Anisotropic materials issues

[momchil-flex]

I am pretty sure most functionality with anisotropic materials will fail in the current formulation, because the return of eps_model() for those materials is different than the return of the same function for any other material (it returns a 3-tuple instead of a single complex number). For example, simulation.epsilon with a given component should select not only the specific locations, but also the correct component for an anisotropic material (something like this should be added as a test).

I see several possible ways to deal with this:

• Handle Anisotropic materials separately everywhere (seems cumbersome)
• Have all materials return a three-tuple from eps_model() (seems a bit redundant for many uses)
• Have eps_model() return the averaged epsilon for an anisostropic material. Introduce another function, e.g. eps_components() that returns a three-tuple for all materials, and use that when needed (seems best?)

[momchil-flex]

I can do that myself actually I guess I was thinking out loud here. Just let me know if option 3 sounds good to you.

[tylerflex]

#3 sounds good to me.

…

On Thu, Dec 9, 2021 at 1:57 PM momchil-flex ***@***.***> wrote: I can do that myself actually I guess I was thinking out loud here. Just let me know if option 3 sounds good to you. — You are receiving this because you were assigned. Reply to this email directly, view it on GitHub <#106 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AWDVXGDJ3ZISK5VS5V3ARSTUQE67ZANCNFSM5JXYYFGQ> .

[tylerflex]

Hey, I actually had a thought about this after the fact. I wonder if this makes sense:

Would it also make sense to overload the .eps_model() for anisotropic materials to return an (3, N) shaped array for N frequencies? (and just (3,) for 1 frequency)?

If we can ensure that the downstream functions that use .eps_model() are broadcasted correctly, I dont see a problem with this. What do you think? Too complicated maybe?

[momchil-flex]

I would say too complicated yeah. I already implemented a separate eps_diagonal() function. eps_model for an anisotropic material now returns the average value of the three components.

[tylerflex]

ok, that's fine for now. We can maybe warn that eps_model for anisotropic returns average and to use the tensor components' eps_model functions for more granularity (ie. anisotropic_med.xx.eps_model().