Yambo Kerr

[michele guerrini]

Dear developers, I would need to perform optical simulations at RPA level resolved in polarization and then to access (or extract in a log file) the full dielectric tensor by using three orthogonal polarizations per each frequency. I know there is the routine Yambo_kerr but, since I have never used it before, I was wondering how should I proceed.

Many thanks!

Michele

[Davide Sangalli]

Dear Michele,
it depends a bit on what you need to compute.

OPT 1
yambo kerr is able to directly compute the off-diagonal matrix elements of the dielectric tensor.
But i is limited to the xy plane. It works as follow.
(i) prepare the input file for a calculation in eh space (yambo -o b) using yambo_kerr in place of yambo
(ii) uncomment the EvalKerr flag
(iii) specify the direction "i" along which you want to compute the dielectric function. It must lie in the xy plane
For "i=x"

Code: Select all

% BLongDir
 1.000000 | 0.000000 | 0.000000 |        # [BSS] [cc] Electric Field
%

yambo_kerr will give you in output eps_ii and eps_ij where "j" is the direction rotated by 90 degree (in the xy plane) compared to i.
Accordingly "i=x" then "j=y".

OPT 2
Alternatively you can do six standard calculations with yambo specifying each time a different direction and then reconstruct the whole dielectric tensor.
Yo will need (1,1,0), (1,0,1), (1,0,0), (0,1,1), (0,1,0) and (0,0,1). There are relations you can find in the literature.
Again the relevant variable is

Code: Select all

% BLongDir
 1.000000 | 0.000000 | 0.000000 |        # [BSS] [cc] Electric Field
%

In this second case you will obtain only the symmetric part of the off diagonal elements, while in the first there will be an eventual contribution from the anti-symmetric part in case it exist in your system (for example in presence of a magnetizaton and SOC or of an external magnetic field)

Best,
D.

[michele guerrini]

Dear Davide,
thank you very much for the advices. So, regarding Option 1 you are suggesting, if I would like to calculate the component eps_yz, I should choose to polarize the E field along i=y and for component eps_zx , along i=z ? Right ?

Bests

Michele

[Davide Sangalli]

Option one, for now, is limited to compute eps_xy basically.
So a way to get eps_yz without touching the code would be to rotate the atoms in the supercell, in such a way that "x" beomes "y" and "z" becomes "y" ...

Otherwise one could change the code. It is quite easy.
Just open the subroutine
src/pol_function/DIPOLE_project_along_q.F

Then "field_dir" is the direction defined in the input (what I called "i") and "dipole_dir" is the direction rotated by 90_degrees (what I called "j").
You can just manually change the two to get the ij component you prefer.
Pay attention to the "q0_def_norm" factor.