Since Matlab is the reference implementation, it is crucial that it is correct. There should therefore be automated tests that can be run on CI. I propose using https://github.com/MOxUnit/MOxUnit (a MIT licensed unit test framework for Matlab/Octave) for this.
This would also make it possible to check which parts of the code work correctly with Octave.
Plus, having good tests would make it easier to create a Python API (or port parts of the code to Python) and be sure that the new code has identical behavior to the original Matlab code.
Since Matlab is the reference implementation, it is crucial that it is correct. There should therefore be automated tests that can be run on CI. I propose using https://github.com/MOxUnit/MOxUnit (a MIT licensed unit test framework for Matlab/Octave) for this.
This would also make it possible to check which parts of the code work correctly with Octave.
Plus, having good tests would make it easier to create a Python API (or port parts of the code to Python) and be sure that the new code has identical behavior to the original Matlab code.