S4 deals with only linear electromagnetism, so Maxwell's equations are scale invariant. This means that in some sense there are no units. To think of it a different way, you can decide that all length numbers correspond to microns, then frequency is in units of c/microns, where c is the speed of light, and so on.
What is the coordinate system?
All spatial coordinates are cartesian. They are not in terms of the lattice vector basis.
Where is the unit cell located and do I need to define periodic copies of patterns?
This a tricky topic. If you use an FMM formuation which is based on analytic Fourier transforms (which includes the default settings), then all patterns are automatically periodized, and in some sense, the shape and location of the unit cell is irrelevant. For spatially sampled FMM formulations (those that rely on an FFT), the unit cell is the Wigner-Seitz cell centered at the origin. For these formulations you need to specify periodic copies. You should always check that the patterning is specified correctly using GetEpsilon() or OutputLayerPatternRealization().
What happens if I get a quantity "outside" the structure?
For example, if you specify a negative z coordinate, that corresponds to being within the first layer, since it is assumed the first layer extends infinitely towards negative infinity in the z direction. Similarly, if you specify a z coordinate past the total thickness of all layers, it is within the last layer. This is why you can specify zero thickness for the first and last layers.
How do I specify frequency dependent material parameters?
If you have a loop iterating over frequencies, then within the loop simply set new epsilon values for the materials. In order to incorporate values from tabulated data, you can use the Interpolator object and pass it a table of values. Large tables can be defined in separate files and included using the Lua dofile() function.
Why is the POVRay output wrong?
The POVRay feature is only to provide a starting point for writing an actual POVRay script. It is not guaranteed to be correct. You should not rely on this feature.
Should the 4 in S4 be superscripted in publications?
Although we would prefer that it be superscripted, it does not have to be. The preferred way to typeset it in LaTeX is: S$^4$.