Iterative solution methods for mesh approximation of control and state constrained optimal control problem with observation in a part of the domain (Q2263100)

Considering a homogeneous Dirichlet problem for Poissons's equation with linear control \(u\) as state equation, in state \(y\) and control positively definite quadratic functional as objective as well as point wise control and state constraints, the authors use for the numerics the approach ``first discretize then optimize''. They get the finite-dimensional problem by a finite element method with piecewise linear elements for the state and piecewise constant elements for the control. The corresponding finite-dimensional problem has a unique solution and its -- with respect to the pointwise constraints -- nonsmooth Karush-Kuhn-Tucker (KKT) conditions has a unique solution w.r.t. the state and may be solved non-uniquely w.r.t. the multipliers. In a numerical example the authors show that for solving the KKT conditions by suitable transformation the Uzawa method can be much faster than some linear convergent gradient like method for KKT conditions with quadratic penalty relaxation of point wise constraints.