报告时间、地点：

报告时间：2026年1月16日，星期五，下午15:00

报告地点：华南理工大学 大学城校区，B10中座101

报告摘要：

       This talk presents our research findings from two key perspectives in exchange market design. First, from the perspective of incentive analysis, we examine participant behavior in peer-to-peer (P2P) exchange markets. Specifically, we consider a networked market where each agent allocates resources to its neighbors proportionally based on their contributions. While proportional sharing dynamics are known to converge to a market equilibrium, we investigate their strategic robustness as a resource allocation mechanism. A central question in Internet market design is whether agents can gain more resources by misreporting private information. In this work, we provide the first mathematical proof that this practical distributed sharing protocol is truthful against manipulation through weight misreporting and edge deletion, whether applied individually or in combination. Furthermore, by introducing the incentive ratio—defined as the percentage of utility after a Sybil attack relative to the benchmark—we quantify the strategic incentive for such attacks and establish a tight bound of two for general networks.

      Second, from the perspective of mechanism design, we study hierarchically structured exchange markets. Although market equilibrium theory highlights the benefits of free competition, real-world markets require regulations and rules to ensure stability and operability, often resulting in a hierarchical structure: an upper-level operator sets trading rules, and lower-level prosumers exchange resources within that framework. We propose a novel two-stage model to capture this hierarchy. In the first stage, agents report private information to the operator, who then designs market rules via a specified mechanism. In the second stage, agents trade freely under these rules until reaching a stable outcome. We adapt the concept of liquid welfare from budget-feasible auctions to exchange markets as a metric for mechanism performance and introduce profitability to assess the market’s financial sustainability. Our goal is to design truthful mechanisms that achieve (approximately) optimal welfare. Our main contributions are: (1) For large markets with many participants, we design a truthful and profitable mechanism that with high probability achieves a tight approximation ratio of 1/2. (2) For general markets, we first show an impossibility result: no mechanism can simultaneously guarantee truthfulness, profitability, and a constant approximation ratio. We then design a truthful mechanism that attains a 1/2-approximation. This work highlights the critical role of regulatory design in decentralized trading and advances the understanding of mechanism design for structured resource exchange.

       主讲人简介：

       程郁琨：博士，博士生导师，江南大学商学院教授。中国运筹学会理事兼副秘书长、中国工业与应用数学学会区块链专委会秘书长、中国计算机学会计算经济学专委会副主任，江苏省运筹学会理事，担任《运筹学学报》《运筹与管理》《Journal of Dynamics and Games》《Blockchain》等期刊编委。主要研究领域为组合优化、博弈论、运筹优化等。曾主持国家自然科学基金项目4项，省部级科研项目2项。在运筹学领域国际顶级刊物MOR，博弈论领域顶级期刊GEB，计算机科学领域重要国际期刊IEEE TCC、IEEE TC、TCS，计算经济学领域国际顶级会议EC、WINE、WWW，区块链领域顶级国际会议FC 2022，人工智能领域顶级会议IJCAI上发表高水平学术论文80余篇，获得发明专利2项。江苏省“科技副总”，江苏省高校“青蓝工程”中青年学术带头人，江苏省高校“青蓝工程”优秀青年骨干教师。获得江苏省第十八届哲学社会科学优秀成果奖三等奖，2021年度江苏省高等学校科学技术研究成果奖三等奖。无锡市委二十届三中全会精神宣讲团、二十届四中全会宣讲团成员。