Open questions/known issues with current implementation of direct-hybrid method for aeroacoustics

[lchristm]

This whole process seems to be overly convoluted to me. It's a good prototype that works, so I wouldn't change it now, but it seems to me like we are abusing a lot of the OrdinaryDiffEq machinery here. We should somehow find another way to make this work in a cleaner fashion.

If not already done, can you maybe start an issue with open questions and known issues with this implementation, such that we can collect them in one place? First item from my side:

• Find a cleaner way to implement the coupled Euler-Acoustics evolution than manually using OrdinaryDiffEq internals

_Originally posted by @sloede in https://github.com/trixi-framework/Trixi.jl/pull/712#discussion_r669669399_

List of open questions and known issues:

• Find a cleaner way to implement the coupled Euler-Acoustics evolution than manually using OrdinaryDiffEq internals
• .h5 files generated by AveragingCallback are currently not compatible with Trixi2Vtk (discussed here)

[DanielDoehring]

FYI:

I recently adapted this to my custom integrators and did not find this too difficult to understand, nor too ugly implemented.

The only thing I removed was this bit of code

https://github.com/trixi-framework/Trixi.jl/blob/abf8a08bd40fbf894b4cb6bfd62fe17f448a6bca/src/callbacks_step/euler_acoustics_coupling.jl#L190-L194

Of course, performing a step! within a callback is somewhat hacky and probably not intended, but this worked out quite well in my case.

[sloede]

Interesting, thanks a lot for your feedback!