momchil-flex

Yeah, this is definitely not nonsense, and has been making my life hard every now and then, since the beginning of Tidy3D. The approach we have under the hood is...

By the way @aisichenko note that the `outer_dot` has been extended to work for `FieldData` too in our `pre/2.3` branch. Related to that @flaport @Jan-David-Black note that the method now...

@weiliangjin2021 actually I think the issue is simply in our Green2008 material fit. What Pascal is showing is that it has much higher loss at that wavelength than the reference.

@weiliangjin2021 in 2.7, apart from still creating the two separate Green2008 models, let's plan to switch the default materials to "Palik_Lossless" for both cSi and SiO2. This should be done...

Yes, the other part of the solution would be to split Green2008 into a lossless or nearly lossless model for the range 1.2-1.45, and the current 5-pole lossy model which...

I think this is unlikely to be 2.7 right?

In fact looking at the solver, TFSF where the source surface crosses a custom medium would straight up not work, or the results may be wrong, even if the custom...

> What about checking if custom medium is uniform, if not, warn that it should be uniform on the plane? If we actually do the check, we should error if...

Is there a cheap way to check this? I guess something like `np.amin(arr) == np.amax(arr)`? But would this be slow for a large array?

That's true, I forgot amin/amax is not sorting but just O(N). I think this is fine then.