承办单位：东北财经大学管理科学与工程学院

报告题目（一）：量子计算驱动的两阶段大类资产投资组合优化研究

报告人：余乐安

【报告人简介】

余乐安，四川大学特聘教授、博士生导师，国家高层次人才计划入选者、国际系统与控制科学院院士、国际信息技术与量化管理学会会士、亚太人工智能学会会士。

【报告摘要】

本报告主要介绍量子计算在大类资产投资组合优化的基本框架、关键模型与核心技术。报告首先简要介绍大类投资组合优化的研究背景与科学问题；其次，针对大类资产投资组合中的科学问题，重点讲述基于量子退火算法的两阶段大类资产投资组合优化总体框架、关键优化模型及其核心量子求解技术；再次，为验证所提两阶段投资组合模型及量子技术求解方法的有效性，采用美国股市及三类国际大宗商品相关数据，对关键优化模型及量子求解技术进行了实证分析。最后，对研究成果进行总结，并阐述未来发展方向。

报告题目（二）：From ORLM to Decision Systems: Self-Evolving LLMs for Optimization and Beyond

报告人：葛冬冬

【报告人简介】

葛冬冬，上海交通大学智能计算研究院院长、安泰经济与管理学院特聘教授，杉数科技联合创始人兼首席科学家。研究方向为大规模数学优化算法与理论，主持国家自然科学基金杰出青年项目、重大项目及原创探索项目。

【报告摘要】

Optimization modeling is central to operations research, yet translating real-world problems into correct mathematical formulations remains a major bottleneck. Recent advances in large language models (LLMs) offer a new pathway, but challenges in reliability, scalability, and continual improvement persist.

In this talk, we present a unified framework for self-evolving LLM-based decision systems. We develop training-based approaches (ORLM, Solver-Informed RL, StepORLM), extend modeling to dynamic programming, and analyze intrinsic limits via OPT-Engine. Moving beyond static models, we introduce agentic systems with solver-grounded validation, decentralized debate, and memory, enabling reliable and reusable decision-making. These results suggest a new paradigm in which LLMs evolve from generators into verifiable, adaptive decision systems for operations research and beyond.

报告题目（三）：复杂装备可靠性评估与维护决策

报告人：刘宇

【报告人简介】

刘宇， 博士， 二级教授、博士生导师，国家高层次人才计划入选者，装备可靠性学科创新引智基地（国家111计划）主任，四川省装备可靠性国际联合研究中心主任、电子科技大学教务处处长。

【报告摘要】

本报告围绕复杂装备研制阶段、运行阶段和维护阶段面临的“信息不精确”“数据多来源”“运维高成本”可靠性共性难题，介绍团队在复杂系统不确定性量化与可靠性评估、复杂系统剩余寿命预测与动态可靠度评估、复杂系统（集群）/基础设施维护级韧性提升决策方面的研究成果。

 

报告题目（四）：A Learning and Rectification Algorithm for Nonstationary Online Linear Programming

报告人：陈彩华

【报告人简介】

陈彩华，国家重点研发计划首席科学家、国家优秀青年基金获得者、国家自然科学基金重大项目课题负责人、美国斯坦福大学访问学者，现任南京大学教授、博士生导师、工程管理学院副院长、民建江苏省委大数据与人工智能委员会主任。

【报告摘要】

This talk investigates online linear programming (OLP) under resource capacity constraints, aiming to maximize total reward over a finite horizon. To address the limitations of stochastic models in nonstationary settings and the conservatism of adversarial models, we introduce a globally nonstationary and locally stationary arrival model to the OLP context. This framework partitions the horizon into known periods where requests are independent and identically distributed within each period, while the underlying distribution varies across periods, balancing environmental evolution with local stability.

Assuming the true distribution is unknown, but an inaccurate prior is available, we propose the Gradient Descent with Learning and Rectification (GDLR) framework. This dual-based method optimizes budget consumption for each request through two complementary components: a learning component that leverages local stationarity to refine estimates by combining prior knowledge with real-time observations, and a rectification component that counters global nonstationarity by adaptively adjusting estimates based on cumulative resource imbalance.

Theoretically, we demonstrate that rectification serves as robust optimization against estimation errors in dual and ensures primal feasibility. The regret bound of GDLR decomposes into intrinsic stochasticity, estimation error, and allocation suboptimality, with the latter two significantly reduced via learning and rectification. Empirically, our approach consistently outperforms baselines on synthetic and real-world datasets. To our knowledge, this work presents the first enterprise OLP deployment at Alipay, achieving a 2.2% average revenue gain in online A/B tests.

 

撰稿：赵永丽 朱晗 审核：吴志樵 印明鹤 单位：管理科学与工程学院