Willem Diepeveen
Portrait
Logo Hedrick Assistant Adjunct Professor, Department of Mathematics, UCLA

I am a Hedrick Assistant Adjunct Professor in the Department of Mathematics at UCLA, hosted by Deanna Needell and Andrea Bertozzi, where I also work closely with Oscar Leong. Previously, I completed my PhD at University of Cambridge, supervised by Carola-Bibiane Schonlieb, and hold a double MSc from TU Delft and TU Berlin, and a double BSc from TU Delft.

My research bridges Riemannian geometry, machine learning, and signal processing, which I use to develop interpretable and theory-grounded machine learning approaches for inverse problems, manifold learning, and data-driven modeling in the sciences.

Beyond research, I enjoy teaching mathematics at UCLA, mentoring graduate and undergraduate students, and engaging in interdisciplinary collaborations. I am also an avid sailor, hiker, skier and snowboarder, and a classics and poetry enthusiast.

Google Scholar GitHub Twitter
Education
  • University of Cambridge
    University of Cambridge
    PhD Mathematics
    Thesis: Riemannian Geometry for Inverse Problems in Cryogenic Electron Microscopy
    Advisor: Carola-Bibiane Schonlieb
    Oct 2020 – Jun 2024
  • TU Delft & TU Berlin (double degree)
    TU Delft & TU Berlin (double degree)
    MSc Applied Mathematics & MSc Scientific Computing
    Sep 2018 – Sep 2020
  • TU Delft (double degree)
    TU Delft (double degree)
    BSc Applied Mathematics & BSc Applied Physics
    Sep 2014 – Aug 2017
Experience
  • University of California, Los Angeles
    University of California, Los Angeles
    Hedrick Assistant Adjunct Professor, Department of Mathematics
    Host: Deanna Needell and Andrea Bertozzi
    Jul 2024 - Present
  • Harvard University
    Harvard University
    Visiting Postdoctoral Fellow, John A. Paulson School of Engineering and Applied Sciences
    Host: Melanie Weber
    Jun 2025 - Sep 2025
  • Institute for Pure and Applied Mathematics
    Institute for Pure and Applied Mathematics
    Fellow, Long Program on "Computational Microscopy"
    Oct 2022 - Dec 2022
News
2026
New preprint out with Oscar on generative modeling and Riemannian manifold learning from corrupted data.
Jan 27
2025
Our paper with Joyce and Deanna curvature-corrected submanifold approximation has been accepted to Information and Inference.
Dec 06
New preprint out with Melanie on arguing for iso-Riemannian geometry as the language for optimization on learned manifolds.
Oct 24
Our paper with Jon and Andrea on interpretable latent dynamics modeling has been accepted to Applied Mathematics Letters.
Jul 23
For the summer, I joined Harvard University as a Visiting Postdoctoral Fellow in the Geometric Machine Learning Group.
Jul 01
New preprint out with Deanna on iso-Riemannian geometry for Riemannian manifold learning.
May 13
Our paper with Georgios, Zak and Carola on scaling up learned pullback geometry has been accepted to ICML 2025.
May 01
2024
I joined the Department of Mathematics at UCLA as Hedrick Assistant Adjunct Professor.
Jul 02
Selected Publications (view all )
Curvature Corrected Tangent Space-based Approximation of Manifold-valued Data

Willem Diepeveen, Joyce Chew, Deanna Needell

Information and Inference: A Journal of the IMA 2025

[Paper] [Code]

Curvature Corrected Tangent Space-based Approximation of Manifold-valued Data

Willem Diepeveen, Joyce Chew, Deanna Needell

Information and Inference: A Journal of the IMA 2025

[Paper] [Code]

Iso-Riemannian Optimization on Learned Data Manifolds

Willem Diepeveen, Melanie Weber

Preprint 2025

[Paper] [Code]

Iso-Riemannian Optimization on Learned Data Manifolds

Willem Diepeveen, Melanie Weber

Preprint 2025

[Paper] [Code]

Score-based Pullback Riemannian Geometry: Extracting the Data Manifold Geometry using Anisotropic Flows

Willem Diepeveen*, Georgios Batzolis*, Zakhar Shumaylov, Carola-Bibiane Schonlieb (* equal contribution)

ICML 2025

[Paper] [Code]

Score-based Pullback Riemannian Geometry: Extracting the Data Manifold Geometry using Anisotropic Flows

Willem Diepeveen*, Georgios Batzolis*, Zakhar Shumaylov, Carola-Bibiane Schonlieb (* equal contribution)

ICML 2025

[Paper] [Code]

Riemannian Geometry for Efficient Analysis of Protein Dynamics Data

Willem Diepeveen, Carlos Esteve-Yague, Jan Lellmann, Ozan Oktem, Carola-Bibiane Schonlieb

Proceedings of the National Academy of Sciences 2024

[Paper] [Code]

Riemannian Geometry for Efficient Analysis of Protein Dynamics Data

Willem Diepeveen, Carlos Esteve-Yague, Jan Lellmann, Ozan Oktem, Carola-Bibiane Schonlieb

Proceedings of the National Academy of Sciences 2024

[Paper] [Code]

All publications