Yambo Community Forum

Dear developer,

I was trying to use yambo_kerr to calculate the full dielectric tensor.

I first tested it on iron to reproduce the result in the paper PRB 86, 125139 (2012).

I can correctly obtain the diagonal part, i.e. the Fig1(a) and Fig1(b).
However, for the off-diagonal term, neither the order or the pattern, match the Fig2(a) and Fig2(b).

I use quantum espresso pwscf with spin-orbital coupling turned on and the full relativistic norm-conserving pp from pseudo dojo.
14x14x14 k-grid, 130.0 Ry cutoff used

I used latest version 4.5.0 yambo_kerr with flag -o b to generate the input file.
I used 3000 RL plane wave and 32 bands, both for which I have checked the convergence.

Part of the input file is in following:
FFTGvecs= 3000 RL
BSEmod= "retarded" # [BSE] resonant/retarded/coupling
BSKmod= "IP" # [BSE] IP/Hartree/HF/ALDA/SEX
Lkind= "Lbar" # [BSE] Lbar (default) / full
Gauge= "velocity" # [BSE] Gauge (length|velocity)
EvalKerr # [BSE] Compute the Kerr effect
DrudeWBS= ( 4.9 , 5.652 ) eV (for diagonal)
DrudeWBS= ( 0.0 , 0.0 ) eV (for off-diagonal)


I found no where to modify.
Can you point out what I may miss?

Or is there new version of tutorial to follow?

Thanks.

Best,
Hsiao-Yi

Ciao Hsiao-Yi,
sorry, I saw your message few days ago but then I forgot to reply.

Can you send here:
- pw input files and pseudo potentials
- yambo input files
- yambo report and output files

I'll check everything and let you know.

Best,
D.

Ciao Hsiao-Yi,
I had a look to the files you sent me.

I think the reason for the discrepancy is that you put the magnetization in the xy plane. Instead in the tutorial for the kerr the magnetization is along the z-axis as in the standard polar geometry for kerr.
The latter is actually the only one implemented, or better yambo, so far, is able to compute only eps_ij and eps_ij with both "i" and "j" axis in the xy plane.

Hope it helps

D.

Ciao Davide,

Thanks. It works now.

One last check, if I want to use yambo_kerr to calculate the off-diagonal dielectric tensor for non-magnetization semiconductor,
do I need to specify this angel1(1)=0 as well?

Further, if so, then if I want to calculate for other direction, besides modifying the routine "src/pol_function/DIPOLE_project_along_q.F",
I need to re-run pwscf with different angel1(1), right?

Thanks again.

Best,
Hsiao-Yi

One last check, if I want to use yambo_kerr to calculate the off-diagonal dielectric tensor for non-magnetization semiconductor,
do I need to specify this angel1(1)=0 as well?

You don't.
That is the angle which decides the initial direction of the magnetization.
In a non magnetic material it is meaningless.

Further, if so, then if I want to calculate for other direction, besides modifying the routine "src/pol_function/DIPOLE_project_along_q.F",
I need to re-run pwscf with different angel1(1), right?

It depends on what you want and what they have experimentally.
Given the ground state of your material, changing src/pol_function/DIPOLE_project_along_q.F you can get all the components of the dielectric tensor.
Then if you change the ground state (i.e. you choose one with a different direction of the magnetization), the dielectric tensor in general will change and you need to recompute everything.

D.

P.S.: in the next release DIPOLE_project_along_q.F will disappear and the direction will be chosen on the fly inside the BSE subroutines.
It will thus be easier to get different components of the dielectric tensor