Implementation of Maxwell RV

[tvwenger]

Motivation for these changes

Following this discussion, I have submitted this PR to implement a random variable (RV) that follows a Maxwell distribution. The Maxwell distribution appears often in physical systems and having access to it in pymc and pytensor would be generally useful.

Implementation details

I have attempted to simply follow the style for other RV implementations that extend ScipyRandomVariable. I have also added tests to tests/tensor/random/test_basic.py, which is now running (slowly...).

I've added the relevant implementation and tests to numba and jax, but I would appreciate someone double checking what I've done since I was unsuccessful in running tests/link/numba/test_random.py, and tests/link/jax/test_random.py (the tests are skipped, probably because I don't have jax/numba set up in my environment).

Checklist

• Explain motivation and implementation 👆
• Make sure that the pre-commit linting/style checks pass.
• Link relevant issues, preferably in nice commit messages.
• The commits correspond to relevant logical changes. Note that if they don't, we will rewrite/rebase/squash the git history before merging.
• Are the changes covered by tests and docstrings?
• Fill out the short summary sections 👇

Major / Breaking Changes

• None

New features

• New MaxwellRV implementation

Bugfixes

• None

Documentation

• Included docstrings as necessary

Maintenance

• N/A

[tvwenger]

As I suspected, the numba and jax tests are failing. I didn't really know how to set up these tests, so any help would be appreciated!

[ricardoV94]

Hi @tvwenger, there's no need to split this in 2 libraries, both the RV and distribution should be implemented in pymc-experimental.

For more details in how to implement and test see this guide: https://www.pymc.io/projects/docs/en/latest/contributing/implementing_distribution.html A CustomDist sounds enough so that guide isn't needed

[tvwenger]

Related PR on pymc-experimental: pymc-devs/pymc-extras#239