Characterization of strong stability for C-stationary points in MPCC (Q2429468): Difference between revisions

Let \(f,h_{i},g_{j},F_{1,m},F_{2,m}:\mathbb{R}^{n}\longrightarrow \mathbb{R}\), for \(i\in I\), \(j\in J\) and \(m=1,\dots,k\), with \(\left| I\right| \leq n\) and \(J\) finite, be of class \(C^{2}\). Then the authors consider the problem of minimizing \(f(x)\) subject to \(h_{i}(x)=0\) \((i\in I)\), \(g_{j}(x)\geq 0\) \((j\in J)\), and \(F_{1,m}(x)F_{2,m}(x)=0\), \(F_{1,m}(x)\geq 0\), \(F_{2,m}(x)\geq 0\) \((m=1,\dots,k)\). A feasible point is said to be C-stationary if it satisfies the KKT necessary optimality conditions. The main result characterizes strong stability in the sense of \textit{M. Kojima} [Proc. Symp., Univ. Wis. 1979, 93--138 (1980; Zbl 0478.90062)] for C-stationary points satisfying the Linear Independence Constraint Qualification. The characterization involves a nondegeneracy property of the complementarity conditions relative to the constraints \(F_{1,m}(x)\geq 0\), \(F_{2,m}(x)\geq 0\) as well as certain matrices which contain first and second order information on the defining functions being nonsingular with the same determinant sign.