New second-order directional derivative and optimality conditions in scalar and vector optimization (Q1035899)

A new second order manner of computing the directional derivative is introduced in connection with the asymptotic second order tangent cone. It generates the derivatives named asymptotic second order directional derivatives of Hadamard and of Dini types. Some calculus properties are derived. Necessary and sufficient conditions are deduced for a general type of nonsmooth optimization problems in terms of parabolic and asymptotic second order derivatives of the objective function and second order tangent sets to the feasible set. The sufficiency requires a finite dimensional decision space, while the necessity holds in an arbitrary normed space. The results are applied to a general vector optimization problem, in which the partial order is defined by a convex cone with nonempty interior. A convenient scalarization is used to get necessary and sufficient optimality conditions in the vectorial case. It has second order directional derivatives at zero and various types of minimal elements to the vector problem are corresponding special kinds of solutions to the associated scalarized problem.