I am broadly interested in the mathematics of machine learning and statistics. My work focuses on applications of optimal transport to computational problems arising in these fields, such as log-concave sampling (see my book draft above) and variational inference (Lambert et al. (2023); Diao et al. (2023); Jiang, Chewi, and Pooladian (2024)).

Publications

• Theory and computation for structured variational inference. Shunan Sheng, Bohan Wu, Bennett Zhu, Sinho Chewi, and Aram-Alexandre Pooladian. November 2025. [arXiv] [.bib]
• Shifted composition II: shift Harnack inequalities and curvature upper bounds. Jason Altschuler and Sinho Chewi. November 2025. [arXiv] [IZS 2024] [IEEE Transactions on Information Theory] [Slides] [.bib]
• Sublinear iterations can suffice even for DDPMs. Matthew Zhang, Stephen Huan, Jerry Huang, Nicholas Boffi, Sitan Chen, and Sinho Chewi. November 2025. [arXiv] [Slides] [.bib]
• Stability of the Kim–Milman flow map. Sinho Chewi, Aram-Alexandre Pooladian, and Matthew Zhang. November 2025. [arXiv] [.bib]
• Algorithms for mean-field variational inference via polyhedral optimization in the Wasserstein space. Yiheng Jiang, Sinho Chewi, and Aram-Alexandre Pooladian. August 2025. [arXiv] [COLT 2024] [Foundations of Computational Mathematics] [Slides] [.bib]
• Gaussian mixture layers for neural networks. Sinho Chewi, Philippe Rigollet, and Yuling Yan. August 2025. [arXiv] [.bib]
• Shifted composition IV: toward ballistic acceleration for log-concave sampling. Jason Altschuler, Sinho Chewi, and Matthew Zhang. June 2025. [arXiv] [Slides] [.bib]
• DDPM score matching and distribution learning. Sinho Chewi, Alkis Kalavasis, Anay Mehrotra, and Omar Montasser. April 2025. [arXiv] [ICLR 2025 DeLTA Workshop (Best Short Paper Award)] [Deep Learning Theory Workshop @ RIKEN AIP (Video)] [Slides] [.bib]
• Shifted composition III: local error framework for KL divergence. Jason Altschuler and Sinho Chewi. December 2024. [arXiv] [Online Monte Carlo Seminar (Video)] [Slides] [.bib]
• The ballistic limit of the log-Sobolev constant equals the Polyak–Łojasiewicz constant. Sinho Chewi and Austin Stromme. November 2024. [arXiv] [.bib]
• Shifted composition I: Harnack and reverse transport inequalities. Jason Altschuler and Sinho Chewi. October 2024. [arXiv] [IZS 2024] [IEEE Transactions on Information Theory] [Slides] [.bib]
• Uniform-in-N log-Sobolev inequality for the mean-field Langevin dynamics with convex energy. Sinho Chewi, Atsushi Nitanda, and Matthew Zhang. September 2024. [arXiv] [Slides] [.bib]
• Query lower bounds for log-concave sampling. Sinho Chewi, Jaume de Dios Pont, Jerry Li, Chen Lu, and Shyam Narayanan. August 2024. [arXiv] [FOCS 2023] [Journal of the ACM] [Slides] [.bib]
• Analysis of Langevin Monte Carlo from Poincaré to log-Sobolev. Sinho Chewi, Murat Erdogdu, Mufan Li, Ruoqi Shen, and Matthew Zhang. July 2024. [arXiv] [COLT 2022 (Extended Abstract)] [Foundations of Computational Mathematics] [.bib]
• Fast parallel sampling under isoperimetry. Nima Anari, Sinho Chewi, and Thuy-Duong Vuong. July 2024. [arXiv] [COLT 2024] [.bib]
• Sampling from the mean-field stationary distribution. Yunbum Kook, Matthew Zhang, Sinho Chewi, Murat Erdogdu, and Mufan Li. July 2024. [arXiv] [COLT 2024] [.bib]
• Faster high-accuracy log-concave sampling via algorithmic warm starts. Jason Altschuler and Sinho Chewi. June 2024. [arXiv] [FOCS 2023] [Journal of the ACM] [Slides] [.bib]
• Learning threshold neurons via the "edge of stability". Kwangjun Ahn, Sébastien Bubeck, Sinho Chewi, Yin Tat Lee, Felipe Suárez-Colmenares, and Yi Zhang. December 2023. [arXiv] [NeurIPS 2023] [.bib]
• The probability flow ODE is provably fast. Sitan Chen, Sinho Chewi, Holden Lee, Yuanzhi Li, Jianfeng Lu, and Adil Salim. December 2023. [arXiv] [NeurIPS 2023] [Slides] [.bib]
• An entropic generalization of Caffarelli's contraction theorem via covariance inequalities. Sinho Chewi and Aram-Alexandre Pooladian. November 2023. [arXiv] [Comptes Rendus Mathématique] [.bib]
• The entropic barrier is n-self-concordant. Sinho Chewi. September 2023. [arXiv] [GAFA Seminar Notes] [.bib]
• Forward-backward Gaussian variational inference via JKO in the Bures–Wasserstein space. Michael Diao, Krishnakumar Balasubramanian, Sinho Chewi, and Adil Salim. July 2023. [arXiv] [ICML 2023] [Slides] [.bib]
• Improved discretization analysis for underdamped Langevin Monte Carlo. Matthew Zhang, Sinho Chewi, Mufan Li, Krishnakumar Balasubramanian, and Murat Erdogdu. July 2023. [arXiv] [COLT 2023] [.bib]
• Sampling is as easy as learning the score: theory for diffusion models with minimal data assumptions. Sitan Chen, Sinho Chewi, Jerry Li, Yuanzhi Li, Adil Salim, and Anru Zhang. May 2023. [arXiv] [NeurIPS 2022 Workshop on Score-Based Methods] [ICLR 2023 (Notable Top 5%)] [Georgia Tech ARC Colloquium (Video)] [Poster] [Slides] [.bib]
• Fisher information lower bounds for sampling. Sinho Chewi, Patrik Gerber, Holden Lee, and Chen Lu. February 2023. [arXiv] [ALT 2023] [Slides] [.bib]
• On the complexity of finding stationary points of smooth functions in one dimension. Sinho Chewi, Sébastien Bubeck, and Adil Salim. February 2023. [arXiv] [ALT 2023 (Best Student Paper)] [.bib]
• Gaussian discrepancy: a probabilistic relaxation of vector balancing. Sinho Chewi, Patrik Gerber, Philippe Rigollet, and Paxton Turner. December 2022. [arXiv] [Discrete Applied Mathematics] [.bib]
• Variational inference via Wasserstein gradient flows. Marc Lambert, Sinho Chewi, Francis Bach, Silvère Bonnabel, and Philippe Rigollet. December 2022. [arXiv] [NeurIPS 2022] [UMass Amherst Reading Seminar on Mathematics of Machine Learning (Video)] [Poster] [Slides] [.bib]
• Improved analysis for a proximal algorithm for sampling. Yongxin Chen, Sinho Chewi, Adil Salim, and Andre Wibisono. July 2022. [arXiv] [COLT 2022] [COLT 2022 (Video)] [Georgia Tech ARC Colloquium (Video)] [Slides] [.bib]
• The query complexity of sampling from strongly log-concave distributions in one dimension. Sinho Chewi, Patrik Gerber, Chen Lu, Thibaut Le Gouic, and Philippe Rigollet. July 2022. [arXiv] [COLT 2022] [.bib]
• Towards a theory of non-log-concave sampling: first-order stationarity guarantees for Langevin Monte Carlo. Krishnakumar Balasubramanian, Sinho Chewi, Murat Erdogdu, Adil Salim, and Matthew Zhang. July 2022. [arXiv] [COLT 2022] [Slides] [.bib]
• Rejection sampling from shape-constrained distributions in sublinear time. Sinho Chewi, Patrik Gerber, Chen Lu, Thibaut Le Gouic, and Philippe Rigollet. March 2022. [arXiv] [AISTATS 2022] [.bib]
• Averaging on the Bures–Wasserstein manifold: dimension-free convergence of gradient descent. Jason Altschuler, Sinho Chewi, Patrik Gerber, and Austin Stromme. December 2021. [arXiv] [NeurIPS 2021 (Spotlight)] [NeurIPS 2021 (Video)] [Slides] [.bib]
• Efficient constrained sampling via the mirror-Langevin algorithm. Kwangjun Ahn and Sinho Chewi. December 2021. [arXiv] [NeurIPS 2021] [Poster] [.bib]
• Dimension-free log-Sobolev inequalities for mixture distributions. Hong-Bin Chen, Sinho Chewi, and Jonathan Niles-Weed. December 2021. [arXiv] [Journal of Functional Analysis] [Slides] [.bib]
• Optimal dimension dependence of the Metropolis-adjusted Langevin algorithm. Sinho Chewi, Chen Lu, Kwangjun Ahn, Xiang Cheng, Thibaut Le Gouic, and Philippe Rigollet. August 2021. [arXiv] [COLT 2021] [Sampling Algorithms and Geometries on Probability Distributions (Video)] [Poster] [Slides] [.bib]
• Fast and smooth interpolation on Wasserstein space. Sinho Chewi, Julien Clancy, Thibaut Le Gouic, Philippe Rigollet, George Stepaniants, and Austin Stromme. March 2021. [arXiv] [AISTATS 2021] [.bib]
• Exponential ergodicity of mirror-Langevin diffusions. Sinho Chewi, Thibaut Le Gouic, Chen Lu, Tyler Maunu, Philippe Rigollet, and Austin Stromme. December 2020. [arXiv] [NeurIPS 2020] [Slides] [.bib]
• SVGD as a kernelized Wasserstein gradient flow of the chi-squared divergence. Sinho Chewi, Thibaut Le Gouic, Chen Lu, Tyler Maunu, and Philippe Rigollet. December 2020. [arXiv] [NeurIPS 2020] [Slides] [.bib]
• Gradient descent algorithms for Bures–Wasserstein barycenters. Sinho Chewi, Tyler Maunu, Philippe Rigollet, and Austin Stromme. July 2020. [arXiv] [COLT 2020] [COLT 2020 (Video)] [Optimal Transport: Regularization and Applications 2020 (Video)] [TGDA@OSU 2020 (Video)] [Slides] [.bib]
• Matching observations to distributions: efficient estimation via sparsified Hungarian algorithm. Sinho Chewi, Forest Yang, Avishek Ghosh, Abhay Parekh, and Kannan Ramchandran. September 2019. [arXiv] [Allerton 2019] [.bib]
• A combinatorial proof of a formula of Biane and Chapuy. Sinho Chewi and Venkat Anantharam. March 2018. [arXiv] [E-JC] [.bib]

PhD thesis: An optimization perspective on log-concave sampling and beyond. 2023.