[Example] Axisymmetric initial-value problem (2D r - z plane)

[neuton]

Describe the example
A finite-time pulse propagation starting from r=0 in axisymmetric cylindrical coordinates.

Additional context
It's hard to figure out how to correctly setup the axisymmetric initial-value problem (2D).
The present examples do include the axisymmetric case but as far as I understand are limited to the continuously-driven source of some frequency. When trying the simplest setup by manually setting p0 the simulation seems to produce unexpected results. It is not clear where the 2D grid centre is located. Neither it is clear what happens for a non-rectangular grid. A clear example showing the minimal required configuration for a successful IVP would be invaluable.

[djps]

Thanks for the suggestion.

This gist is first pass at an example: it is a initial value probelm for the a given pressure p0 in a axisymmetric, non-rectangular setup with source as mask and sensors as cartesian grid points. I hope it helps. Note that it is a draft, as the output of the time signal from the fifth sensor is a bit larger than it should be and so the example code needs some modification.

[neuton]

Thanks for this example! This seems like a very useful example in general and it does clarify multiple confusions I had.

After running the simulation, I have a couple of questions though:

1. Is pml_size == 20 the default implicit simulation value?
2. Not sure how general of a problem this is, but when I simply set is_gpu_simulation=True --- turns out the produced simulation results are consistently wrong on Ubuntu with Nvidia L4 (haven't tested other GPUs yet) as compared to the CPU execution. I attach the CPU (1) vs GPU (2) plots.

[djps]

I don't think there is an axisymmetric code on the gpu - see here.

I had found that the cpu axisymmetric code was as quick as the equivalent 3d gpu code on my laptop.