Fix driven solver SurfaceCurrent with no pre-existing excitation

[phdum-a]

Closes #616.

SurfaceCurrent can be used in the driven solver, but don't have any "Excitation" index. Instead they are added to all excitations. There is a bug where if there is no already existing excitation, this fails.

More generally, the Excitation interface was only designed for lumped or wave ports even through there are other times of excitations (like SurfaceCurrent, Dipole, etc) and this should be reworked to be more general and user friendly.

[Sbozzolo]

Do you maybe want to add some of the (very useful) content of #616 (comment) in the documentation?

[phdum-a]

Do you maybe want to add some of the (very useful) content of #616 (comment) in the documentation?

Happy to, but could you let me know what parts in particular? I think that everything I have written is already in the documentation in https://awslabs.github.io/palace/dev/guide/boundaries/ and https://awslabs.github.io/palace/dev/reference/

[Sbozzolo]

Do you maybe want to add some of the (very useful) content of #616 (comment) in the documentation?

Happy to, but could you let me know what parts in particular? I think that everything I have written is already in the documentation in https://awslabs.github.io/palace/dev/guide/boundaries/ and https://awslabs.github.io/palace/dev/reference/

I find that your comment provides insight on how the features should be used, not just a description of the math/interface (e.g, you describe the difference between LumpedPort and SurfaceCurrent from a modeling/physics perspective). I also don't see references to dissipation or even the different normalizations in the boundaries page.

[phdum-a]

So I want to close this PR as this is an actual bug fix and it is waiting on documentation improvements that will take a while.

I also don't see references to dissipation or even the different normalizations in the boundaries page.

It's there, if you know where to look and how to read the Palace docs....

From https://awslabs.github.io/palace/dev/guide/boundaries/#Lumped-and-wave-port-excitation

A lumped port applies a similar boundary condition to a surface impedance boundary, but takes on a special meaning for each simulation type.

For each port, the excitation is normalized to have unit incident power over the port boundary surface.

From https://awslabs.github.io/palace/dev/guide/boundaries/#Surface-current-excitation:

This option prescribes a unit source surface current excitation on the given boundary in order to excite the model. It does does not prescribe any boundary condition to the model and only affects the source term on the right hand side.

I know that this is hard to find and interpret. Making the documentation more didactic is work and I have not found a place to naturally put my comment from #616 with more rewrites. For now:

• I've added a note in [WIP] Update port description in reference.md #696 that we should document surfacecurrent v lumpedport differences in the reference doc.
• In the tutorial on the driven solver Add tutorial for driven solver (adaptive and uniform) #702, I have a small discussion mentioning conditioning effects.

[hughcars]

Lookahead failures are unrelated, tracked in #700