Distributed control of systems governed by a general class of quasilinear elliptic equations (Q687498): Difference between revisions

Let \(\Omega\) be a bounded open subset of \(\mathbb{R}^ N\) with Lipschitz continuous boundary; let \(a: \Omega\times \mathbb{R}^ N\to \mathbb{R}^ N\), \(a_ 0: \Omega\times\mathbb{R}\to \mathbb{R}\) be two Carathéodory functions, \(C^ 1\) in the second variable, such that \[ \sum_{i,j=1}^ N \partial_{N+i} a_ j (.,\eta) \xi_ i\xi_ j\geq \Lambda_ 1(k+| \eta|)^{\alpha-2} |\xi|^ 2, \qquad \sum_{i,j=1}^ N | \partial_{N+i} a_ j (.,\eta)|\leq \Lambda_ 2 (k+| \zeta|)^{\alpha-2}, \] \(0\leq \partial_{N+1} a_ 0(.,s)\leq f(| s|)\), \(a_ 0(.,0) =0\), \(a(.,0) =0\), for all \(\xi,\eta\in\mathbb{R}^ N\), \(s\in\mathbb{R}\), for some \(\Lambda_ 1,\Lambda_ 2>0\), \(k\in [0,1]\), \(\alpha\in ]1,+\infty[\), positive and non-decreasing function \(f\); \(y_ d\in L^ 2(\Omega)\); \(\varepsilon,\nu\geq 0\); let \({\mathbf K}\) be a nonempty, convex and closed subset of \(L^ 2(\Omega)\). Let \(A_ \varepsilon y=- \varepsilon\Delta y-\text{div } a(.,\nabla y)+ a_ 0(.,y)\), if \(u\in L^ 2(\Omega)\) let \(y_ \varepsilon(u)\in W_ 0^{1,\alpha} (\Omega)\cap L^ \infty(\Omega)\) such that \(A_ \varepsilon y_ \varepsilon(u)u\) in \(\Omega\), \(y_ \varepsilon(u)=0\) on \(\partial\Omega\); \(J:{\mathbf K}\to \mathbb{R}\), \(J(v)= {1\over 2} \int_ \Omega | y_ 0(v)- y_ d|^ 2 dx+ {\nu\over 2} \int_ \Omega | v|^ 2 dx\), if \(\varepsilon>0\) let \(J_ \varepsilon(v)= {1\over 2} \int_ \Omega | y_ \varepsilon(v)- y_ d|^ 2 dx+{\nu\over 2} \int_ \Omega | v|^ 2 dx+ {1\over 2} \int_ \Omega | v-\overline{u}|^ 2 dx\) where \(\overline{u}\) is a point of minimum for \(J\). Then the authors prove under suitable hypothesis the existence of \(y_ \varepsilon(u)\) and of the minimum of \(J\) and \(J_ \varepsilon\) and if \(\alpha\geq 2\) and \(k\neq 0\) the differentiability of the state with respect to the control and the optimality system of the problem. In order to deal with the case in which \(\alpha>2\) and \(k=0\) and with the case in which \(\alpha<2\) and \(k\neq 0\) they show in what sense the problems associated to \(J_ \varepsilon\) approximate that associated to \(J\) and pass to the unit in the optimality systems of the approximated problems also deducing the regularity of the optimal solution and of the optimal state.