Angled plane wave sources incompatible with absorbing boundaries - workaround? Plans to address this?

[samuelkim16]

If I try to use a angled plane wave source with absorbing boundaries in the lateral dimensions (e.g. propagation in z, absorbing in x or y), I get the following error:

ValidationError: 1 validation error for Simulation
boundary_spec
  Angled plane wave sources are not compatible with the absorbing boundary along dimension 0. Either set the source ``angle_theta`` to ``0``, or use Bloch boundaries that match the source angle. (type=value_error.setup)

Is there a workaround for this? The couple I can think of:

• TFSF source - this technically works, but does not have a fixed angled mode making it cumbersome/impossible to do broadband simulations.
• Bloch/periodic boundaries for plane wave and move the boundaries far away from my structure - increases simulation cost; I am interested in far field patterns, but this would also only allow scattering at diffraction orders

Are there plans to add this feature into Tidy3D?

[momchil-flex]

There are currently no plans to expand the existing functionalities.

• TFSF source - this technically works, but does not have a fixed angled mode making it cumbersome/impossible to do broadband simulations.

In general, the best solution would be broadband fixed-angle support in the TFSF source. However, this is somewhat involved, and not really on our to-do list right now (maybe some day in the future). However, I would say that it it not impossible, and maybe only a bit cumbersome to do broadband simulations by sticking several fixed-k simulations with several different center frequencies. I think just a few of those would give you pretty good results, because each is close to accurate for some small bandwidth around the center. This might even be cheaper to do in some cases, because broadband fixed angle simulations can require a significant reduction of the time step to be stable.

• Bloch/periodic boundaries for plane wave and move the boundaries far away from my structure - increases simulation cost; I am interested in far field patterns, but this would also only allow scattering at diffraction orders

I would say, the other option to consider along these lines is a GaussianBeam source with absorbing boundaries. Then you can use the field projection monitors. Of course it is not equivalent to plane-wave illumination, but depending on what you're trying to do, you could get useful results.