教育部人文社科重点研究基地中央财经大学中国精算研究院学术活动

精算论坛第292期讲座

(2026年10月23日)

报告一：Successive approximations for stochastic optimal control problems

报告摘要：Based on the stochastic maximum principle, a modified method of successive approximations (MSA for short) is established for stochastic recursive optimal control problems. The second-order adjoint processes are introduced in the augmented Hamiltonian minimization step.We give a delicate proof of the error estimate and then prove the convergence of the modified MSA algorithm.

报告人简介：嵇少林现为山东大学金融研究院教授，师从彭实戈院士。1999年至今在山东大学工作。研究领域为金融数学、金融经济学、随机优化和非线性期望理论。嵇少林与彭实戈、Larry Epstein、周迅宇等合作者在《Review ofFinancialStudies》, 《OperationsResearch》,《ProbabilityTheory and theRelatedFields》和《SIAM Control and Optimization》等杂志上发表了一系列的成果。对金融市场中的学习理论、资本资产定价、随机优化问题和非线性期望理论进行了系统的研究。

报告二：Parisian stopping in Merton's problems

报告摘要：In this talk, I will present recent work on Parisian stopping in Merton's problems. The Parisian stopping criteria is described as if the wealth remains continually under the barrier over a pre-specified time period, then the investment on the risky assets ends up. There is a separate clock set up to record the total time that the wealth has passed below the barrier. Therefore one more dimension (clock) will be increased in the domain with wealth below the barrier. This leads to a two-three-dimensional system of nonlinear HJB equations. The dual methods are developed to convert the two-three-dimensional system of nonlinear HJB equations into a linear PDE system with free boundaries and the verification theories are rigorously established. After approximating the free boundary by a piecewise constant function, a recursive moving window algorithm is developed to obtain the closed-form solution to the linear system and then the solutions to the prime problems are attained by the exact conjugate relation. Numerical examples are carried out to verify the effectiveness of the methods and to compare the performance of the investment with and without Parisian stopping criteria.

报告人简介：马敬堂，西南财经大学数学学院光华英才特聘教授、博士生导师、院长，教育部新世纪优秀人才,四川省学术与技术带头人。现任四川省数学会副理事长，中国运筹学会理事，中国运筹学会金融工程与金融风险管理分会副理事长。主要研究方向为：金融数学、随机控制计算。在SIAM Journal on Control and Optimization, Mathematics of Operations Research, European Journal of Operational Research, Insurance: Mathematics and Economics等期刊发表论文。

报告三：Model-freereinforcementlearning forcontinuoustime andstate

报告摘要：We develop a model-free reinforcement learning (RL) algorithm based on the stochastic maximum principle for continuous-time stochastic control problems with continuous state and action spaces. For a parameterized Markovian policy, we establish the existence of the decoupling field for the adjoint backward stochastic differential equation, which allows the Hamiltonian gradient to be represented as a deterministic function of time and state. We then learn this Hamiltonian gradient directly from data and incorporate it into a policy gradient scheme with inexact gradients. We establish the convergence of the resulting policy gradient algorithm and establish the corresponding error estimates. Under suitable conditions on learning rates, exploration parameters, and approximation errors, the objective values converge andthe $L^2$-norm of the policy gradient vanishes asymptotically. We further show that the proposed SMP-based policy gradient representation is equivalent to the existing continuous-time deterministic policy gradient representation based on the advantage-rate function. The effectiveness and efficiency of the proposed RL algorithm are demonstrated by numerical experiments.

报告人简介：薄立军，西安电子科技大学数学与统计学院教授，概率与数理统计专业博导、本科毕业于西安电子科技大学数学系、分别于2006年和2009年获南开大学概率论与数理统计专业理学硕士和理学博士学位。主持国家自然科学基金面上项目3项、陕西数理基础科学研究重点项目、中国科学院前沿科学重点研究计划项目；牵头获得陕西省自然科学奖二等奖、陕西高等学校科学技术研究优秀成果奖特等奖和陕西省教学成果奖二等奖; 在概率统计、随机控制和金融数学等领域权威学术期刊《Automatica》《Ann. Appl. Prob.》《Math. Oper. Res.》《Production Oper. Manag.》《Math. Finan.》《Science China: Math.》《SIAM J. Contr.Optim.》《SIAM J. Finan. Math.》等发表论文70余篇。出版教材《随机过程》《哈佛概率公开课》（译著）《最优化模型》（译著）和《高等概率论》（科学出版社“十四五”高等学校本科规划教材）。

报告四：Representation theorems for temporal copula families and corresponding Markov processes

报告摘要：Using the copula method, we develop a unified framework for studying the temporal dependence structure of Markov processes based on their connection with consistent temporal copula families. First, representation theorems for homogeneous consistent temporal copula families are presented under different continuity conditions, and the probability structure of the generated Markov processes is provided. Building on these results, we introduce a methodology for constructing Markov martingales by adding marginal information to the generated Markov processes. Inparticular, self-similar fake alpha-stable Levy motions are constructed with fake Brownian motions as their special cases. Finally, the long-term dependence in the generated Markov processes is discussed. We obtain sufficient conditions for discrete-time Markov processes to be beta-mixing and geometrically beta-mixing, and prove that these properties can be extended to continuous-time Markov processes.(It is a joint work withLiuleiSun,QiliangHuangandHengxuanCheng)

报告人简介：杨静平，北京大学数学科学学院教授，博士生导师，国家二级教授。研究兴趣有金融和保险中的风险相依性、风险度量、信用风险管理以及资产支持证券等。在金融数学期刊Mathematical Finance、Finance and Stochastics、SIAM Journal onFinancial Mathematics、Journal of Computational Finance、精算学期刊Insurance：Mathematics and Economics、ASTIN Bulletin、ScandinavianActuarial Journal、North American Actuarial Journal以及概率论期刊Bernoulli等发表了多篇学术论文。主持完成了中国国债发行策略的随机模拟模型、国债收益率曲线的拟合、信贷资产证券化、含权债估值模型等金融行业的课题。完成《寿险精算基础》和《非寿险精算学》等教材。获2023年国家级教学成果二等奖（2/7）。

讲座时间：2026年10月23日（周五） 下午13:30-17:00

报告地点：学院南路校区 学术会堂606

邀请人：池义春