Tensors Come of Age: Why the AI Revolution will help HPC

This addresses the problem of improving software reliability and performance in high-performance computing for AI applications, though it appears incremental as it builds on existing mathematical concepts.

The paper tackles the challenge of automating tensor algorithms by introducing a framework based on A Mathematics of Arrays, Psi Calculus, and Unum Arithmetic, resulting in software that is mechanically provable, scalable, portable, and more numerically accurate.