Build M-series (arm64) macOS wheels on GitHub Actions hosted runners

[agriyakhetarpal]

Description

GitHub Actions M1 runners are free for open-source repositories now – I have configured the PyPI workflow to add a job for these runners.

Important

The wheels can be tested on M-series hardware via the following link: https://github.com/pybamm-team/PyBaMM/actions/runs/7721068374/

Closes #3772, related to #3462

Type of change

Please add a line in the relevant section of CHANGELOG.md to document the change (include PR #) - note reverse order of PR #s. If necessary, also add to the list of breaking changes.

• New feature (non-breaking change which adds functionality)
• Optimization (back-end change that speeds up the code)
• Bug fix (non-breaking change which fixes an issue)

Key checklist:

• No style issues: $ pre-commit run (or $ nox -s pre-commit) (see CONTRIBUTING.md for how to set this up to run automatically when committing locally, in just two lines of code)
• All tests pass: $ python run-tests.py --all (or $ nox -s tests)
• The documentation builds: $ python run-tests.py --doctest (or $ nox -s doctests)

You can run integration tests, unit tests, and doctests together at once, using $ python run-tests.py --quick (or $ nox -s quick).

Further checks:

• Code is commented, particularly in hard-to-understand areas
• Tests added that prove fix is effective or that feature works

[agriyakhetarpal]

Note to self:

• Look into if archflags configuration is needed for CMake

Not needed because we are not cross-compiling at this time. CMake sets the variable CMAKE_CROSSCOMPILING to ON and cibuildwheel sets archflags as required.

• Add CHANGELOG entry (af9c299)

• Investigate whether to build universal2 wheel via delocate-fuse, delocate-addplat, or similar or to keep macOS wheels architecture-specific

We don't need dual-architecture wheels – pip prefers an architecture-specific binary distribution, which shall save both disk space and bandwidth for users shall we distribute it as such. The applications of universal2 binaries are more oriented towards .dmg or .pkg archives, or towards Python-based standalone applications such as those compiled by PyInstaller or Nuitka.

[agriyakhetarpal]

We should start testing PyBaMM on M1 runners as well if we are going to be distributing them – I can do that in a follow-up PR, though I am not sure if #3140 (particle cracking submodels) has been resolved or not.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (063a383) 99.59% compared to head (310fbbf) 99.59%.

Additional details and impacted files

@@           Coverage Diff            @@
##           develop    #3789   +/-   ##
========================================
  Coverage    99.59%   99.59%           
========================================
  Files          257      257           
  Lines        20802    20802           
========================================
  Hits         20718    20718           
  Misses          84       84           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[Saransh-cpp]

Thanks, @agriyakhetarpal! This looks really good!

[Saransh-cpp]

Also, all green ticks in this PR, nice!

[kratman]

I am not sure if #3140 (particle cracking submodels) has been resolved or not.

I put in a stop-gap fix for that ticket. I did not mark it as closed because all I did was adjust the grid. Not really a full fix or investigation. Having tests that would alert us sooner would probably make fixing stuff like that more visible and we could discuss closing that issue if we start testing more thoroughly on M-series hardware.

[agriyakhetarpal]

Having tests that would alert us sooner would probably make fixing stuff like that more visible and we could discuss closing that issue if we start testing more thoroughly on M-series hardware.

Sure, happy to merge this after you can download and test a wheel or two (I tested them, they are working for me since the IDAKLUSolver can be instantiated and can be used to solve a basic Simulation – even though that seems to not be a reliable test, see #3783). I have set up testing for M1 in #3791.

[kratman]

@agriyakhetarpal I downloaded the wheels to test them. I think they are working

pybamm.have_idaklu()
True