Where does SciPy’s adaptive step size method for finite differences originate? Inside the KrylovJacobian class from SciPy, there is this method:
