Yifu ZhangMathematics

Research

Numerical analysis, randomized numerical linear algebra, approximation theory and operator learning.

Current position

First-year PhD

First-year Mathematics PhD student at Imperial College London, supervised by Nicolas Boullé, Greg Pavliotis, Matthew Colbrook.

Projects

Approximation theory · Numerical analysis

Function approximation under heteroskedastic noise

HeteroChebtrunc adapts Chebyshev-based approximation to the heteroskedastic setting, where noise variance depends on the sampling location. The paper gives a high-probability infinity-norm error comparison and studies sample-variance estimation under subgaussian noise.

Joint work with Yuji Nakatsukasa.

Randomized numerical linear algebra

CUR approximation for Gram matrices

My Oxford MMath dissertation studied approximation of Gram matrices given a column subset selection, applying techniques from CUR approximation and sketch-and-solve least-squares.

Supervised by Yuji Nakatsukasa.

Analytic and probabilistic number theory

Permutations and fixed sets

This project investigated how questions about permutations fixing sets of specified sizes connect with number-theoretic counting problems.

Supervised by Ofir Gorodetsky.

Conferences and workshops

  1. Upcoming · Workshop

    3rd Workshop on Gradient Flows for Sampling, Inference, and Learning

    170 Queen’s Gate, Imperial College London

    Poster presentation

  2. Attended · Workshop

    ENLARGED

    Emerging Numerical Linear Algebra Research Gathering for Exchanging Developments

    Rutherford Appleton Laboratory, Harwell

  3. Attended · Workshop

    OMDW01

    Operator approaches to dynamics – new connections

    Isaac Newton Institute, Cambridge