Primal and dual stability results for variational inequalities (Q1841552)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Primal and dual stability results for variational inequalities
scientific article

    Statements

    Primal and dual stability results for variational inequalities (English)
    0 references
    0 references
    0 references
    18 February 2001
    0 references
    It is considered the variational inequality problem (\text{VI}): find \(\overline {x} \in C\) and \(x^* \in A(\overline x)\) satisfying: \((x- \overline{x})^Tx^* \geq 0\) \(\forall x \in C\), where \(A:\mathbb{R}^n\to 2^{(\mathbb{R}^n)}\) is a maximal monotone operator, and \(C= \{x \in \mathbb R^n \mid f_i(x) \leq 0\), \(i=1 \dots n\), \(Lx=l\) \}, with \(f_i\) a closed proper convex function from \(\mathbb{R}^n\) to \( \mathbb{R} \bigcup {\infty},\) for each \(i\) and \(L-(q,n)\)-matrix , \(l \in \mathbb{R}^q\). Other assumptions are established on \(C\), one of them: \(C \neq \emptyset\) and subset of \(\operatorname {dom} A\) (domain of \(A\)). The solution set of (VI) is denoted by \(S\). The authors investigate the continuous dependency of solutions of (VI) with respect to perturbations of the data in the following way: \(A=A+ \varepsilon \widetilde{A}\), \(f=f + \varepsilon \widetilde{f}\) , \(L=L+ \varepsilon \widetilde{L}\) and \(l=l+ \varepsilon \widetilde{l}\), where \(\widetilde{A}\) is a maximal monotone operator, \(\widetilde{f_i}\) are closed proper convex functions and \(\widetilde{L}\) is a \((q,n)\)-matrix, \(\widetilde{l} \in \mathbb{R}^n\). The set \(C(\varepsilon)\), the perturbed variational inequality \((\text{VI}(\varepsilon)),\) its solution set \(S(\varepsilon)\) and the multi-valued \(S(.)\) are defined analogously. The concepts of \(\text{VI}(\varepsilon)\)-regular and \(\text{VI}(0)\)-stable, (given also in the paper), are central in the investigation. \(\text{VI}(\varepsilon)\)-regular is characterized by the generalized Slater condition and \(S( \varepsilon) \neq \emptyset\), compact. Then the authors prove that if for all \(\widetilde {A}, \widetilde{f_i}, i \dots m, \widetilde{L}, \widetilde{l}\) it is possible to find \( \overline{\varepsilon}\) such that \(\text{VI}(\varepsilon)\) is regular, for \(0< \varepsilon< \overline{\varepsilon}\) then \(\text{VI}(0)\) is said to be stable. Besides, they define the dual variational inequality problem (\(D\text{VI}(\varepsilon)\)) associated to \(\text{VI}(\varepsilon)\) using its description as a generalized equation. Denoting \(U(\varepsilon)\) as the solution's set of \(D\text{VI}(\varepsilon)\), it is proven that: \(\text{VI}(0)\)-regular \(\Rightarrow \text{VI}(0)\)-stable, \(S(.)\) and \(U(.)\) locally stable at \(0^+\). In addition: \(\text{VI}(0)\) regular and \(C\) subset of the interior of \(\operatorname {dom} A\) implies \(S(.)\) and \(U(.)\) closed at \(0^+\).
    0 references
    variational inequalities
    0 references
    stability theory
    0 references
    perturbations
    0 references
    duality theory
    0 references
    recession functions
    0 references

    Identifiers