Say I have a non-convex objective function loss that takes a np.ndarray named X whose shape is (n,) and returns a float n
loss
np.ndarray
X
float