electron-phonon matrix elements are not hermitian

[simba2828]

Dear experts,

From my EPW calculation, the electron-phonon matrix elements at q = 0 are not Hermitian, i.e,

g_{n,m}^\nu ≠ g_{m,n}^\nu*

Could anyone please tell the possible reason?

I am using EPW6.0 where I can print the complex electron-phonon matrix elements.

Sincerely,
Shubham

[Shashi]

Dear Shubham,

You may want to check the behavior at a finite q-point instead of Gamma, since q = 0 can sometimes be special (e.g., degeneracies or phase/gauge issues) that may affect the apparent Hermiticity.

Also, does your system preserve time-reversal symmetry? This can influence the symmetry properties of the electron–phonon matrix elements.

In addition, please check the quality of your Wannierization (e.g., spreads, band interpolation), as poor Wannierization can also lead to such inconsistencies.

It would be useful to verify these points and see if the issue persists away from gamma.

Best regards,
Shashi

[simba2828]

Dear Shashi,

Thank you for your reply. Yes the system is TRS preserved. I haven't checked the q\=0 case but will definitely check.

Meanwhile I am trying to print the el-ph matrix elements from PHonon routine of ph.x. Is there any way I can print the matrix elements for given k points. I am using the following ph.in file but I can print only one k point at a time. Is there any tag from which I can print all the k points together.
Currently I am doing nscf calculation for some k points and then using the following input I can print el-ph matrix elements individually.

Code: Select all

&inputph
  tr2_ph=1.0d-14
  prefix=aiida'
  fildvscf = 'dvscf'
  amass(1)= 118.71
  amass(2)= 78.96
  outdir='./out/'
  fildyn=aiida.dyn'
  electron_phonon = 'prt'
  el_ph_sigma=0.04
  el_ph_nsigma = 25
  ldisp = .true.
  nq1 = 5
  nq2 = 5
  nq3 = 1
  start_q =1
  last_q =1
  trans = .false.
  kx = -0.5000000
  ky = -0.4885337
  kz = 0.0000000
 /

I also tried nk1, nk2, nk3 for isotropic grid but that too did not work. ALso, the above input file, read only 2pi/alat kpoints and not the crystal coordinates.

Please let me know if there is any functionality through which I can print the matrix elements in one go.


Thanks very much.

Regards,
Shubham