Thus, in ยง 1 states as # can be converted multiplicative performance indicator additive. Sometimes having aftereffects can be overcome by introducing an additional state variable. For example, if the income of the k-th step, the problem is not only one of the allocated funds and the remainder of xk to the top step | / g_ but from a certain value of UC \, depending on the previous period, b = F (lh- b xk, yk-1). In that case, we shall characterize the state of the system (k-1) th step, the two parameters of the state and = yk-\ -
The same remark concerns the decomposition process into steps, which sometimes follows naturally from the properties of the process, and sometimes introduced artificially. All of this suggests that the formulation of the problem in the form of a model DP sometimes requires the use of artificial methods. With some of them we will introduce in the following chapters.
Building a model DP is reduced to the following main points:
1) Select the method of dividing the process into steps; __
2) Input parameters of the state lk == (lk \
and control variables uk = (u (k \ u {k \ ..., ais)) at each step of the process;

In terms of volume calculations both schemes are the same, but with some additional research is preferable to one or the other.
Copyright ©