scientific article; zbMATH DE number 3994385

From MaRDI portal
Publication:4721118





In Hilbert spaces the extension of the method \[ Ux^{n+1}=Ux^ n- \alpha_ ny^ n,\quad y^ n\in Ax^ n,\quad n=1,2,..., \] to conditional optimization problems (x\(\in \Omega\), \(\Omega\) a closed convex set of constraints) or to the more general problem of solving variational inequalities is most often realized by projecting the right- hand side of (1) on the set \(\Omega\) : \(x^{n+1}=P_{\Omega}(x^ n- \alpha_ nAx^ n)\), \(n=1,2,...\). In Banach spaces it does not seem possible to prove even convergence of such a method in the framework of the traditional schemes of investigation, chiefly because (in contrast to the Hilbert case) the projection operator is not nonexpanding in Banach spaces. We consider three methods for solving variational inequalities. The first uses the operation of projection on \(\Omega\) of the element \(x^ n\) under the sign of the operator A. A distinctive feature of it is the presence of an additional term which plays the role of a kind of penalty for leaving the constraints. The second method uses support functionals of the set \(\Omega\) and has been considered previously only in Hilbert spaces for the problem of minimizing convex functionals. The third method is connected with problems in which the constraints are given in the form of inequalities, and uses Lagrange multipliers.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4721118)