Difference Current-Current / Density-Density Picture

[Vasp1a]

Dear Vasp users,

I am using vasp 5.4.4 to calculate the dielectric functions for a multitude of compounds.
In the Outcar, I get the current-current "picture" and the density-density "picture" (as I call them), but I could not find out what the difference is (in principle), when to use it, what to be careful about. The difference I see in my results is that I get a spike at 0eV for the imaginary part of the dielectric function.

I was told everyone uses the density-density results, but with no explanation about the current-current one...

I would be happy if you could help me understand better how this works!

Thanks a lot and all the best,

Carl-Friedrich Schön, RWTH Aachen University

[alexey.tal]

Dear Carl-Friedrich Schön,

Here are a few papers that are discussing the difference between density-density and current-current response functions in great detail. That should give you a good overview of this topic and if you have some questions more specific to VASP, I would be happy to help you.

1. Sangalli et al., Phys. Rev. B 95, 155203 (2017)
2. Sangalli et al., Phys. Rev. B 86, 125139 (2012)
3. Kunes, Phys. Src. T109, 116 (2004)

[Vasp1a]

Dear Alexey,

thank you for referring me to these papers.
What I have understood from the first one is that the Current Model is better/more accurate if the sum rule is imposed:

"The current-based approach is instead exact at q = 0. In practice,
however, one needs to impose the conductivity sum rule in a
way which suppresses the intraband transitions at q = 0, thus
making the small q expansion needed for metalic systems also
in the current based approach."

Is the Sum Rule automatically included in vasp, or is this somethin that needs to be done after/manually?

Thanks and all the best,

Carl-Friedrich

[alexey.tal]

VASP does automatically enforce the conductivity sum rule in the calculation of the dielectric function via the current-current response function, so you don't need to do anything manually for that.