Minimization of stochastic functionals and random variational inequalities

From MaRDI portal





The author studies the problem of minimizing stochastic functionals of the form \[ \widetilde f(u)(\omega)=f\big(\omega,u(\omega)\big),\quad \omega\in\Omega, u\in L^0(\Omega,{\mathcal F},P,H). \] Here \((\Omega,{\mathcal F},P)\) is a complete probability space, \(H\) is a Hilbert space, and \(f\) is an \({\mathcal F}\times{\mathcal B}(H)\)-measurable real function. Given a closed convex set \(K\subset H\), he provides sufficient conditions for the existence of \(u\in L^0(\Omega,{\mathcal F},P,K)\) such that \(\widetilde f(u)(\omega)\leq f(\omega,x)\) for all \(x\in K\). As a departure from the literature, none of these conditions is imposed on all sample paths. The author shows that little is lost in comparison with the situation with sample path conditions. On the other hand, he proves that the Itô integral is an example of a functional where such conditions are not valid. In the second part of the paper, the author establishes a link between minimization of stochastic inequalities and random variational inequalities associated with a pair of functionals one of which is stochastically bilinear and the other is stochastically linear.











This page was built for publication: Minimization of stochastic functionals and random variational inequalities

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4393943)