CurrentDetector and FFT progress

[0xDBFB7]

Hi!

This is a little more towards #9. I think FrequencyRoutines.FFT() might perhaps cover #1 when it's done (prerequisite).

You can merge #23 now if desired; this PR is also unfinished, you can review later if you want.

It's such a simple procedure, (apply a pulse, measure the voltage and current at each node, FFT both V and I, divide the two spectra to get Z-param, multiply and divide conjugates to get S-param), I'm just taking a while to implement it.

Several parts might not be what you're looking for, feel free to suggest API improvements:

• New functions only work on a single cell in the Detector. Haven't figured out how one should "sum" E-detector cells over a certain Block for S-param purposes (averaging?). Maybe relates to #25. CurrentDetector only reports one polarization, to be fixed.
• Some holdouts still formulated in terms that make sense in the microwave systems I've been working with (specifically, SoftArbitrary taking a "voltage" amplitude input, and the antenna test case), but less so for the optical domain that /fdtd/ seems to mostly be used for.
• Varying magnetic permeability has not been accounted for in the current detector, because I don't have any test cases for that and I'm not well-versed enough to figure it out from first principles :P

Thanks for your patience :)

[flaport]

This is amazing! I'll start reviewing this asap! (might take me a while though...)

Short answers to the points you brought up:

• I think that's indeed related to #25
• That's fine, we can work on that later.
• Varying magnetic permeability is pretty rare in purely optical materials and I would leave it to the people who actually need those features to implement that themselves ;-). That said, maybe you can check for a varying permeability before doing the calculations and raise a NotImplementedError or a warning if a varying permeability is present.

[flaport]

Hi @0xDBFB7 , thanks for this PR. What you've added so far looks good to me! I've merged it. If there are any new changes you'd like to add feel free to open a new PR 🙂

Also, would you be willing to add an example notebook going over the new features you've added?

[0xDBFB7]

@flaport Wonderful! Thanks for the review and ping. I'll work on a notebook.

[flaport]

Thanks, @0xDBFB7 ,

I also saw you left some comments in the PR with some good ideas! I'll leave a comment here if I implement any of them.

Currently implemented:

• [X] fp16 (numpy and torch) and fp128 (numpy only) support