Optimal control in coefficients of elliptic equations with state constraints (Q1198459): Difference between revisions

The author considers state constrained optimal control problems of second order linear elliptic equations, where the control is a coefficient of the partial differential equations. Precisely, the following optimal control problem on a bounded convex domain \(\Omega\) in \(\mathbb{R}^ n\) \((1\leq n\leq 3)\) is considered: \[ \text{Minimize}\quad J(u)={1\over 2}\int_ \Omega (y_ u-y_ d)^ 2 dx+{\nu\over 2} \int_ \Omega u^ 2 dx \] subject to \(u\in U_{ad}\) and \(| y_ u(x)|\leq \delta\) for all \(x\in \Omega\), where \(y_ d\in L^ 2(\Omega)\), \(\nu\geq 0\), \(0<\delta<+\infty\), and \(U_{ad}=\{u\in C^{0,1}(\overline\Omega): a\leq u(x)\leq b\) and \(|\nabla u(x)|\leq c\}\), with \(0< a< b< +\infty\) and \(0<c<+\infty\), and where \(y_ u\in H^ 1_ 0(\Omega)\cap H^ 2(\Omega)\) is a unique solution to the Dirichlet problem \(- \text{div}(u\nabla y)=f\) in \(\Omega\), and \(y=0\) on \(\partial\Omega\), with \(f\in L^ 2(\Omega)\). Under the assumption of the existence of \(u_ 0\in U_{ad}\) with \(| y_{u_ 0}|\leq\delta\), the existence of the optimal control is proved. Also the optimality conditions are derived and the numerical discretization of this problem is performed.