if you never played with #Numpy, it can be rather magical, it is kind of not normal #Python, it is something else.
The two most impressive ideas for me are:
- broadcasting https://numpy.org/doc/stable/user/basics.broadcasting.html (even the garden variety arithmetic operations between an array and scalars)
- using an array of indices to another array and "masks" (an array of booleans that get you parts of another array)
From time to time I revisit Rougier's "From Python to Numpy"
https://www.labri.fr/perso/nrougier/from-python-to-numpy/






