High-order tangent cones and their application in optimization (Q4374201)

The paper is concerned with the minimization of functionals in Banach spaces under both equality and inequality constraints, where the equality constraints are determinated by the set \(Q= \{x\in X: F()= 0\}\) in which \(F: X\to Y\) is an operator between Banach spaces. A generalization of the classical Lyusternik theorem determining the structure of high-order approximations to \(Q\) at a point \(x^*\) when \(F'(x^*)\) is not onto is presented but proved elsewhere, \(F'\) being the Fréchet derivative of \(F\). The result works when the Lyusternik condition is not satisfied. A precise characterization of \(p\)-order polynomial approximations to \(Q= \{x\in X: F(x)= F(x^*)\}\) when \(F'(x^*)\) is not onto is furnished when a suitable operator \(G_p\), which takes into account derivatives up to and including order \(p\), is onto. Then \(p\)-order tangent sets and cones are defined and their structure is determined. This result, combined with \(p\)-order, approximating sets to other types of constraints, provides a tool to derive nontrivial necessary conditions for optimality. The result obtained applies also to some of the so-called abnormalities of second-order and to a variety of extremum and optimal control problems where classical theorems cannot be applied. Some applications are also given.