Add a notebook that replicates the setup and results of the benchmark 4 of Jean-Francois Aubry et al paper

[d-lucena]

Describe the new feature or enhancement

A notebook that describes how to replicate the results as those in the benchmark 4 using NDK

Describe your proposed implementation

Show how to setup the scenario, setup material properties, transducer position and properties, and run the simulation that replicates the results in Jean-Francois Aubry, et al.. The results from NDK and from the paper are then presented together to show they match.

[charlesbmi]

For reference, linking code to compare results https://github.com/ucl-bug/transcranial-ultrasound-benchmarks and the paper's published modeling results here: https://zenodo.org/record/6020543

[charlesbmi]

Making a note here: the maximum pressure is different between the 2-D and 3-D versions. This might have to do with how the wave equation units work out in 2-D vs 3-D.

The 3-D simulation is substantially slower, however; a GPU helps get it to ~1 minute.

[charlesbmi]

Some initial explorations that are relevant:

• https://github.com/agencyenterprise/ndk-research/pull/5
• https://github.com/agencyenterprise/ndk-research/pull/8

[charlesbmi]

We will leave this as an in-repo Jupyter notebook with results already-run rather than on CI (which will be very slow).