Enhanced Karush-Kuhn-Tucker conditions for mathematical programs with equilibrium constraints (Q481046)

The authors provide generalized Karush-Kuhn-Tucker conditions and associated constraint qualifications for nonsmooth optimization problems with equilibrium constraints according to \[ \begin{gathered} \min f(x)\quad\text{s.t.}\quad x\in X,\quad h_i(x)= 0,\quad i= 1,\dots, p,\quad g_j(x)\leq 0,\quad j= 1,\dots, q,\\ G_l(x)\geq 0,\quad H_l(x)\geq 0,\quad G_l(x)H_l(x)= 0,\quad l= 1,\dots, m.\end{gathered} \] The functions \(f,h_i,g_j: \mathbb{R}^n\to\mathbb{R}\) are assumed to be Lipschitz continuous (not necessary smooth), the functions \(G_l,H_l: \mathbb{R}^n\to \mathbb{R}\) are assumed to be continuously differentiable and \(X\subset\mathbb{R}^n\) is assumed to be a closed set. All conditions are formulated in terms of the limiting subdifferentials of the functions \(f\), \(h_i\), \(g_j\) and the limiting normal cone of the set \(X\). First it is pointed out that the linear independence constraint qualification is not a suitable constraint qualification for such optimization problems if the objective function is nonsmooth. Hence a new stationarity condition (the so-called enhanced Mordukhovich stationarity condition) and some corresponding constraint qualifications are introduced. The authors discuss relations between these conditions and relations to the existence of local error bounds for the feasible set.