Numba reduced set

[samkitshah1262]

towards #544

[samkitshah1262]

@liadomide @maedoc could you please review the following PR. Moreover, can you please guide me on how to test the reproducibility of similar results in old and new code. Thanks!

[maedoc]

hi, thanks for your contribution. In fact, we cannot accept the current modifications because the existing NumPy code is the reference implementation. Instead, I would suggest approaching this by creating a separate Numba implementation, not modifying in-place the existing NumPy code. This would then allow you to test your Numba against the existing NumPy code.

[samkitshah1262]

@maedoc added back the numpy code. Can you please tell how can I carry out the testing ( I am quite new to TVB, apologies if it's something trivial). Thanks! :)

[maedoc]

sorry for late reply @samkitshah1262, you can write a function which calls both definitions and compares the results with numpy.testing.assert_allclose, checking that for different values of the states, coupling and parameters, the results are close enough.