Gradient-Free Optimization for Matrix functions

Sawyer Allen, Cash Cherry, Aidan Eck +2

We consider the task of optimizing smooth, possibly non-convex functions of a matrix variable given access only to directional derivatives rather than full gradients. This setting…