| Q（f）－过程非唯一时连续时间折扣目标MDP |
| |
| 引用本文： | 郭先平.Q（f）－过程非唯一时连续时间折扣目标MDP[J].湖南师范大学自然科学学报,1996(3). |
| |
| 作者姓名： | 郭先平 |
| |
| 作者单位： | 湖南师范大学数学系 |
| |
| 基金项目： | 国家自然科学基金,湖南省自然科学基金 |
| |
| 摘 要： | 考虑的是可数状态空间连续时间ＭＤＰ的折扣模型，与以往不同的是，我们放弃了由策略ｆ所确定的Ｑ（ｆ）－过程唯一的传统假设．而首次考虑Ｑ（ｆ）－过程非唯一的情形，借助于Ｑ－过程的构造理论，用拓扑分析的方法，证明了最优策略的存在性．
|
| 关 键 词： | 连续时间MDP 非唯一 折扣目标 最优策略 |
Countinuous Time MDP with Discounted Reward Criterion-the Case of Q(f)-Processes being not Unique |
| |
| Guo Xianping.Countinuous Time MDP with Discounted Reward Criterion-the Case of Q(f)-Processes being not Unique[J].Journal of Natural Science of Hunan Normal University,1996(3). |
| |
| Authors: | Guo Xianping |
| |
| Abstract: | We consider MDP discounted model with countable state space.The difference between this paper and others is that we give up the conditions that Q(f)-processes decided by policy f are unique in this paper,and first probe into the case of Q(f)-processes being not unique.By the structure theory of Q(f)-processes,we prove the existence of optimalpolicies in topological and analytic method. |
| |
| Keywords: | countinuous time MDP not uniqueness discounted reward criterion optimal policies |
| 本文献已被 CNKI 等数据库收录！ |
