circular dichroism

[malwi]

Dear Yambo Team,

how to set input for circular dichroism?
I tried just to use BSE with BSKmod= "IP" and BSEprop= "dichr"
and it stopped before the end.

Is something published using this piece of code?

With best regards,
Gosia

[Daniele Varsano]

Dear Gosia,
can you post your input/report/log files in order to see where the codes does stop?

Best,
Daniele

[malwi]

Thank you very much Daniele. I attach the report file.

[Daniele Varsano]

Dear Gosia,
it would be useful to look also at the input file and one of the log files to see where it stops.
Also see if in one of the log files there is any error message.
You can upload the log and input without converting in pdf by changing the nell of the file with one of the allowed suffix e.g. .txt

Daniele

[malwi]

Dear Daniele,
there is an error in some of the log files.
These errors refer to reading dipoles - I attach the error file.

My procedure was two step:

[1] epsilon (input below) with writing to dir "bse-2"

Code: Select all

rim_cut                          # [R] Coulomb potential
em1s                           # [R Xs] Static Inverse Dielectric Matrix
dipoles                        # [R   ] Compute the dipoles
Chimod= "HARTREE"              # [X] IP/Hartree/ALDA/LRC/PF/BSfxc
CUTGeo = "ws xy"
CUTwsGvec= 0.700000
% BndsRnXs
    1 | 600 |                 # [Xs] Polarization function bands
%
NGsBlkXs= 6 Ry                # [Xs] Response block size
% DmRngeXs
  0.10000 |  0.10000 | eV     # [Xs] Damping range
%
% LongDrXs
 1.000000 | 1.000000 | 0.000000 |        # [Xs] [cc] Electric Field
%

2. BSE kernel with "dichr" instead of BSE, reading data from dir "bse-2"
(input attached)

Thanks a lot!

[Daniele Varsano]

Dear Gosia,

the error can be probably be solved by removing the ndb.dipol* files and let yambo recalculate them.
Next, as I already told you, the circular dichroism in the GPL release can be calculated only at in independent-particle level "IP", the CD in BSE framework
it is present in a developer's version and it is actually in phase of testing.

Best,
Daniele

[malwi]

Dear Daniele,
Thank you a lot!
CD works and looks similar to the experiment when I play with a number of bands and electric field.

Just a few more questions:
1. You said that it is only for the case of Gamma, yes?
When I give only the first k-point or all k-points, results do not change.
But the system is weakly interacting (almost dissociated), thus it might be a good approximation.

2. It would be good to have the equations implemented in this calculation.

3. There is also an anisotropy factor in the experiment.
It is defined as g = (I_L − I_R )/(I_L + I_R )^1/2 with left and right polarized light.

I attach input/output files for your check. There is still an issue of the energetic shift
of the CD spectrum - like it was in the BSE spectrum (taken from GW or hyb-DFT).
Actually, after the hyb-DFT calculations, the experimental truth is between hyb-DFT and GW.

Best regards,
Gosia

[Daniele Varsano]

Dear Gosia,

1) Yes, I think that it is implemented for gamma only, actually I do not know what the code does in presence of a k point sampling. Someone else could answer on that
2) If I'm not wrong it is the trace of the equation 2.13 you can find in the paper D. Varsano et al Phys. Chem. Chem. Phys., 2009, 11, 4481–4489
3) What is I here, the absorption? so it is a frequency dependent quantity? If so, the numerator is the CD itself, but you do not have access to I_R and I_L separately.

About the shift, take into account the you are in independent particle framework, so excitonic effect are not present. Such effects should shift your CD spectra to lower energies.

@Davide, can you confirm that?

Best,

Daniele

[malwi]

Thank you Daniele.

I am just thinking:
if we need a sum of absorption/emission with right and left polarization (I_R+I_L),
would it be proper to take just the absorption with the same approximation (to replace dichr by abs in the same input)?

[Daniele Varsano]

Dear Gosia,
yes the absorption mediated on the three cartesian direction should be the quantity you need.

Bsst,
Daniele