We develop a systematic mathematical framework for dynamic Bayesian persuasion through the lens of martingale transport. Under suitable regularity conditions, we establish strong duality and characterize optimal sequences of posterior distributions via contact conditions. Building on these results, we develop an entropy-regularized algorithm that combines posterior discretization with backward recursion. At each stage, the algorithm solves a convex dual optimization problem and explicitly recovers the corresponding regularized optimal transition probabilities. We characterize the algorithm’s approximation error in terms of the time horizon, discretization resolution, and logit error. Numerical experiments in entertainment design and information-constrained sales campaigns demonstrate the algorithm’s computational efficiency relative to existing methods. This talk is based on a joint work with Zhenyu Shen (CUHK) and Nizar Touzi (NYU).

Biography

Nan Chen is a Professor of Systems Engineering and Engineering Management at the Chinese University of Hong Kong (CUHK). His research interests lie in financial engineering and FinTech, with a focus on reinforcement learning, quantitative modeling in finance and risk management, Monte Carlo simulation, and applied probability. Prof. Chen earned his Ph.D. in Operations Research from Columbia University in 2006, following his M.S. and B.S. degrees in Probability and Statistics from Peking University in 2001 and 1998, respectively. He has served as an associate editor for Operations Research Letters (2007–2008) and is currently an associate editor for Operations Research, Mathematical Finance, International Review of Finance, and Digital Finance. At CUHK, Prof. Chen serves as the Directors of the Bachelor of Engineering Program and Master of Science Program in Financial Technology. He also directs the Master of Science Program in Financial Engineering at CUHK Shenzhen.