Optimal control of the convection-diffusion equation using stabilized finite element methods (Q2373394)

An optimal control problem is considered for a polygonal domain \(\Omega \subset\mathbb R^d\), with \(d=2,3\). The cost functional is quadratic with respect to control and state and the state equation is a linear convection-diffusion reaction one. The state variable \(u\) belongs to the space \(V=H^1_0(\Omega)\), while the control \(q\) belongs to a pointwise box constraint, i.e., the admissible control set is \(Q_{\text{ad}}= \{ q \in L^2(\Omega)\); \(a \leq q \leq b\) a.e. in \(L^2(\Omega) \}\), where \(a<b\) are real values, possibly \(a=-\infty\) and \(b=\infty\). For the problem considered, being a convection dominated one, it is well known that the finite element method parameter \(h\) should be sufficiently small with respect to a ratio between constants in the state equation in order to avoid strongly oscillatory solutions. The main contribution of this paper is the a priori error analysis of the discretization of the optimal control problem by stabilized finite elements. It is demonstrated that the approximation error is of order \(O(h^{3/2})\) in \(L^2(\Omega)\) for the control, state and adjoint state in both cases, the unconstrained case (\(Q_{\text{ad}}= L^2(\Omega)\)) and the constrained one. It is known that the above error estimate is optimal. Numerical results are also presented.