Terms for SphericalGrid1D not implemented

[mhvwerts]

This is not very urgent. I am trying out PyFVTool and it seems that the 1D spherical mesh has not been implemented yet.

I was trying to port the example concerning 1D spherical diffusion from the original Matlab FVTool to PyFVTool, and found that the createMeshSpherical1D does not exist (yet).

For now, I will go play in a different coordinate system. Anyway, thanks for PyFVTool which is an interesting option for using the finite-volume method in teaching transport phenomena and for setting up toy(?) models for research.

[simulkade]

I spent my last hours of holidays on implementing most of the spherical terms for 1D and 3D in a separate branch. It still needs to be tested, e.g. with @mhvwerts own example that was previously done in FVTool here. Feel free to have a look @mhvwerts. Since it is -kind of- linked to an ongoing project of ours at DTU, I will try to spend some work hours on it this week.

[simulkade]

With the latest commit, all .py tests pass (I could not get the notebook tests to work on my machine; not a priority at the moment). The diffusion example seems to give reasonable results although not yet compared with an analytical solution. The visualization is also broken for 3D spherical grid. I am in favor of drawing several circular wedges either in parallel or perpendicular. However, for incomplete spheres it is not the best option.

[simulkade]

For now, one must avoid $\theta = \pi$ when creating a 3D spherical grid.

[mhvwerts]

It is great that you are implementing the terms for spherical coordinates (1D... and even 3D). I am eager to take a closer look and hope to do that somewhere in the coming days.

Indeed, the notebook tests are not easy to run and need a special conda environment to make pytest-notebook happy. I am planning to export the code from the notebooks to conventional test scripts so that they run with vanilla pytest. We would then simply keep the notebooks as documentation/examples and have separate test scripts running the same code. The notebook examples may then evolve, while the (more basic) test scripts can stay the same.

[mhvwerts]

There are some additional subtleties with notebook-based testing:

• It is necessary to run (and save) the final notebooks once by hand, when moving to a different system
• Rigorously avoid any notebook output that changes upon each run of the notebook. Typically, that may be memory addresses of objects that are automatically output (e.g. by plt.plot()). Ironically, these can be silenced by adding a trailing ";" after the last command in a cell!

This is quite a hassle, I realize. I will take care of occasional notebook testing for now, and keep in mind to create stand-alone conventional test scripts.

[mhvwerts]

I pushed some small changes that went directly to your "pyfvtool-spherical" branch. Don't forget to pull them in to your local branch. Later on, I will probably work again on a separate branch in my fork and file pull requests as necessary.

[mhvwerts]

Anyways, notebook testing is working for me now and all tests pass, so your new code also runs on my computer. I hope to get back to this soon (the coming days?).

[simulkade]

Thanks, @mhvwerts. Just pulled all your changes. I am struggling a bit with the boundary conditions at the north and south poles of the spherical grid. Otherwise, everything is fine.

[simulkade]

I also forgot that the range for $\theta$ is [0, $\pi$]. I fix the warning message.

[mhvwerts]

The 1D spherical diffusion example is working. There is a notebook illustrating this. Nice to see this working with PyFVTool!

[simulkade]

This looks great! I will clean up and make a pull request later today.