Research

research interests, published papers, and working papers.

My research lies at the intersection of numerical analysis, stochastic control, reinforcement learning, and machine learning. I am particularly interested in quasi-Monte Carlo methods for high-dimensional integration and stochastic differential equations, continuous-time reinforcement learning and stochastic optimization, and learning-based numerical methods for nonlinear partial differential equations.

Published Papers

  1. Achieving high convergence rates by quasi-Monte Carlo and importance sampling for unbounded integrands
    Du Ouyang, Xiaoqun Wang, and Zhijian He. SIAM Journal on Numerical Analysis, 62 (2024), 2393-2414.
    Links: DOI: 10.1137/23M1622489 | arXiv: 2310.00650

  2. Generalization error analysis of deep backward dynamic programming for solving nonlinear PDEs
    Du Ouyang, Jichang Xiao, and Xiaoqun Wang. Applied Numerical Mathematics, 217 (2025), 319-339.
    Links: DOI: 10.1016/j.apnum.2025.06.013 | arXiv: 2407.14566

  3. Accuracy of discretely sampled stochastic policies in continuous-time reinforcement learning
    Yanwei Jia, Du Ouyang, and Yufei Zhang. SIAM Journal on Control and Optimization, 64 (2026), 1889-1929. Alphabetical order.
    Links: DOI: 10.1137/25M1742199 | arXiv: 2503.09981

Working Papers

  1. Why Quasi-Monte Carlo Can Remain Effective in SDE Simulation: An Effective Dimension Perspective
    Du Ouyang and Xiaoqun Wang. Submitted, 2026.
    Links: ResearchGate | SSRN

  2. A zeroth-order deep learning method for fully nonlinear parabolic partial differential equations with unknown coefficients
    Yanwei Jia, Du Ouyang, Huyên Pham, and Xun Yu Zhou. Submitted. Alphabetical order.
    Links: arXiv: 2606.24999

  3. Convergence and regret of the policy gradient for multi-armed bandits in diffusion environment
    Yanwei Jia and Du Ouyang. Submitted. Alphabetical order.
    Links: arXiv: 2607.29593

  4. Quasi-Monte Carlo for SDE simulation: Error analysis and dimensionality reduction
    Du Ouyang, Zexin Pan, and Zhijian He. Submitted.
    Links: arXiv: 2606.24207

  5. Quasi-Monte Carlo integration over $\mathbb{R}^s$ with boundary-damping importance sampling
    Zexin Pan, Du Ouyang, and Zhijian He. Submitted.
    Links: arXiv: 2509.07509

  6. Uncertainty quantification using importance-sampled quasi-Monte Carlo with dimension-independent convergence rates
    Zexin Pan, Du Ouyang, and Zhijian He. Submitted.
    Links: arXiv: 2603.00843

  7. Adaptive Two-Level Quasi-Monte Carlo for Soft Actor-Critic
    Du Ouyang, Zhenpeng Shi, Aodong Guo, Huaze Tang, Hejin Wang, Chao Wang, and Wenbo Ding.
    ICML 2024 Workshop: Aligning Reinforcement Learning Experimentalists and Theorists.
    Links: OpenReview