摘要
本文讨论了一类非时齐马氏决策模型向时齐马氏决策模型的转化问题，从而得到若πｔ是非时齐模型的ｋ（ｋ≥１）矩ε（ε≥０）最优策略，则由其导出的时齐策略是相应的时齐模型的ｋ（ｋ≥１）矩Ｓｉ－ε（ε≥０）最优策略的结论．
in this paper, it has discussed the transformation of from a non-stationarydiscounted Markovian decision model to stationary Markvian decision model. It′s proved that if π′ is a k(k≥1) moment ε(ε≥0) optimal policy, then the policy π whitch is induced from π′ is the corresponding stationary model′s k (k≥1) moment St-ε (ε≥0)optimal policy.
关键词
非时齐
马氏决策模型
矩优化
non-stationary, Markvian decision model, moment optimal
