transition dipole

[xixi]

Dear All,

how to get the transition dipole moment in a specific transition (like HOMO to LUMO at Gamma),

Thanks !

SSC

National University Singapore

[myrta gruning]

Dear Xixi,

as far as I know the transition dipole matrix is not directly accessible.
It is saved in the ndb.dipole database in the SAVE directory.
To inspect the file you need to use ncdump (usage ncdump ndb.dipole in the SAVE directory).
ncdump is the bin directory of your netcdf installation.
(Note that it may not be in the standard path, so you may need to specify the path to the bin directory of your netcdf installation when invoking ncdump).
The structure of the file (that can be deduced from src/io/io_DIPOLES.F) is the following:
for each k point and direction (e.g. Gamma along x will be DIP_iR_k_0001_xyz_0001_spin_0001 =)
you have 2 times Nv (where Nv is the number of valence bands) matrices, as the real and imaginary part of the dipole are saved separately.
Note that only valence-conduction dipoles are computed, so you have zeros between valence-valnece states.

Let me know how it goes. In principle I had written a python utility to read the dipoles, but I should find it and test if it is still working.

Regards,
Myrta

[Flex]

Hello,

I am looking for the transition dipoles of one my calculations as well, the input is attached. I am looking for the first and second transitions around the gap (bands 18-19) at the K symmetry point.

I used ncdump to extract the dipoles database, a part of the file is attached as example (the full file is too big)

I partially understand the format : it is dip_ir_[k point]_[direction]_[spin]. So, I can find the K point.

What are the indexes in the brackets (:,:,:) ?

The first index is 0 or 1. So, imaginary and real part. Which one is which ? The real part is the number I'm looking for, isn't it ?

The second number goes from 0 to 17. I am not sure what that could be. Index of valence bands ?

The last value is 0-79, listed below. I guess polarisation bands.

Where do I find the 18-19 transition ? Would it be (0,17,18) ?

Is there a way to find indirect transitions for different K points ? (from gamma to K for example)

Finally, to obtain the value of the full transition dipole, do I have to sum over directions ? In fact, I want the square modulus, so that would be m_x^2+m_y^2+m_z^2, I guess.

Can you confirm me all of this ?

Thanks in advance

[Daniele Varsano]

Dear Thierry,
I would say that all what you argued it is correct.

Daniele