NonlinearSolve.jl should be much faster than Roots.jl.

For an exponentially moving box average of the vector data x with box length 'n' implementation, check movingaverages. It seems like we don't need this often in space local coordinate transformations, so I will remove it for now.

There are still remaining scripts for plotting MHD-EPIC simulation outputs. They should be moved to a script folder under Batsrus.jl or somewhere else.

Precise control of colorbar position in Matplotlib is not an easy task. axis(“scaled”) or axis(“equal”) will cause issue with the present layout, such as overlapping, cutoff, or too much white spaces. Things are improving, but it takes time. The scripts for plotting PIC variables present some examples of controlling the layouts.

I've learned that Parquet is in general better than CSV, so we can consider supporting this new format. It won't matter so much for small datasets, but will be worthwhile for large data.

I need to implement the Shue model and maybe other empirical models for data comparison. Reference from SpacePy.

Since this pkg is not registered itself, and it relies on other unregistered pkg (e.g. TestParticle.jl), the document generation in CI failed. Currently I don't know what is a proper way to fix this; maybe just register TestParticle.jl?

This issue keeps being annoying. EmpiricalModeDecomposition.jl is even worse, because I am not the author of that package, and I cannot decide whether it should be registered or not.

Surprisingly, after fixing the matrix transpose issue with BATSRUS IDL outputs, the PC IDL output from FLEKS is still wrong.
I will try to fix it (again!).

The issue arises from the different input types of the contour function in Matplotlib. I used to have x,y,z all in matrix form: it works well. However, when I switched to x,y in vector form, the different in row/column-major between Python and Julia comes up and it is really tricky to remember the correct order. For some other plotting packages like Makie, only vector-like input for the positions are supported, which is why I made the decision to modify the APIs.

Currently this package sits in a very odd situation: in the actual module, it contains only the plotting functions for BATSRUS; in the space folder, there are many standalone scripts for handling different space physics related tasks. Technically this should not even be called a package.

A while ago I decided to split the Batsrus data processing part completely into a standalone package Batsrus.jl. Now it is a good time to consider splitting the rest into more modularized parts. In the future I also want to add more data processing methods related to space physics, so it would be better to collect them into something like Space.jl. Meanwhile, the visualization of Batsrus data should belong to Batsrus.jl!

Currently Julia does not have a clear way of letting one unregistered package depend on another unregistered package without first adding the other package manually. See the issue for details.

Now that with the new structure and package separation, this is no longer an issue.