computing the extraction efficiency of slab

弄了好久，一直没有好的结果，sigh.

> In the post "How to calculate the extraction efficiency of spontaneous
> emission from slabs using C++ interface?" by Jun She, he constructed a
> geometry without slab and defined the flux through the top surface as total
> emitted flux, and then put a slab in the geometry and re-compute the flux
> through the top surface, and the extraction efficiency is the ratio of the
> later flux to the former flux. The computation looks like that for
> transmission spectrum. However, you have ever said that "...the same current
> source puts on a different amount of power depending on its surroundings..."
Exactly, but this difference is a real physical effect and directly influences the extraction efficiency. The reason is that the total power emitted by a dipole source at a point in a given structure is proportional to the local density of states at that point, and the local density of states in turn is proportional to the rate of spontaneous emission (by Fermi's Golden Rule). So, the ratio of the emitted flux spectra from a point current source between two structures (e.g. slab vs. no slab) is proportional to the ratio of the spontaneous emission extraction efficiency of those two structures as a function of frequency.
> In my intuitionistic opinion, the total flux and the flux through top
> surface should be computed in the same run and in the same geometry.
It all depends upon what you want to compute. If you want to compute the fraction of the power that goes upwards vs. in all directions, then yes you want to have a single structure and compute both the total flux and the upwards flux spectra and take the ratio. If you want to know how much a particular structure enhances spontaneous emission relative to another structure, including the effect of changes in the local density of states, then you want to take the ratio of the upwards ("extracted") flux spectra in the two different structures. And there might be other questions that you would answer in different ways. The point is, you should always keep in mind what reference you want to compare to when looking at flux spectra. Steven