多阶段决策过程（ multistep decision process ）是指 这样一类特殊的活动过程，过程可以按时间顺序分解成若干个相互联系的阶段，在每一个阶段都需要做出决策，全部过程的决策是一个决策序列。 动态规划 （ dynamic programming ）算法 是解决 多阶段决策过程最优化问题 的一种常用方法，难度比较大，技巧性也很强。利用动态规划算法，可以优雅而高效地解决很多贪婪算法或分治算法不能解决的问题。动态规划算法的基本思想是：将待求解的问题分解成若干个相互联系的子问题，先求解子问题，然后从这些子问题的解得到原问题的解； 对于重复出现的子问题，只在第一次遇到的时候对它进行求解，并把答案保存起来，让以后再次遇到时直接引用答案，不必重新求解 。动态规划算法将问题的解决方案视为一系列决策的结果，与贪婪算法不同的是，在贪婪算法中，每采用一次贪婪准则，便做出一个不可撤回的决策；而在动态规划算法中，还要考察每个最优决策序列中是否包含一个最优决策子序列，即问题是否具有最优子结构性质。

-multi-stage decision-making process (multistep decision process) is that like a special kind of process, the process can be time-sequence decomposed into a number of interrelated stage, in every stage of a need to make a decision, all of the decision-making process is a sequence of decision-making. Dynamic Programming (dynamic programming) algorithm is a multi-stage decision-making process optimization asked that's a common method was more difficult, skills also have a strong character. Using dynamic programming algorithm can be elegant and efficient solution for many greedy algorithm or partition algorithm to solve the problem. Dynamic programming algorithm is the basic idea : to be solving the problem is decomposed into a number of interrelated issues son, Solving the problem first s