On constrained Newton linearization and multigrid for variational inequalities (Q1606045): Difference between revisions

Let \(\Omega\) be a bounded, polyhedral domain in the Euclidean space \(\mathbb{R}^{2}\) and let \(H\) be a closed subset in \(H^{1}(\Omega)\). The problem of minimization of the functional \(J(v)+I(v)\) in the set \(H\) is considered, where \(J(v)\) is a quadratic functional is which induced by a continuous symmetric and \(H\)-elliptic bilinear form and a linear functional belonging to the dual space, \(I(v)\) is an integral functional generated by some convex function. For lack of smoothness, the usual Newton multigrid method cannot be applied. The author's approach is based on a combination of convex minimization methods with constrained Newton linearization. Global convergence of the resulting monotone multigrid methods is shown and polylogarithmic upper bounds for the asymptotic convergence rates are given. The method is illustrated by numerical examples.