NumPy is Python's implementation of array-oriented computing techniques such as APL that emerged in the 1960s and have been used effectively to enable high-level descriptions of large-scale data-processing problems in science, engineering, and business. In this talk I will review the concept of array oriented computing and discuss why it is important and speculate as to why it and other central programming abstraction have been neglected in common computer languages. I will then provide a brief history of NumPy that will also describe its central features.

I also plan to discuss the future roadmap of NumPy and how some natural abstractions of its current implementation will enable high-level descriptions of common algorithms that can be seamlessly mapped across machines with many cores, GPUs, and even clusters of computers.