报告题目：Dynamic Pricing Strategies in Hub-and-Spoke Networks: Structural Characteristics and Theory-Driven AI Agents

报 告 人：萧柏春

报告时间：2026年10月9日（星期五）10:00-11:30

报告地点：明哲楼517

主办单位：东北财经大学现代供应链管理研究院

【报告人简介】

Dr. Baichun Xiao graduated from the Department of Mathematics at Nanjing University in 1982. He earned an MBA from the University of Leuven in Belgium in 1985 and a PhD in Management Science from the Wharton School of the University of Pennsylvania in 1990. From 1990, he taught at the School of Business at Seton Hall University in the Department of Decision and Computing Sciences, where he served as Assistant Professor, Tenured Professor, Department Chair, and Full Professor. From 1998 to 2024, he worked at the School of Management at Long Island University as a Full Professor, Chair of the Department of Management, and Senior Full Professor.

Professor Xiao's research interests include service management, revenue management, supply chain and logistics management, and non-smooth function optimization. His research has been published in leading international academic journals such as Management Science, Operations Research, European Journal of Operational Research, and Decision Sciences. He has also served as a visiting professor at numerous universities, including Peking University, Tsinghua University, Fudan University, Nanjing University, Sichuan University, the Chinese University of Hong Kong, and City University of Hong Kong.

Professor Xiao has held various leadership positions, including serving as a member of the Board of Trustees at Nanjing University, an independent director at Hong Kong H&H Group, Director of the Service Management Research Institute at Sichuan University, and Director of the Sino-US Service and Operations Management Research Center at Southwest Jiaotong University.

【摘要】

This paper investigates dynamic bid-price strategies within a capacitated hub-and-spoke network, where $n$ spoke cities funnel traffic to a single hub. This topology gives rise to complex inventory interactions, encompassing both complementarity (e.g., connecting flights) and competition (e.g., between local and connecting passenger flows). The study demonstrates that the optimal value function exhibits coordinate concavity and simultaneously displays a mix of submodularity and supermodularity. These fundamental structural properties rigorously characterize the bid prices—representing the shadow costs of seat capacity—across the network. Building on these theoretical insights, the paper proposes a "Theory-Informed AI" framework for automated revenue management agents. This approach employs constrained deep learning to estimate unknown demand intensities while strictly adhering to the proven geometric properties of bid prices; this effectively mitigates the computational challenges associated with high dimensionality and yields robust, stable pricing strategies for high-dimensional networks.