For the problemP(λ): Maximizec T z subject toz∈Z(λ), whereZ(λ) is defined by an in general infinite set of linear inequalities, it is shown that the value-function has directionalderivatives at every point \(\ba...
详细信息
For the problemP(λ): Maximizec T z subject toz∈Z(λ), whereZ(λ) is defined by an in general infinite set of linear inequalities, it is shown that the value-function has directionalderivatives at every point \(\bar \lambda \)such thatP( \(\bar \lambda \) ) and its dual are both superconsistent. To compute these directionalderivatives a min-max-formula, well-known in convex programming, is derived. In addition, it is shown that derivatives can be obtained more easily by a limit-process using only convergent selections of solutions ofP(λ n ), λ n → \(\bar \lambda \)and their duals.
暂无评论