Page 1 of 1
How to get the el-ph coupling matrix in complex number form?
Posted: Tue May 17, 2022 8:59 am
by edward
Dear Developers,
Along with the growth of the diversity of the electron-phonon topic greatly powered by your code, it's sometimes necessary to get the coupling matrix element in complex number form rather than |g|. I wonder if there is already a way to get it or maybe would you like to consider it to be a new feature in the next version of EPW code.
Re: How to get the el-ph coupling matrix in complex number form?
Posted: Fri May 27, 2022 4:59 pm
by hlee
Dear edward:
For the moment, we don't have any plan to add it.
One of the reasons to print out |g| instead of g is to confirm the Wannier-interpolated electron-phonon matrix elements are comparable to the corresponding values from the direct DFPT calculations. As you know well, g is not a gauge-invariant quantity and if we print out g, it is not trivial to compare it with g from the phonon code.
I think that you can easily print out these values by modifying the subroutine of print_gkk in printing.f90.
Sincerely,
H. Lee
Re: How to get the el-ph coupling matrix in complex number form?
Posted: Sat Dec 24, 2022 11:33 am
by baiyunfei
As far as I know, EPW use wannier interpoltation to interpolate dvscf and wavefunction on a finer mesh ,then use them to construct electron-phonon matrix. However wannier interplotation is known to bring some gauge problems.
My question is : if I properly modify the ''print_gkk in printing.f90.'' ; set the coarse mesh of nscf calculation the same as fine mesh of wannier interpolation , and get the output of complex electron-phonon matrixs, will the el-ph matrixs' gauge be in consistent with nscf *SAVE file? if not ,what should I do to use wannier interpoltated dvscf and uninterpoltated nscf wave function to construct electron-phonon matrix?
Re: How to get the el-ph coupling matrix in complex number form?
Posted: Wed Dec 28, 2022 1:27 pm
by baiyunfei
I am quite interested in complex value of electron-phonon matrix in order to do exciton-phonon interactions.I have read your discussion about it. As far as I know, EPW use wannier interpoltation to interpolate dvscf and wavefunction on a finer mesh ,then use them to construct electron-phonon matrix. However wannier interplotation is known to bring some gauge problems.
My question is : if I properly modify the ''print_gkk in printing.f90.'' ; set the coarse mesh of nscf calculation the same as fine mesh of wannier interpolation , and get the output of complex electron-phonon matrixs, will the el-ph matrixs' gauge be in consistent with nscf *SAVE file? if not ,what should I do to use wannier interpoltated dvscf and uninterpoltated nscf wave function to construct electron-phonon matrix?