Neither operator overloading nor built-in numerical types are necessary (or arguably even useful) for scientific computing. Further, Python also lacks built-in vector/matrix types.