Krylov Methods are (nearly) Optimal for Low-Rank Approximation

Open in new window