摘要
研究了一类具有可数状态空间的Markov控制过程在无限水平平均代价准则下的最优平稳策略问题.对此类过程,引入了折扣Poisson方程,运用无穷小矩阵和性能势的基本性质,导出了平均代价模型在紧致行动集上的最优性方程,并证明了其解的一个存在性定理.
The problems of optimal stationary policies are studied for a class of Markov control processes with countable state spaces and infinite horizon average-cost criteria.The discounted Poisson equation is introduced for these processes.Using the basic properties of infinitesimal generators and performance potentials,the optimality equation is given for the average-cost model on a compact action set,and an existence theorem of the solution to this equation is proved.
出处
《控制理论与应用》
EI
CAS
CSCD
北大核心
2005年第1期43-46,共4页
Control Theory & Applications
基金
国家自然科学基金资助项目(60274012)
安徽省自然科学基金资助项目(01042308).
关键词
可数Markov控制过程
性能势
平均代价准则
紧致行动集
最优平稳策略
countable Markov control process
performance potential
average-cost criteria
compact action set
optimal stationary policy