Publications (45)
Fast algorithms for integral formulations of steady-state radiative transfer equation
Yuwei Fan, Jing An, Lexing Ying
We investigate integral formulations and fast algorithms for the steady-state radiative transfer equation with isotropic and anisotropic scattering. When the scattering term is a s…
Shifted CholeskyQR for sparse matrices
Haoran Guan, Yuwei Fan
In this work, we focus on Shifted CholeskyQR (SCholeskyQR) for sparse matrices. We provide a new shifted item for Shifted CholeskyQR3 (SCholeskyQR3) based on the number of non-…
A Framework on Moment Model Reduction for Kinetic Equation
Zhenning Cai, Yuwei Fan, Ruo Li
By a further investigation on the structure of the coefficient matrix of the globally hyperbolic regularized moment equations for Boltzmann equation in [Z. Cai, Y. Fan and R. Li, C…
An entropic fourier method for the Boltzmann equation
Zhenning Cai, Yuwei Fan, Lexing Ying
We propose an entropic Fourier method for the numerical discretization of the Boltzmann collision operator. The method, which is obtained by modifying a Fourier Galerkin method to…
Solving Optical Tomography with Deep Learning
Yuwei Fan, Lexing Ying
This paper presents a neural network approach for solving two-dimensional optical tomography (OT) problems based on the radiative transfer equation. The mathematical problem of OT…
13-Moment System with Global Hyperbolicity for Quantum Gas
Yana Di, Yuwei Fan, Ruo Li
We point out that the quantum Grad's 13-moment system [R. Yano, Physica A: Statistical Mechanics and its Applications, 416:231-241, 2014] is lack of global hyperbolicity, and even…
Linear Stability of Hyperbolic Moment Models for Boltzmann Equation
Yana Di, Yuwei Fan, Ruo Li +1
Grad's moment models for Boltzmann equation were recently regularized to globally hyperbolic systems, and thus the regularized models attain local well-posedness for Cauchy data. T…
Meta-learning Pseudo-differential Operators with Deep Neural Networks
Jordi Feliu-Faba, Yuwei Fan, Lexing Ying
This paper introduces a meta-learning approach for parameterized pseudo-differential operators with deep neural networks. With the help of the nonstandard wavelet form, the pseudo-…
Hyperbolic Model Reduction for Kinetic Equations
Zhenning Cai, Yuwei Fan, Ruo Li
We make a brief historical review to the moment model reduction to the kinetic equations, particularly the Grad's moment method for Boltzmann equation. The focus is on the hyperbol…
Solving Electrical Impedance Tomography with Deep Learning
Yuwei Fan, Lexing Ying
This paper introduces a new approach for solving electrical impedance tomography (EIT) problems using deep neural networks. The mathematical problem of EIT is to invert the electri…
LoPT: Lossless Parallel Tokenization Acceleration for Long Context Inference of Large Language Model
Wei Shao, Lingchao Zheng, Pengyu Wang +3
Long context inference scenarios have become increasingly important for large language models, yet they introduce significant computational latency. While prior research has optimi…
Solving Inverse Wave Scattering with Deep Learning
Yuwei Fan, Lexing Ying
This paper proposes a neural network approach for solving two classical problems in the two-dimensional inverse wave scattering: far field pattern problem and seismic imaging. The…
Beyond Token Length: Step Pruner for Efficient and Accurate Reasoning in Large Language Models
Canhui Wu, Qiong Cao, Chang Li +5
Large Reasoning Models (LRMs) demonstrate strong performance on complex tasks but often suffer from excessive verbosity, known as "overthinking." Existing solutions via reinforceme…
Tackling the Data-Parallel Load Balancing Bottleneck in LLM Serving: Practical Online Routing at Scale
Tianci Bu, Yuan Lyu, Zixi Chen +6
Data-parallel (DP) load balancing has emerged as a first-order bottleneck in large-scale LLM serving. When a model is sharded across devices via tensor parallelism (TP) or expert p…
Multi-Level Fine-Tuning: Closing Generalization Gaps in Approximation of Solution Maps under a Limited Budget for Training Data
Zhihan Li, Yuwei Fan, Lexing Ying
In scientific machine learning, regression networks have been recently applied to approximate solution maps (e.g., potential-ground state map of Schrödinger equation). In this pap…
CANN Bench: Benchmarking Agent Generated Kernels against Real NPU and Algorithmic Limits
Xue-Jian Gao, Deng Pan, Yueming Su +12
AI agents are now capable of writing, compiling, and iteratively optimizing low-level operator kernels on different hardware platforms. Existing benchmarks, however, focus almost e…
A multiscale neural network based on hierarchical nested bases
Yuwei Fan, Jordi Feliu-Faba, Lin Lin +2
In recent years, deep learning has led to impressive results in many fields. In this paper, we introduce a multi-scale artificial neural network for high-dimensional non-linear map…
AIS: Adaptive Importance Sampling for Quantized RL
Jiajun Zhou, Wei Shao, Lingchao Zheng +2
Reinforcement learning (RL) for large language models (LLMs) is dominated by the cost of rollout generation, which has motivated the use of low-precision rollouts (e.g., FP8) paire…
A multiscale neural network based on hierarchical matrices
Yuwei Fan, Lin Lin, Lexing Ying +1
In this work we introduce a new multiscale artificial neural network based on the structure of -matrices. This network generalizes the latter to the nonlinear case by…
A Nonlinear Moment Model for Radiative Transfer Equation in Slab Geometry
Yuwei Fan, Ruo Li, Lingchao Zheng
This paper is concerned with the approximation of the radiative transfer equation for a grey medium in the slab geometry by the moment method. We develop a novel moment model inspi…
Analysis of randomized CholeskyQR for sparse matrices
Haoran Guan, Yuwei Fan
This work is about rounding error analysis of randomized CholeskyQR-type algorithms for sparse matrices. We often encounter QR factorization of the sparse matrices in many real pro…
Solving Traveltime Tomography with Deep Learning
Yuwei Fan, Lexing Ying
This paper introduces a neural network approach for solving two-dimensional traveltime tomography (TT) problems based on the eikonal equation. The mathematical problem of TT is to…
Resolving Knudsen Layer by High Order Moment Expansion
Yuwei Fan, Jun Li, Ruo Li +1
We model the Knudsen layer in Kramers' problem by linearized high order hyperbolic moment system. Due to the hyperbolicity, the boundary conditions of the moment system is properly…
Model Reduction of Kinetic Equations by Operator Projection
Yuwei Fan, Julian Koellermeier, Jun Li +2
By a further study of the mechanism of the hyperbolic regularization of the moment system for Boltzmann equation proposed in [Z. Cai, Y. Fan, R. Li, Comm. Math. Sci. 11(2): 547-571…
AMLA: MUL by ADD in FlashAttention Rescaling
Qichen Liao, Chengqiu Hu, Fangzheng Miao +8
Multi-head Latent Attention (MLA) significantly reduces KVCache memory usage in Large Language Models while introducing substantial computational overhead and intermediate variable…
On Hyperbolicity of 13-Moment System
Zhenning Cai, Yuwei Fan, Ruo Li
We point out that the thermodynamic equilibrium is not an interior point of the hyperbolicity region of Grad's 13-moment system. With a compact expansion of the phase density, whic…
A Simple Multiscale Method for Mean Field Games
Haoya Li, Yuwei Fan, Lexing Ying
This paper proposes a multiscale method for solving the numerical solution of mean field games which accelerates the convergence and addresses the problem of determining the initia…
An improved error analysis of CholeskyQR with the randomized model
Haoran Guan, Yuwei Fan
This work is about an improved error analysis of CholeskyQR with the randomized model for the tall-skinny . Due to the structure of CholeskyQR, we uti…
A Nonlinear Hyperbolic Model for Radiative Transfer Equation in Slab Geometry
Yuwei Fan, Ruo Li, Lingchao Zheng
Linear models for the radiative transfer equation have been well developed, while nonlinear models are seldom investigated even for slab geometry due to some essential difficulties…
BCR-Net: a neural network based on the nonstandard wavelet form
Yuwei Fan, Cindy Orozco Bohorquez, Lexing Ying
This paper proposes a novel neural network architecture inspired by the nonstandard form proposed by Beylkin, Coifman, and Rokhlin in [Communications on Pure and Applied Mathematic…
Globally Hyperbolic Regularization of Grad's Moment System in One Dimensional Space
Zhenning Cai, Yuwei Fan, Ruo Li
In this paper, we present a regularization to 1D Grad's moment system to achieve global hyperbolicity. The regularization is based on the observation that the characteristic polyno…
Multi-Scale Dequant: Eliminating Dequantization Bottleneck via Activation Decomposition for Efficient LLM Inference
Lingchao Zheng, Yuwei Fan, Jun Li +5
Quantization is essential for efficient large language model (LLM) inference, yet the dequantization step-converting low-bit weights back to high-precision for matrix multiplicatio…
A Novel Randomized XR-Based Preconditioned CholeskyQR Algorithm
Yuwei Fan, Yixiao Guo, Ting Lin
CholeskyQR is a simple and fast QR decomposition via Cholesky decomposition, while it has been considered highly sensitive to the condition number. In this paper, we provide a rand…
Quantum Hydrodynamic Model by Moment Closure of Wigner Equation
Zhenning Cai, Yuwei Fan, Ruo Li +2
In this paper, we derive the quantum hydrodynamics models based on the moment closure of the Wigner equation. The moment expansion adopted is of the Grad type firstly proposed in \…
Fine-Tuning Pre-trained Large Time Series Models for Prediction of Wind Turbine SCADA Data
Yuwei Fan, Tao Song, Chenlong Feng +3
The remarkable achievements of large models in the fields of natural language processing (NLP) and computer vision (CV) have sparked interest in their application to time series fo…
Filtered Hyperbolic Moment Method for the Vlasov Equation
Yana Di, Yuwei Fan, Zhenzhong Kou +2
In this paper, we investigate the effect of the filter for the hyperbolic moment equations(HME) [15] of the Vlasov-Poisson equations and propose a novel quasi time-consistent filte…
Analytical Provisioning for Attention-FFN Disaggregated LLM Serving under Stochastic Workloads
Chendong Song, Meixuan Wang, Hang Zhou +5
Attentio-FFN disaggregation (AFD) is an emerging architecture for LLM decoding that separates state-heavy, KV-cache-dominated Attention computation from stateless, compute-intensiv…
Globally Hyperbolic Regularization of Grad's Moment System
Zhenning Cai, Yuwei Fan, Ruo Li
In this paper, we propose a globally hyperbolic regularization to the general Grad's moment system in multi-dimensional spaces. Systems with moments up to an arbitrary order are st…
A new randomized CholeskyQR based on LU decomposition with partial pivoting
Yuwei Fan, Haoran Guan, Zhonghua Qiao
The paper introduces RCLUPP, a randomized QR factorization method that uses LU decomposition with partial pivoting and matrix sketching to improve efficiency and robustness over ex…
An improved Shifted CholeskyQR based on columns
Yuwei Fan, Haoran Guan, Zhonghua Qiao
Among all the deterministic CholeskyQR-type algorithms, Shifted CholeskyQR3 is specifically designed to address the QR factorization of ill-conditioned matrices. This algorithm int…
Domain-specific ReAct for physics-integrated iterative modeling: A case study of LLM agents for gas path analysis of gas turbines
Tao Song, Yuwei Fan, Chenlong Feng +3
This study explores the application of large language models (LLMs) with callable tools in energy and power engineering domain, focusing on gas path analysis of gas turbines. We de…
Burnett Spectral Method for the Spatially Homogeneous Boltzmann Equation
Zhenning Cai, Yuwei Fan, Yanli Wang
We develop a spectral method for the spatially homogeneous Boltzmann equation using Burnett polynomials in the basis functions. Using the sparsity of the coefficients in the expans…
Globally Hyperbolic Moment System by Generalized Hermite Expansion
Yuwei Fan, Ruo Li
In a recent paper [Z.-N. Cai, Y.-W. Fan, and R. Li. Tech Report, Institude of Math, Peking Univeristy(2013)], it was revealed that a modified 13-moment system taking intrinsic heat…
RCLUPPr: a new randomized CholeskyQR with LU preconditioning
Haoran Guan, Zhenyu Zou, Yufeng Wei +3
In this work, we present the comprehensive rounding error analysis of RCLUPPr proposed in \cite{RCLUPP}, which is a novel randomized CholeskyQR-type algorithm performing LU decompo…
Coarse-Proxy Reduced Basis Methods for Integral Equations
Philip A. Etter, Yuwei Fan, Lexing Ying
In this paper, we introduce a new reduced basis methodology for accelerating the computation of large parameterized systems of high-fidelity integral equations. Core to our methodo…