Upcoming Event: Oden Institute Seminar

CUR Approximation and Polynomial Interpolation: Connections and Applications

Yuji Nakatsukasa, Professor of Mathematics, University of Oxford

Abstract

Low-rank approximation and polynomial (e.g. Chebyshev) interpolation are pillars of numerical analysis. They are usually introduced as distinct tools. In this talk, we reveal a surprising bridge between them via the CUR approximation, by showing that polynomial interpolation can be seen as a CUR approximation of an infinity-by-infinity matrix. This observation generalises naturally to various settings and leads to new insight and algorithms, often resulting in dramatically faster computation. I will present several illustrating examples in model-order reduction and randomized algorithms in linear algebra.

Biography

Yuji Nakatsukasa is a Professor at Oxford University Mathematical Institute, specializing in numerical linear algebra. After earning his PhD from UC Davis in 2011, he held positions at Manchester, Tokyo, and the National Institute of Informatics, before his current post.