Perturbation theory of nonlinear programs when the set of optimal solutions is not a singleton (Q1108933)

The present article continues previous articles of the author about sensitivity analysis in nonlinear programming problems. A parametric mathematical programming problem is considered with more parameters (x) in the objective function and also in the feasible set. To this problem an optimal value function \(\phi\) (x) and an optimal solution set multifunction M(x) is associated. Local properties of the function \(\phi\) (x) and the multifunction M(x) are investigated in a firm point \(x_ 0\), when a set of optimal solution \(M(x_ 0)\) is not one point. The author shows that under certain regularity conditions the multifunction M(x) is upper Lipschitzian at \(x_ 0\) and the function \(\phi\) (x) possesses second-order directional derivatives at \(x_ 0\). As an illustrative example the smallest eigenvalue of a symmetric matrix is investigated.