Yu Wang

Yu Wang

Ph.D., MIT EECS & CSAIL

Yu Wang

Computer Science and Artificial Intelligence Laboratory
Electrical Engineering and Computer Science
Massachusetts Institute of Technology
Harvard Medical School

E-mail: wangyu9 at mit.edu
Address: D475A, 32 Vassar St, Cambridge, MA 02139

This is the webpage of Yu Wang and hosted on GitHub pages.

About Me

I received my Ph.D. in Electrical Engineering & Computer Science from MIT with PhD minors in Mathematics and in Statistics, where I was a member of the Geometric Data Processing (GDP) Group in the MIT CS & AI Lab. I am currently a Research Fellow in the Lab for Computational Neuroimaging at Harvard Medical School.

Honorably mentioned by 2024 ACM SIGGRAPH Doctoral Dissertation Award, my Ph.D. thesis titled “Geometric Computing beyond the Laplacian” is in the area of Geometric Computing, where we design efficient algorithms to process, analyze, and optimize shapes or geometric objects. My research specifically bridges Applied Mathematics and Visual Computing, computationally leveraging the intricate interplay between geometry/topology and PDEs (partial differential equations) to design efficient algorithms.

Research Areas & Interests

Graphics & Geometric Computing: Geometry Processing, Shape Analysis, Vision & Imaging, Applied Mathematics & Geometry: Applied Geometric Analysis, Partial Differential Equations (PDEs), Computational Physics, Discrete & Computational Geometry, Inverse Problems & Optimal Control, Medical Imaging, Geometric Deep Learning, Scientific Machine Learning & Data-driven PDE, Numerical Algorithms, Convex Optimization, Differentiable Programming, Physical Simulation.

Representative Work

  • Geometric Computing beyond the Laplacian.
    Ph.D. Thesis, MIT, 2023. ACM SIGGRAPH Doctoral Dissertation Honorable Mention. PDF
    Joint work with Vladimir Kim, Michael Bronstein, Justin Solomon, Mirela Ben-Chen, Iosif Polterovich, Minghao Guo. Committee: Justin Solomon, Frédo Durand, Laurent Demanet (Math), Steven Gortler (Harvard). Qualify: Suvrit Sra and Erik Demaine.

  • Variational Quasi-Harmonic Maps for Computing Diffeomorphisms.
    Yu Wang, Minghao Guo, Justin Solomon. ACM Trans. on Graph. 42(4). Paper Dropbox.
    In short, we show and computationally leverage that
    Injectivity = {Quasi-Harmonic} + {Dirichlet & Neumann Boundary}

Summary: My approach to solving geometry problems incorporates ways to computationally leverage the intricate interplay between geometry/topology and PDEs (partial differential equations) to design efficient algorithms. My research draws on the rich synergy among differential geometry, topology, geometric PDE analysis, inverse problems and optimal control of PDEs, geometric graph theory, spectral geometry, harmonic and complex analysis, conformal/global geometry, discrete structures, and convex optimization.

These are vibrant subfields of mathematics whose successes I strive to translate into advances in computer graphics, geometry processing, medical imaging, and computational sciences.

Selected Publications

  • Schwarz–Schur Involution and Dirichlet-to-Neumann Condensing: Partial Differential Equations and Sparse Linear Systems Solved 1000x Faster.
    Yu Wang, Mazdak Abulnaga, Yaël Balbastre, Bruce Fischl.
    Submitted.

  • Variational Quasi-Harmonic Maps for Computing Diffeomorphisms.
    Yu Wang, Minghao Guo, Justin Solomon.
    ACM Trans. on Graph. 42(4). ACM SIGGRAPH 2023 (Journal Track). OpenAccessPaper Dropbox 26 pages. Code (Email me for access).

  • Fast Quasi-Harmonic Weights for Geometric Data Interpolation.
    Yu Wang and Justin Solomon.
    ACM Trans. on Graph. 40(4). ACM SIGGRAPH 2021. OpenAccessPaper. 15 pages. Code

  • Intrinsic and Extrinsic Operators for Shape Analysis.
    Yu Wang and Justin Solomon.
    Processing, Analyzing and Learning of Images, Shapes, and Forms, 2019. Preprint Publisher

  • Learning Geometric Operators on Meshes.
    Yu Wang, Vladimir Kim, Michael Bronstein and Justin Solomon.
    International Conference on Learning Representations (ICLR) 2019 Workshop.
    Representation Learning on Graphs and Manifolds. Paper

  • Steklov Spectral Geometry for Extrinsic Shape Analysis.
    Yu Wang, Mirela Ben-Chen, Iosif Polterovich and Justin Solomon.
    ACM Trans. on Graph. 38(1). Presented at ACM SIGGRAPH 2019.
    OpenAccessPaper. arXiv. Code

  • Steklov Geometry Processing: An Extrinsic Approach to Spectral Shape Analysis.
    Master Thesis, Massachusetts Institute of Technology. Paper

  • Linear Subspace Design for Real-Time Shape Deformation.
    Yu Wang, Alec Jacobson, Jernej Barbič and Ladislav Kavan.
    ACM Trans. on Graph. 34(4). ACM SIGGRAPH 2015. Paper

  • Grid-Based Nonlinear Elasticity with Spline Constraints for Image Deformation.
    Rajsekhar Setaluri, Yu Wang, Nathan Mitchell, Ladislav Kavan, Eftychios Sifakis.
    ACM SIGGRAPH / Eurographics Symposium on Computer Animation (SCA) 2015. Paper

  • Vision-based Probabilistic Localization for Soccer Robots.
    Yu Wang, Senior Thesis, Tsinghua University.

Education

Teaching Experience

Selected Awards

Work Experience

  • Research Intern, Creative Intelligence Laboratory, Adobe Research.
  • Research Intern, Visual Computing Group, Microsoft Research.
  • Visiting Research Fellow, Institute for Pure and Applied Mathematics, Los Angeles.
  • Research Assistant, Tsinghua National Laboratory for Information Science and Technology (TNList).

Professional Service

Reviewer

  • ACM Transactions on Graphics (TOG), 6 times
  • IEEE Transactions on Systems, Man, and Cybernetics (TSMC)
  • SIAM Journal on Imaging Sciences (SIIMS)
  • IEEE Transactions on Visualization and Computer Graphics (TVCG)
  • Nature Scientific Reports
  • ACM SIGGRAPH and ACM SIGGRAPH ASIA, 10+ times
  • International Conference on Machine Learning (ICML)
  • Neural Information Processing Systems (NeurIPS)
  • AAAI Conference on Artificial Intelligence (AAAI)
  • Artificial Intelligence and Statistics (AISTATS)
  • Eurographics and Pacific Graphics (EG & PG)
  • Computers & Graphics
  • and a few others

Program Committee Member

  • Graphics Replicability Stamp Initiative (GRSI)
  • Shape Modeling International (SMI)

Technical Skills & Subjects

(at the graduate level)

Math & Phys Differential Geometry and Manifold, Measure Theory, Real and Functional Analysis, Partial Differential Equations, Complex Analysis, Mathematics of Modern Physics, Modern and Optimal Control Theory, Inverse Problems, Optimal Transport, Computational Physics and Math, Physically Based Simulation, Numerical Methods for PDEs;

AI & ML Artificial Intelligence, Robotics, Advanced Computer Vision, Deep Learning, Advanced Machine Learning, Computational Learning Theory, Bayesian Modeling and Computing, Information (Theoretic) Inference;

Stats & Opt & EE Statistical Inference II, Topics in Probability and Statistics, Modern Convex Optimization, Numerical Nonlinear Optimization, Signal Processing and Analysis, Operations Research, Statistical Computing and Monte Carlo Methods;

CS Advanced Algorithms, Distributed System Engineering, Computer Architecture, Software System Engineering, Modern Computer Networks, Modern Operating System, Algorithms for Big Data, Theoretical Computer Science.

Recent Highlights

  • Schwarz–Schur Involution and Dirichlet-to-Neumann Condensing: Partial Differential Equations and Sparse Linear Systems Solved 1000× Faster: we present a somewhat surprising result that an algorithm customized to modern GPUs can significantly improve the speed of solving sparse linear systems, or exactly inverting convolution (deconvolution), up to 1000 faster than common sparse solvers.