(look at the end of readmat subroutine).

thus the  above transformation projects the el-ph matrix on a given 
phonon mode (whose mode number is nu).

hope this helps.

Stefano de Gironcoli