| not the full distribution information, of the random variables involved in the given stochastic programming problems. Then a sample problem can be formed. The optimal solution x N of this sample program is a point estimation of the unknown ``true" optimal solution x * of the original stochastic program. Then we have to make a statistical inference for the true solution. The methods of making statistical inference in classical statistics do not apply for stochastic programs. In this paper we study how this kind of statistical inference can be made. We will construct the confidence regions(including confidence intervals) for $x^{*}$ and the linear form v'x *. |
|
|
