Introduction
So far each function produced a single number. But a neural-network layer maps a vector to a vector, and a normalizing flow maps one distribution's samples to another's. The derivative of a vector-valued function is a matrix — the Jacobian — which is the best linear approximation of the function near a point.