Hong Wang 王泓

I’m Hong Wang, a Ph.D. student in the Artificial Intelligence track at the School of Information Science and Technology, University of Science and Technology of China (USTC).
My research lies at the intersection of AI and scientific computing, with current interests in AI for scientific computing, matrix computation acceleration, large language models, and reinforcement learning for reasoning and code tasks.
Recently, I have been working on neural operators, efficient PDE data generation, learning-based numerical solvers, RL for LLM reasoning, and code agents.
I welcome discussions and collaborations related to AI for science, scientific computing, LLM post-training, and reinforcement learning.
For the latest materials, please see [Download CV], [Research Slides], and [Lab Discussion Slides].

Research Interests

• AI for scientific computing and AI for Science
• Matrix computation acceleration and numerical linear algebra
• Neural operators and PDE learning
• Large language models, RL for reasoning, and code agents

Contact

• Email: wanghong1700@mail.ustc.edu.cn
• WeChat: wh1732060534
• QQ: 1732060534
• Phone: (+86) 17318588680

Education

• Ph.D. student, Artificial Intelligence track
  School of Information Science and Technology, USTC, Hefei, China, 2023-present
• M.S., Computational Mathematics
  School of Mathematical Sciences, USTC, Hefei, China, 2021-2023
  Advisor: Prof. Kuan Xu
• B.S., Theoretical Physics; B.S., Mathematics (double degree)
  School of the Gifted Young, USTC, Hefei, China, 2017-2021

Internship and Projects

• Tencent AI Code Assistant (CSIG CodeBuddy) Team, Qingyun Program — Research Intern
  Tencent CodeBuddy, Shenzhen, China, May 2025 - December 2025.
  Worked on LLM post-training for code and mathematical reasoning, with a focus on reasoning-tree-based curriculum design, entropy dynamics in RLVR, and data curation / sample generation for code RL.

• Huawei Project: Fast Similar Matrix Computation via Low-Rank Decomposition
  Huawei contract No. TC20211015677, September 2021 - June 2022.
  Led the full research pipeline from literature review and algorithm design to theory, experiments, and reporting, and developed accelerated matrix algorithms with 1.2×–20× speedups in real scenarios.

Awards

• Doctorate National Scholarship, USTC
  
• Outstanding Graduate, USTC
  
• ICLR 2024 financial assistance
  
• NeurIPS 2025 financial assistance
  

Professional Services and Mentoring

• Reviewer for ICLR, ICML, NeurIPS, AAAI, and other top AI venues
• Mentored more than 10 students on research projects during my graduate studies

Selected Publications (Google Scholar)

Large Language Models and Reinforcement Learning

• Scheduling Your LLM Reinforcement Learning with Reasoning Trees
  Authors: Hong Wang^*, Zhezheng Hao^*, Jian Luo, Chenxing Wei, Yao Shu, Lei Liu, Qiang Lin, Hande Dong^†, Jiawei Chen^†.
  ICLR 2026, Tencent internship work.
  [arXiv] [code]

• Rethinking Entropy Interventions in RLVR: An Entropy Change Perspective
  Authors: Zhezheng Hao^*, Hong Wang^*, Haoyang Liu, Jian Luo, Jiarui Yu, Hande Dong^†, Qiang Lin, Can Wang, Jiawei Chen^†.
  ACL 2026 Oral, Tencent internship work.
  [arXiv] [code]

• Exploiting Edited Large Language Models as General Scientific Optimizers
  Authors: Qitan Lv, Tianyu Liu, Hong Wang^†.
  NAACL 2025.
  [arXiv]

• ReCreate: Reasoning and Creating Domain Agents Driven by Experience
  Authors: Zhezheng Hao, Hong Wang, Jian Luo, Jianqing Zhang, Yuyan Zhou, Qiang Lin, Can Wang, Hande Dong^†, Jiawei Chen^†.
  ACL 2026, Tencent internship work.
  [arXiv]

• Perturbation-Restrained Sequential Model Editing
  Authors: Jun-Yu Ma, Hong Wang, Hao-Xiang Xu, Zhen-Hua Ling, Jia-Chen Gu^†.
  ICLR 2025.
  [arXiv]

• Energy-Regularized Sequential Model Editing on Hyperspheres
  Authors: Qingyuan Liu^*, Jia-Chen Gu^*, Yunzhi Yao, Hong Wang, Nanyun Peng.
  ICLR 2026.
  [arXiv] [code]

• Plug-and-Play Data Module for Code RL: Adaptive Ambiguity Replay
  ACL 2026 Findings, Tencent internship work.

• GAPO: Robust Advantage Estimation for Real-World Code LLMs
  Authors: Jianqing Zhang, Zhezheng Hao, Wei Xia^†, Hande Dong, Hong Wang, Chenxing Wei, Yuyan Zhou, Yubin Qi, Qiang Lin, Jian Cao.
  ACL 2026 Findings, Tencent internship work.
  [arXiv] [code]

• LEPO: Latent Reasoning Policy Optimization for Large Language Models
  Authors: Yuyan Zhou^*, Jiarui Yu^*, Hande Dong^†, Zhezheng Hao, Hong Wang, Jianqing Zhang, Qiang Lin.
  ACL 2026 Findings, Tencent internship work.
  [arXiv] [code]

AI for Scientific Computing

• Accelerating Data Generation for Neural Operators via Krylov Subspace Recycling
  Authors: Hong Wang^*, Zhongkai Hao^*, Zijie Geng, Zhen Wang, Bin Li, Feng Wu.
  ICLR 2024 Spotlight.
  [arXiv] [code]

• Mixture-of-Experts Operator Transformer for Large-Scale PDE Pre-Training
  Authors: Hong Wang^*, Haiyang Xin^*, Xuanze Yang, Fei Zha, Huanshuo Dong, Yan Jiang.

  NeurIPS 2025.
  [arXiv] [code]

• SymMaP: Improving Computational Efficiency in Linear Solvers through Symbolic Preconditioning
  Authors: Hong Wang, Minghao Ma, Haoran Shao, Haoyang Liu.
  NeurIPS 2025.
  [arXiv] [code]

• STNet: Spectral Transformation Network for Solving Operator Eigenvalue Problem
  Authors: Hong Wang^*, Jiang Yixuan^*, Xinyi Li, Jian Luo, Huanshuo Dong.
  NeurIPS 2025.
  [arXiv] [code]

• Accelerating Eigenvalue Dataset Generation via Chebyshev Subspace Filter
  Authors: Hong Wang, Jian Luo, Huanshuo Dong, Yeqiu Chen, Runmin Jiang, Zhen Huang.
  ICLR 2026.
  [arXiv]

• HGATSolver: A Heterogeneous Graph Attention Solver for Fluid-Structure Interaction
  Authors: Qin-Yi Zhang^*, Hong Wang^*, Siyao Liu, Haichuan Lin, Linying Cao, Xiao-Hu Zhou, Chen Chen, Shuangyi Wang^†, Zeng-Guang Hou^†.
  AAAI 2026 Oral.
  [arXiv] [code]

• Learning Neural Operators from Partial Observations via Latent Autoregressive Modeling
  Authors: Jingren Hou^*, Hong Wang^*, Pengyu Xu, Chang Gao, Huafeng Liu, Liping Jing^†.
  AAAI 2026.
  [arXiv]

• Self-Attention to Operator Learning-based 3D-IC Thermal Simulation
  Authors: Zhen Huang^*, Hong Wang^*, Wenkai Yang, Muxi Tang, Depeng Xie, Ting-Jung Lin, Yu Zhang, Wei W. Xing, Lei He.
  DAC 2025.
  [arXiv]

• Accelerating PDE Data Generation via Differential Operator Action in Solution Space
  Authors: Huanshuo Dong, Hong Wang, Haoyang Liu, Jian Luo.

  ICML 2024.
  [arXiv]

• Neural Krylov Iteration for Accelerating Linear System Solving
  Authors: Jian Luo, Hong Wang, Huanshuo Dong, Zijie Geng, Hanzhu Chen, Yufei Kuang.
  NeurIPS 2024 Spotlight.
  [NeurIPS]

• Coordinate Transform Fourier Neural Operators for Symmetries in Physical Modelings
  Authors: Wenhan Gao, Ruichen Xu, Hong Wang, Yi Liu^†.
  TMLR 2025.
  [OpenReview] [code]

Manuscripts Under Review

• Test-Time Policy Adaptation for Enhanced Multi-Turn Interactions with LLMs
  Authors: Chenxing Wei, Hong Wang, Ying He, Fei Yu, Yao Shu^†.
  Submitted to ICML 2026.
  [arXiv]

• Words & Weights: Streamlining Multi-Turn Interactions via Co-Adaptation
  Authors: Chenxing Wei, Hong Wang, Ying He, Zhongxiang Dai, Bo Jiang, F. Richard Yu, Yao Shu^†.
  Submitted to ICML 2026.
  [arXiv]

• LFPO: Likelihood-Free Policy Optimization for Masked Diffusion Models
  Submitted to ICML 2026.