Frequency domain solver: slab reflection

[dalarev]

Hi,

Hoping to get your thoughts on why a frequency-domain simulation calculating reflection/transmission off a low-index slab does not seem to be producing accurate results.

The notebook is here. Cell 4 shows a 2D plot of the model setup. Indeed it can be reduced to a 1D problem, but I wanted to test amp_func.

I run a tol sweep to confirm that the error is minimized as the tolerance decreases. This test PASSES, but the field distribution (output of cell 5 in the notebook) already looks incorrect: why is there variation along the horizontal?

Finally, in the last cell, a comparison between the simulated results and the Fresnel-equation-predicted ones shows a discrepancy in reflection. Never mind the > 1 transmission result...

Perhaps this particular model is not a good candidate for the frequency domain solver. Fair enough, but I decided to start with this as I tested the viability of the frequency-domain solver before attempting it on a more complex (and more resonant) structure.

Thanks in advance for your thoughts!

[dalarev]

Just a note that I was able to resolve a similar (but not exact) issue by turning off subpixel averaging.