yambo_kerr

[8813204602]

Dear Yambo community forum

So far, I have calculated the diagonal components of the polarizability matrix for a 2D magnetic semiconductor using BSE calculation, and now I want to calculate the xy component. According to the forum posts, I found out that off-diagonal components should be calculated with yambo_kerr, but nevertheless, I did not find tutorial about this on the Yambo site. I would be very grateful if you could tell me that the following procedure for doing this calculation is correct:

1- Creating an input file with the command: yambo_kerr -o b -k sex -r -y h -V all -F kerr.in -J bse
(bse is the name of the folder in which the databases from the screening calculation are located)
2- Activating Evalkerr flag
3- Setting parameter BLongDir to form:
BLongDir
1.000000 | 0.000000 | 0.000000 |
4- Running yambo_kerr with the command: yambo_kerr -F kerr.in -J bse


I have some other questions in this regard,also:
1) Given that my material is a 2D magnetic semiconductor, is it better to set the Gauge parameter in input "length" or "velocity"?
2) To calculate the diagonal components, I have performed convergence tests related to the BSENGBlk, BSENGexx,BSEBands and k-point parameters completely. Can the same convergent values ​​be used to calculate off-diagonal elements and kerr rotation, or is there a need for re-convergence?
3) According to the parameter BLongDir: 1 | 0 | 0 |, the electric field is in the x direction, so is it correct to say that the incident light is in the z direction?



Thanks and Regards,
Mitra Helmi

[Davide Sangalli]

1- Creating an input file with the command: yambo_kerr -o b -k sex -r -y h -V all -F kerr.in -J bse
(bse is the name of the folder in which the databases from the screening calculation are located)
2- Activating Evalkerr flag
3- Setting parameter BLongDir to form:
BLongDir
1.000000 | 0.000000 | 0.000000 |
4- Running yambo_kerr with the command: yambo_kerr -F kerr.in -J bse

Correct. However notice that, since version 5.0, you do not need to use yambo_kerr anymore, just the standard yambo executable.
I'd suggest to use yambo 5.0, since previous versions had some issues in the evaluation of the kerr parameters.
Moreover in 5.0, instead of the logical "Evalkerr", the following input variable can be used

Code: Select all

BSEprop="abs kerr"

1) The length gauge should work fine if your material has zere Cern number.
2) In general the parameters are fine. Just notice that the off-diagonal components are smaller and in general a bit more sensitive to small errors
3) Yeah, the kerr is always assumed in polar geometry. You need to set BLongDir in the xy plane to defined a first direction "d1"
A second direction "d2" will be defined performing a rotation of 90° in the xy plane. Then eps_{d1,d2} is computed
Light is always assumed to propagate along the z direction.

Also remember that:
- you need a ground state with SOC included
- if, without SOC, eps_xy is already different from zero, then you need to subtract the structural term or symmetric part, eps^S_xy=(eps_xy+eps_yx)*0.5 to get the anti-symmetric term only eps^A_xy=(eps_xy-eps_yx)*0.5
In the future we are going to code a direct calculation of eps^A_xy=(eps_xy-eps_yx)*0.5, but for not you get eps_xy=eps^S_xy+eps^A_xy

Best,
D.

[8813204602]

Dear Davide,

So far, I have calculated the xx and xy components of polarizability for a 2D magnetic material by adjusting the BSEprop= "abs kerr" and BlongDir:1 | 0 | 0 | at the input file.
I have some questions in this regard:
1) In the output of this calculation, in addition to the two files with the alpha and beta names, which represent the xx and xy components of polarizability, respectively, files with the off and jdos names are also created. Can you please tell me what parameters these files specify and what diagrams can be plotted through them?
2) Is the polarizability yx component calculated by setting the BlongDir:0 | 1 | 0 | parameter?

if, without SOC, eps_xy is already different from zero, then you need to subtract the structural term or symmetric part, eps^S_xy=(eps_xy+eps_yx)*0.5 to get the anti-symmetric term only eps^A_xy=(eps_xy-eps_yx)*0.5
In the future we are going to code a direct calculation of eps^A_xy=(eps_xy-eps_yx)*0.5, but for not you get eps_xy=eps^S_xy+eps^A_xy

3) Do you mean that I have to calculate both the xy and yx components of polarizability with SOC, if alpha(xy) = -alpha(yx), the same alpha(xy) can be reported as the final result, but if alpha(xy) ≠ -alpha(yx), alpha(xy)-[(alpha(xy)+alpha(yx))*0.5] must be reported as the final result of the calculation?
4) Is the polarizability reported in the alpha and beta outputs in terms of Bohr?

Thanks and Regards,
Mitra

[8813204602]

Dear Davide,

l would be very grateful if I could use from your answers to my questions.

Best,
Mitra

[Davide Sangalli]

1) In the output of this calculation, in addition to the two files with the alpha and beta names, which represent the xx and xy components of polarizability, respectively, files with the off and jdos names are also created. Can you please tell me what parameters these files specify and what diagrams can be plotted through them?

epsilon and alpha are the diagonal terms dielectric function and the polarizability (xx), while off and beta are the off-diagonal terms of the two (xy)

2) Is the polarizability yx component calculated by setting the BlongDir:0 | 1 | 0 | parameter?

If you set

Code: Select all

% BlongDir
 0 | 1 | 0 |
%

you will get the "yy" and the "y,-x" components
To get the "yx" component you need to change the code (see next point)

3) Do you mean that I have to calculate both the xy and yx components of polarizability with SOC, if alpha(xy) = -alpha(yx), the same alpha(xy) can be reported as the final result, but if alpha(xy) ≠ -alpha(yx), alpha(xy)-[(alpha(xy)+alpha(yx))*0.5] must be reported as the final result of the calculation?

Yeah, more or less. Notice that, at present, there is no easy way to compute the "yx" components. The only solution is to change the code.
Open the subroutine

Code: Select all

src/bseK_IP.F

and change line 153 putting a plus in front of pi/2._SP

Code: Select all

-   if(l_BS_kerr) call rot_v_xy_plane(-pi/2._SP,field_dir(:,1),v_out=field_dir(:,2))
+   if(l_BS_kerr) call rot_v_xy_plane(+pi/2._SP,field_dir(:,1),v_out=field_dir(:,2))

After that setting

Code: Select all

% BlongDir
 0 | 1 | 0 |
%

in input you will get the "yx" component

Code: Select all

 4) Is the polarizability reported in the alpha and beta outputs in terms of Bohr?

It is in atomic units.

[8813204602]

Dear Davide,

Thank you very much for your reply and help.


Best,
Mitra