Ruichen Xu 徐瑞辰

普通话 · Mandarin (Pinyin): Xú Ruìchén
粤语 · Cantonese (Jyutping): ceoi4 seoi6 san4

Ruichen Xu

Ph.D. Candidate

Computational Applied Mathematics
Department of Applied Mathematics & Statistics
Stony Brook University

Also known as Bill Xu
ruichen.xu@stonybrook.edu

Introduction

I am a Ph.D. candidate in Computational Applied Mathematics at Stony Brook University, advised by Dr. Yuefan Deng. My research lies at the intersection of machine learning and scientific computing, with a focus on AI for Science.

I study how learning systems can use incomplete observations and physical structure to solve scientific problems. My work spans neural operators, predictive representation learning, physics-aware generative models, and LLM-guided optimization. I also teach applied mathematics and mentor student research at Stony Brook.

Research

  • Learning from partial observations. Neural operators and benchmarks for reconstructing PDE fields from sparse and irregular measurements. PDE-OBS · Discretization mismatch
  • Predictive representations for scientific data. Joint-embedding learning for PDE inference and multi-resolution graph representations. SCOPE · HP-JEPA
  • Physics-aware generation and optimization. Diffusion, simulated annealing, and LLM-guided search for physical systems and molecular design. APOD · RL-QESA · Molecular optimization

News

All news

Selected publications

All publications
  1. Student–Curriculum Coupling: successive training batches assess current student competence to activate, reactivate, or suspend on-policy distillation supervision.

    Less Data, Better Timing: Student-Curriculum Coupling for VLM On-Policy Distillation in Temporal Video Grounding

    Jiacheng Qiu, Yunsoo Kim, Ruichen Xu, Jian Luo, Petar M. Djurić, Sima Mofakham

    arXiv:2609.40055 · 2026

    Adapts a compact training curriculum to the student model's evolving competence, improving temporal video grounding with fewer examples and less training time.

    arXiv PDF
  2. SCOPE overview showing changing observation coverage, observation-conditioned predictions, and full-field latent and physical reconstruction targets.

    SCOPE: Observation-Conditioned Full-Target Prediction for Sparse PDE Inference

    Ruichen Xu, Siyao Wang, Fang Wan, Jiacheng Qiu, Wenhan Gao, Jiaxing Zhang, Linsey Pang, Ravid Shwartz-Ziv, Prakhar Mehrotra, Yann LeCun, Yuefan Deng

    arXiv:2609.36527 · 2026

    Couples full-field latent prediction with physical reconstruction to recover complete PDE fields from sparse observations in a single pass.

    arXiv PDF Code
  3. PDE-OBS overview linking seven PDE families, controlled observation patterns, and reconstruction and forecasting under matched and shifted observations.

    PDE-OBS: Controlled Evaluation Across Observation Patterns

    Ruichen Xu, Siyao Wang, Fang Wan, Jiacheng Qiu, Wenhan Gao, Jiaxing Zhang, Linsey Pang, Ravid Shwartz-Ziv, Yann LeCun, Yuefan Deng

    arXiv:2609.36521 · 2026

    Benchmarks PDE reconstruction and forecasting across changing observation patterns while holding physical records, prediction targets, and data splits fixed.

    arXiv PDF Code
  4. Frequency-Aware Calibration overview shows input and output frequency-domain calibration around a frozen forecaster, adaptation using the latest matured mini-batch, and re-forecasting of the current mini-batch.

    Towards Principled Test-Time Adaptation for Time Series Forecasting

    Haochun Wang, Ruichen Xu, Georgios Kementzidis, Karen Cho, Sebastian Ramirez Villarreal, Yuefan Deng

    Transactions on Machine Learning Research (TMLR) · 2026 · accepted with minor revision

    Adapts time-series forecasts using frequency-domain calibration around a frozen forecasting model.

    OpenReview arXiv PDF
  5. HP-JEPA motivation and overview: multi-resolution graph structure, limitations of fixed partitions, and adaptive resolution weighting.

    HP-JEPA: Hierarchical Partitioning for Multi-Resolution Graph Joint-Embedding Predictive Learning

    Ruichen Xu, Jingxiang Qu, Wenhan Gao, Jiaxing Zhang, Linsey Pang, Ravid Shwartz-Ziv, Yann LeCun, Yuefan Deng

    arXiv:2608.00491 · 2026

    Learns graph representations across multiple resolutions through hierarchical partitioning and joint-embedding prediction.

    arXiv PDF
  6. GAGA overview showing forward noising and a Gaussian approximation that shortens the reverse generative trajectory.

    GAGA: Gaussianity-Aware Gaussian Approximation for Efficient 3D Molecular Generation

    Jingxiang Qu, Wenhan Gao, Ruichen Xu, Yi Liu

    ICLR 2026

    Uses Gaussianity-aware approximations to shorten diffusion sampling for efficient 3D molecular generation.

    OpenReview arXiv PDF
  7. Multistep protein backmapping pipeline, from coarse beads through intermediate structures to atomistic detail using successive generative networks.

    Multistep Generative Backmapping of Coarse-Grained Structures

    Georgios Kementzidis, Erin Wong, John Nicholson, Ruichen Xu, Yuefan Deng

    Computer Physics Communications · 327, 110286 · 2026

    Restores atomistic detail from coarse-grained structures through a sequence of intermediate generative models.

    DOI
  8. CROP pipeline maps an input through lifting, a neural operator on band-limited latent features, and projection to an output at any discretization.

    Discretization-invariance? On the Discretization Mismatch Errors in Neural Operators

    Wenhan Gao, Ruichen Xu, Yuefan Deng, Yi Liu

    ICLR 2025

    Studies errors caused by changing discretization in neural operators and develops a band-limited latent representation.

    OpenReview Conference page PDF
  9. Kolmogorov–Arnold and multilayer-perceptron Hamiltonian neural network architectures, with energy derivatives feeding a shared training loss.

    Kolmogorov–Arnold Representation for Symplectic Learning: Advancing Hamiltonian Neural Networks

    Zongyu Wu, Ruichen Xu†, Luoyao Chen, Georgios Kementzidis, Siyao Wang, Yuefan Deng · †Co-first author and project lead

    IJCNN 2025

    Combines Kolmogorov–Arnold representations with Hamiltonian neural networks to learn energy-based dynamical models.

    IEEE DOI arXiv PDF
  10. APOD framework turns noisy initial distributions into samples through diffusion steps guided by observation and PDE terms.

    APOD: Adaptive PDE-Observation Diffusion for Physics-Constrained Sampling

    Ruichen Xu, Haochun Wang, Georgios Kementzidis, Chenhao Si, Yuefan Deng

    ICML 2025 Workshop on Assessing World Models · Poster

    Guides diffusion sampling with observations and PDE constraints to recover physically consistent fields.

    OpenReview PDF
  11. A Fourier layer uses a neural kernel generator to create a convolution kernel from input frequencies before inverse transform, bias and activation.

    Dynamic Schwartz–Fourier Neural Operator for Enhanced Expressive Power

    Wenhan Gao, Jian Luo, Ruichen Xu, Yi Liu

    Transactions on Machine Learning Research · 2025

    Introduces input-dependent Fourier kernels to increase the expressive power of neural operators.

    OpenReview PDF

Academic service

Reviewing experience

ICML 2026 Gold Reviewer · May 2026
Recognized by the program chairs for the quality of submitted reviews.

Conference reviewing & program committees: ICLR (2026, 2027) — 2027 invitation accepted; ICML (2026) — Gold Reviewer; NeurIPS (2026); KDD — AI4Sciences Track (2026, 2027) — Reviewer; 2027 Cycle 1; KDD — Research Track (2027) — Reviewer; Cycle 1; AAAI (2027) — Program Committee member; IJCNN (2025, 2026).

Journal reviewer: TMLR (2026), IEEE TNNLS (2025), and Neurocomputing (2026).

Workshop reviewer: AI for Math @ ICML (2025).