Therefore, the methods you can use DP to solve some problems of integer programming. As such, most of the problems discussed above, relates to the problem of integer programming. However, the DP methods require additive (or multiplicative) of the objective function and are very sensitive to the number of constraints of the problem.
Funds go distributed between the two enterprises within n years. X means invested in the company in the beginning of the year I, bringing the year-end perk f \ (x), and return to the amount of f! (X).
Consider the I-step task allocation. Let lh-i - state of the system at the beginning of a k-step (k = l, 2, ..., n). For the stochastic model under the influence of a fixed state control lk-i goes into random state Ik, when it should be a distribution of the random variable
Control u at k-th step should be considered accidental, and the objective function of the stochastic problem is a random variable. Therefore decided to optimize the expectation of the objective function.
Thus, the optimization problem in the stochastic case is reduced to a set of values to determine the control variables based on the condition of the expectation optimization objective function. A feature of such problems is the fact that, in practice, can not decide on the choice of control if you do not know the state of the system to the top step, that is, before making another decision, we must use the knowledge implementation of the random variables that have been observed before.