ABINIT Discussion Forums

Dear all,

after doing all the steps needed to get the d_ij tensor with anaddb, I can also get in the same output the static dielectric tensor. Then I repeated the necessary steps including unoccupied bands to calculate the frequency dependent dielectric function as well as the chi_ijk components with optic.

I see that the value of the compontents of the dielectric tensor are different from the values I can read from the optic output of the real part of the dielectric function at omega=0; in particular I get these non-null components

Electronic dielectric tensor
epsilon_xx 5.28141441
epsilon_yy 4.73276442
epsilon_zz 5.14900599

Relaxed ion dielectric tensor
epsilon_xx 8.68334002
epsilon_yy 6.87470471
epsilon_zz 8.44220858

while from the real part of the dielectric function at omega=0 I can read these values:

epsilon_xx 5.794435
epsilon_yy 5.698954
epsilon_zz 5.821227

I was expecting they were at least close to, if not the same, the relaxed ion dielectric tensor.

Thanks in advance for your kind help

Best

Antonio

1) optic does not include phonons, just the electrons in RPA sum over states, so the low frequency limit is quite different from the correct one. In anaddb (DFPT Sternheimer equation for the electric field response) you include the phonons as well, but the electronic part of the dielectric response is taken to be frequency independent. What you should probably be able to compare is the low freq optic results with the epsilon_infinity in anaddb (infinity = much bigger than phonon energies and smaller than the electronic gap)

2) getting these two to be similar could be a bit tricky for convergence of the bands and k-points in optic.

Matthieu

OK. Many thanks for your reply!

Best

Antonio