Dynamic programming method is used to solve optimization problems with recursive nature, ie finding the optimal method for problems that can lead to find the optimal scheme of a finite number of subproblem. For many recursive algorithms, the principle of divide and rule (spill and Conquer) often play a key role in the design of algorithms. To solve a big problem, I divided it into the same problem with the format to be able to solve it independently. In the dynamic programming method, This principle has been demonstrated: When not know what needs to solve the problem yet, I will go to solve all human problems and storing the solution or answer them for the purpose of reuse in a particular combination to address the more general problem. And that is the difference between planning and recursion and content is also dynamic programming method.

+Recursion starting big problem decays into the problem and solve every problem that children. The award once again brought the problem of children allowed to decay into smaller subproblem and went smaller solve it regardless of whether or not it was the.
+Planning starts from solving all the smallest problem (Base problem) so that gradually solve the larger problem, until the largest solved (original problem).

When using dynamic programming method to solve the problem, we may face difficulties following two:

– One is, not always a combination of subproblem solution also for the larger problem solution

– Second, number of problems to be solved and the answer can store enormous, unacceptable. Hitherto, no one has identified precisely the lessons that can be solved efficiently using dynamic programming method. There are problems too complex and difficult that can not seem to dynamic programming applications to solve, while there are more than simple math makes use planning to address the less efficient than using classic algorithms.

The basic principles of action plan
Action planning is the planning stages of the multi-stage process in which each stage will only optimize a step. But when planning a multi-stage process, Every step must be selected on the basis of the control is not derived from the narrow interests of the steps that the common interests of the entire process.
1. Principle numbered stages from the bottom up.
For the last stage can make it best and not worry about consequences. When this stage it becomes unstable and can be reviewed in the previous period and continued until we went up to the first phase of the process.
2. The principle problem of chemical parameters
At this stage of the final stage monitoring, I do not know the results, so we have to assume at this stage and with the assumption that we find optimal control for the final stage. In another step situation happens so. Thus the optimal control will depend on the specific parameters for the results in the previous step.
3. Principle cage
Cages original problem into a wider problem or a problem they and thus the original problem was a case of their own this problem.
They have this problem because the parameters so we solved. I will try the result of problems with different parameters until a time to be informed of the initial problem is stopped.
4. Optimization guidelines (Bellman)
Optimal posture is nature: whether the initial state and the initial control how they form the next controller is also optimal for state results obtained in the control effect initially.