Plenary Lecture 1

Per-Gunnar Martinsson (UT Austin)

Title:

Randomization in computational linear algebra

Abstract:

The talk will describe how ideas from random matrix theory can be leveraged to effectively, accurately, and reliably solve important problems that arise in data analytics and large scale matrix computations. We will focus in particular on accelerated techniques for computing low rank approximations to matrices. These techniques rely on randomized embeddings that reduce the effective dimensionality of intermediate steps in the computation. The resulting algorithms are particularly well suited for processing very large data sets on modern communication constrained hardware.

While the first part of the talk serves as an overview of the field, there will also be a discussion of more recent work on a posteriori error estimation, and on techniques for computing data sparse representations of structured matrices.