Many recent dynamic programming specifications fail to satisfy traditional contractivity conditions, which are a cornerstone of the standard optimality theory for infinite horizon problems in discrete time. We formulate alternative conditions based around monotonicity and “value” convexity. These conditions lead to an optimality theory that is as strong as the contractive case. Several applications are provided.

中文翻译：

---

具有值凸性的动态编程

许多最新的动态编程规范都无法满足传统的可收缩性条件，这是离散时间中无限层级问题的标准最优性理论的基石。我们根据单调性和“值”凸性制定替代条件。这些条件导致了一个与紧缩情况一样强大的最优性理论。提供了几个应用程序。