Momentum Matrix elements

[kaynat_kalvi]

Hi,
I need to calculate the momentum matrix elements for my supercell.
I think getting a WAVEDAR file will do the job? but I am not sure would it include everything needed for using momentum matrix elements for optical properties?
I am confused in the potentials to use for getting a WAVEDER, as PAW doesnt include core of atoms and for momentum matrix elements and the optical properties we need cores? so should I use GW potential for calculating WAVEDER?
There are some post-processing tools to get (vasp2mat,vaspkit) who accurate are they to do this job ?

[manuel_engel1]

Hi,

You are correct. The WAVEDER file has the information you need. However, after looking at the documentation on the VASP wiki, it is unclear to me if this is still up-to-date. I will get back to you with a full answer once I have confirmation.

[manuel_engel1]

Hi again,

Sorry for the wait, but the situation is a bit more complicated than I originally thought. The WAVEDER file contains exactly what is written on the wiki. If that is what you want, meaning the derivative with respect to the Bloch vector, then you can use it. The derivative is already formulated in the PAW formalism, so it contains one-center terms. The only caveat is that the matrix elements corresponding to degenerate states, where \(\varepsilon{n \mathbf{k}} = \varepsilon{n' \mathbf{k}}\), are most likely meaningless. If they do contain values, those values should be ignored.

There are other ways to calculate this matrix in the code but none of them are currently accessible. If you have access to the code and are able to compile it, then you can perform a small modification that allows you to access a plane-text formatted version of the WAVEDER file called the WAVEDERF file, which should be easier to read. In linear_optics.F, there is a commented line

Code: Select all

!   CALL WRT_CDER_BETWEEN_STATES_FORMATTED(WDES,W,IO%IU0,IU=55)

that can be uncommented and will then produce the WAVEDERF file.

If the matrix elements of the WAVEDER file is not what you are interested in, for example if you require the derivative with respect to the spatial coordinate, then there is currently no easy way to obtaining this matrix in VASP. We are working on making these matrices more accessible in future versions of the code.

I hope that this brings a bit more clarity to the situation.

[kaynat_kalvi]

Hi

I have WAVEDERF file, please help me read that file as there is nothing I can find on Vasp Wiki. What does the rows and coulumns corresponds too exactly. I need to use these derivates of wavefunctions for post-processing.

Also, I can't compile vasp2kp, can you please help me in compilation of that.

thanks a million

[manuel_engel1]

The format of the WAVEDERF file is given as follows. First, a header line is written which contains 3 indices: the number of spin components, the number of k-points and the number of bands considered for constructing the matrix elements.
Each subsequent line then contains information pertaining one particular matrix element, namely the following values:

• Band index \(n'\) of the Bra (outgoing) state

• Kohn-Sham eigenvalue \(\varepsilon_{n' \mathbf{k}}\) corresponding to the Bra state

• Fermi-weight corresponding to the Bra state

• Band index \(n\) of the Ket (incoming) state

• Kohn-Sham eigenvalue \(\varepsilon_{n \mathbf{k}}\) corresponding to the Ket state

• Fermi-weight corresponding to the Ket state

• Three WAVEDER matrix elements corresponding to the three Cartesian directions, \(i\) (in x, y, z order).

Such a line is written for all combinations of spin, k-point and band indices in the following order:

• spin (slowest)

• k-point

• Band index \(n'\)

• band index \(n\) (fastest)

Let me know if this explanation is sufficient to read the WAVEDERF file. Since this file is not usually generated by VASP, the file format as described here might change in future versions.

Regarding vasp2kp, we have no experience with this tool and are not affiliated with it. Therefore, I cannot offer you any support on it. Please, contact the developers of vasp2kp to help you with compilation issues. Thanks for understanding.

[kaynat_kalvi]

Hi, Thanks very much.
I am experiencing a strange behaviour, I had an Ge 8 atom calculation with this incar,

Code: Select all

PREC = Normal ; ENCUT = 350
ISMEAR = 0 ; SIGMA = 0.01
EDIFF = 1.E-7
NBANDS = 70
PRECFOCK = Normal
#ISPIN=2
#WAVEDER file must be made:
LOPTICS = .TRUE.
LPEAD = .TRUE.
OMEGAMAX = 10

and it gives the WAVEDER AND WAVEDERF files having non-zero matrix elements, but I ran the same incar for Ge 2 atoms cell and now the WAVEDERF file has all zeros for the last 6 columns of each line, which means it all matrix elements are zero, which can't be happened.

Then I just certain paramters, like using LSORBIT, INCREASING OMEGAMAX TO 20, USING ISPIN=2, increasing and decreasing the number of bands, but it doesn't work,

SO I am wondering where Am I going wrong.

Plus, after ruuning this cal, I was geting real and imaginary part of dielectric constant for 8 atom cell, which looks reasonable, But for 2 atom cells these files are all zeros.
please help me out here.

[manuel_engel1]

Is the number of atoms the only thing you are changing? To get comparable results to the calculation with 8 atoms, you should increase the number of k-points accordingly in the calculation with 2 atoms. It could be related to other convergence parameters (like ENCUT), but it's difficult to say. I would try to change the k-points first. If that doesn't change your situation, you could choose stricter convergence criteria.

Let me know if that changes anything.

[kaynat_kalvi]

Yeah, I only changes the POSCAR.
jUST A test run cal, I rerun this for 8-atom Ge, and now it's not even generating the WAVEDER and WAVEDERF file, i can't understand why.

[kaynat_kalvi]

Hi, I have changed the ENCUT from 300 to 500 eV and k-mesh, but nothing is working out as I am still having all zeros, in the matrix elements columns of WAVEDERF file. Please suggest me more tests.

[manuel_engel1]

Alright, I'm currently out of ideas. Therefore, I would like to take a look at your calculations. Could you please upload the input and output files of the two different calculations following the minimal reproducible example page on the wiki?

[kaynat_kalvi]

Hi, As the file is too large i have to dell wavecar and some other files, hope this gives you an understanding, why am I getting all zeros in WAVEDERF last 6 columns.