A Type System for Multidimensional Arrays

Abstract

Python is an extremely popular programming language used by many companies and aca- demics around the world. Historically a slow language, it has gained popularity as an in- terface with high performance linear algebra, multidimensional array, and machine learning libraries. Initially designed as a strong, dynamically typed language, recent additions to Python have introduced a static type checker; however, the types do not provide enough in- formation to avoid many runtime errors when typing multidimensional array operations. In this thesis, we develop a type system and extension to Python’s type system that strength- ens the static guarantees of the type checker over multidimensional array operations while preserving the simplicity and ease of use that is core to Python’s spirit.