With Deep Reinforcement Learning, DeepMind has discovered an algorithm no human thought of. It is supposed to significantly accelerate matrix multiplication.

Multiplying two matrices is common in many applications and can be easily parallelized. First, lets look at a simple example of parallelizing a loop, in Fortran.

Let D ⊂ ℂ be a bounded domain, whose boundary B consists of k simple closed continuous curves, and H∞ (D) be the algebra of bounded analytic functions on D. We prove the matrix-valued corona theorem ...