My Community

Dear all,

I'm trying to evaluate a spatially resolved frequency dependent dielectric matrix during a slab calculation, as described
in section III in the supplemental information of the following paper:

https://doi.org/10.1103/PhysRevLett.118.237403

This quantity is useful to explain reflectance anisotropy spectroscopy measurements. Its evaluation involves a
real-space dependent filter function, which is applied to transition matrix elements calculated in the LOPTICS routine.

My main question now concerns the transition matrix elements:
Is it possible to evaluted and output the transition matrix elements (WAVEDER or WAVEDERF) in real-space, as shown in the
linked publications supplemental information, so that one can apply the said filter function and calculate a
spatially resolved frequency dependent dielectric matrix?

Best regards,
Max Großmann

Hi,

The paper you are referring to uses a localized basis set, but VASP is a plane-wave code, so currently it is no possible to calculate the transition matrix elements in real space. You can look into the paper M. Gajdoš, K. Hummer, G. Kresse, J. Furthmüller, F. Bechstedt, Phys. Rev. B 73, 045112 (2006) to see how the transition elements are calculated in VASP.

Hi,

Thank you for your quick responds!

Would it be possible to only use certain G-vectors in the calculation of the cell periodic part of the orbitals u_nk when calculating the transtion matrix elements,
that correspond to the desired region in real space? Than one would basically apply the filter function f(z) in reciprocal space, i.e. f(G_z), or am I missing something?

Best regards,
Max Großmann

I don't think that such a filter function is going to work in reciprocal space because the plane waves are delocalized, but one would need to work out the full expression first. Unfortunately, at the moment such calculation is not possible in VASP.