Approximation theory for stochastic variational and Ky Fan inequalities in finite dimensions (Q1308652): Difference between revisions

Ideas of approximation and continuity are important both for optimization problems and variational inequalities (VIs). The paper studies approximation issues for a class of VIs, called Stochastic Variational Inequalities (SVIs) that arise, e.g. in stochastic programming. The problem of SVIs is a special case of the more general problem of stochastic Ky Fan inequalities (SKFIs): Let \(C\subseteq \mathbb{R}^ n\), \(S\) a probability space and \(\mu\) a probability measure on \(S\). Let \(\phi: C\times C\times S\to \mathbb{R}\) be a function. The problem SKFI \((\phi,C,\mu)\) is to find \(x\in C\) such that \(\int\phi(x,u,\xi)d\mu\geq 0\) for every \(u\in C\). The study of SKFIs is motivated by recent research in the theory of portfolio choice for investors who are not classical expected utility maximizers. The paper provides first some notation and definitions, and discusses the various convergence concepts for functions, sets and integrals that are used later in the approximation results. After studying the approximation theory for SKFIs, the results are applied to SVIs. Defining monotone Ky Fan inequalities, the paper shows how monotone variational inequalities can be important in the approximation theory for monotone Ky Fan inequalities. Finally, the SKFIs are applicated on the above-mentioned new type of portfolio choice problems, which is a response to experimentally observed violations of the classical Von Neumann-Morgenstern-axioms for utility theory.