Shizheng Wen (闻仕政)
PhD Student · Seminar for Applied Mathematics (SAM), ETH Zurich
Affiliated PhD · ETH AI Center · Advised by Prof. Siddhartha Mishra
About
I am Shizheng Wen (闻仕政), a first-year PhD student in AI and applied math at ETH Zurich. My research sits at the intersection of scientific computing and machine learning, with a focus on building fast, scalable, and differentiable solvers for partial differential equations (PDEs).
My current work spans three directions:
- Differentiable solvers. I am the creator of TensorMesh — a GPU-native differentiable finite-element library built on PyTorch — and torch-sla, a differentiable sparse linear algebra library with full autograd support across multiple solver backends. Together they aim to make classical numerical methods first-class citizens in modern ML pipelines.
- Scalable neural solver backbones. I develop neural operator architectures that handle complex geometries and scale to large problems, including GAOT and RIGNO.
- PDE foundation models. I am working on large-scale pretraining of foundation models for PDEs across diverse physical systems (more details coming soon).
In the past, I have had the honor of working with Prof. Earl Dowell at Duke University, and Prof. Wanlin Guo and Prof. Xianglei Liu at NUAA.
PDEs are the language of natural science, governing everything from fluid dynamics to quantum mechanics. As an enthusiast of mathematics, physics, and biology, I love integrating multidisciplinary perspectives to tackle complex problems. If you'd like to connect or collaborate, feel free to reach out!
News
-
May 2026 We open-sourced TensorMesh! A fast, differentiable, JIT-free, debugging-friendly finite element library for PyTorch.
-
May 2026 I received the ETH Medal at my master's graduation ceremony, the highest honor for students with outstanding master's and doctoral theses.
-
Apr 2026 One paper "TensorGalerkin" accepted at ICML 2026.
-
Jan 2026 One paper "MOSIV" accepted at ICLR 2026.
-
May 2025 Started my Ph.D. journey at the Seminar for Applied Mathematics, ETH Zurich.
-
Apr 2024 One paper "Phase-field simulation and machine learning of low-field magneto-elastocaloric effect in a multiferroic composite" accepted at International Journal of Mechanical Sciences (IJMS).
-
Nov 2023 One paper "Feature identification in complex fluid flows by convolutional neural networks" accepted at Theoretical and Applied Mechanics Letters (TAML).
-
Jul 2022 One paper "A machine learning strategy for modeling and optimal design of near-field radiative heat transfer" accepted at Applied Physics Letters (APL).
Publications
-
Learning, Solving and Optimizing PDEs with TensorGalerkin: an efficient high-performance Galerkin assembly algorithm
-
torch-sla: Differentiable Sparse Linear Algebra with Adjoint Solvers and Sparse Tensor Parallelism for PyTorch
-
MOSIV: Multi-Object System Identification from Videos
-
Geometry Aware Operator Transformer as an Efficient and Accurate Neural Surrogate for PDEs on Arbitrary Domains
-
RIGNO: A Graph-based framework for robust and accurate operator learning for PDEs on arbitrary domains
-
Phase-field simulation and machine learning of low-field magneto-elastocaloric effect in a multiferroic composite
-
Feature Identification in Complex Fluid Flows by Convolutional Neural Networks
-
A machine learning strategy for modeling and optimal design of near-field radiative heat transfer
-
High-performance three-body near-field thermophotovoltaic energy conversion
-
Ultrahigh thermal rectification based on near-field thermal radiation between dissimilar nanoparticles
* Equal contribution.
Selected Awards
- ICML Gold Reviewer · 2026
- ETH Medal (top 2.5%) · 2026
- National Scholarship (top 1%) · 2019
- NUAA Presidential Fellowship (top 0.1%) · 2019
Invited Talks
-
Mar 2026 Learning, Solving and Optimizing PDEs with TensorGalerkin: an Efficient High-Performance Galerkin Assembly Algorithm — Simon Fraser University, hosted by Wuyang Chen.
-
Mar 2026 Geometry Aware Operator Transformer as an Efficient and Accurate Neural Surrogate for PDEs on Arbitrary Domains — Tsinghua University, hosted by Angelica Aviles-Rivero.
-
Dec 2025 Accelerating Computational Science: From Differentiable Solvers to Geometry Aware Operator Transformers — Stanford University, hosted by Charbel Farhat.
Teaching
-
Numerical Methods for Partial Differential Equations (10 ECTS)
-
AI in the Sciences and Engineering (6 ECTS)