Sufficient conditions for error bounds and applications (Q1884732)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Sufficient conditions for error bounds and applications
scientific article

    Statements

    Sufficient conditions for error bounds and applications (English)
    0 references
    0 references
    0 references
    0 references
    5 November 2004
    0 references
    The authors develop sufficient conditions in general spaces for error bounds in terms of Fréchet and limiting Fréchet derivatives. The obtained results are then applied to optimization problems. The paper contains first a short introduction and a section for notations and preliminaries. The main theoretical results are given in the sections 3 and 4. The sufficient conditions found in the third and fourth sections are stated using Fréchet and approximate subdifferentials respectively. The final section is devoted to the applications. The first important result given in the section 3 is a sufficient condition ensuring a local error bound for the constraint system \(f(x,y)\leq 0\), where \(f:X \times U \longrightarrow \mathbb{R}\cup\{+\infty\}\) is a lsc function, and \(X\) and \(U\) are Banach and metric spaces respectively. In case \(f\) does not depend on \(u\) a sufficient condition for the existence of a global error bound is also presented. The second kind of constraint system considered in the third section is \(x \in C\) and \(g(x,u) \in D\), where \(C\) and \(D\) are closed sets in \(X\) and \(\mathbb{R}^{m}\) respectively and \(g:X \times U \longrightarrow \mathbb{R}^{m}\) is a mapping. In this case a sufficient condition for the existence of local error bounds is given. In the fourth section a parametrized system of the form \((x,y) \in C \times E\) and \(g(x,y,u)=0\) is considered. Here \(C\) and \(E\) are closed sets in \(X\) and \(Y\), \(g:X \times Y \times U \longrightarrow Z\) is a mapping and \(Y\) and \(Z\) are Banach spaces. For this system and some particular cases of it sufficient conditions for local error bound are given. In the last section the theoretical results obtained are applied to optimization problems. First the nonemptyness and weak-star compacity of the Fritz-John and Karush-Kuhn-Tucker multiplier sets is stablished. As a second application it is proved that the value function of a parametric optimization problem is Lipschitz continuous and its subdifferential is computed. Finally some results concerning the local controllability in control problems of nonconvex unbounded subdifferential inclusions are obtained.
    0 references
    0 references
    0 references
    0 references
    0 references
    error bounds
    0 references
    sensitivity analysis
    0 references
    subdifferential
    0 references
    local controlability
    0 references
    0 references