Generalized derivatives and nonsmooth optimization, a finite dimensional tour (with comments and rejoinder) (Q819950)

The author provides an interesting survey about well-known generalized differentiability notions of first- and of second-order for real-valued, vector-valued and set-valued functions: 1. the classical subdifferential and the classical directional derivative for convex functions, 2. the Clarke subdifferential and the Clarke generalized directional derivative for locally Lipschitz functions, 3. generalized subdifferentials and generalized directional derivatives via tangent and normal cones, 4. graphical derivatives and coderivatives, 5. the quasidifferential of Demyanov, 6. different kinds of second-order directional derivatives. In all sections, the author presents calculus rules (sum and chain rules) which are used for the formulation of necessary and sufficient optimality conditions of Fritz John and of Kuhn/Tucker type. In this way, the close relations between nonsmooth analysis and mathematical optimization are pointed out. At the end of the paper there are some comments from A. Daniilidis, St. Dempe, B. Mordukhovich, J.-P. Penot, A. Rubinov and A. Shapiro. Many references about the subject are suitable for deeper insights.