Optimality conditions in optimization problems with convex feasible set using convexificators (Q2408896): Difference between revisions

Following the works of \textit{J. B. Lasserre} [Optim. Lett. 4, No. 1, 1--5 (2010; Zbl 1180.90237)] and \textit{J. Dutta} and \textit{C. S. Lalitha} [ibid. 7, No. 2, 221--229 (2013; Zbl 1267.90096)], this paper considers an optimization problem with inequality constraints. The assumption is that the constraint set itself is convex, while the constraint functions themselves may be nonconvex, nondifferentiable and even discontinuous. In their analysis, the authors make use of convexificators. Under some conditions, the authors show that the normal cone of the constraint set can be represented by means of upper regular convexificators of the constraint functions. This representation is in turn used to derive Karush-Kuhn-Tucker (KKT) optimality conditions, and these extend some results in [Dutta and Lalitha, loc. cit.], in Lasserre [loc. cit.], and in [\textit{J. E. Martínez-Legaz}, Optim. Lett. 9, No. 5, 1017--1023 (2015; Zbl 1356.90108)]. When the constraint set has a nonempty interior, and under some constraint qualification, the authors obtain KKT conditions from the Fritz-John conditions. Under some additional assumptions, the authors obtain boundedness of the set of KKT multipliers and derive a characterization for the solution set. Finally, they illustrate their results via a linear semi-infinite problem.