Boundary control of semilinear elliptic equations with discontinuous leading coefficients and unbounded controls (Q4351387)

The system is described by the elliptic boundary value problem \[ Ay+ \Phi(x,y,u)= 0\quad (x\in\Omega=\text{bounded domain in }\mathbb{R}^N), \] \[ {\partial y\over\partial n_A}+ \Psi(x,y,v)= 0\quad (0\leq t\leq T,\;x\in\Gamma=\text{boundary of }\Omega) \] under distributed and boundary condrol \(u\), \(v\), respectively. \(A\) is a second-order symmetric uniformly elliptic operator under minimal regularity assumptions (bounded measurable coefficients). The system includes control constraints \((u,v)\in K_U\times K_V\), where \(K_U\subseteq L^r(\Omega)\), \(K_V\subseteq L^s(\Gamma)\) for appropriate \(r\), \(s\); the sets \(K_U\), \(K_V\) may be unbounded. There are also state constraints of the form \(g(x,y(x))\leq 0\) or, more generally \(G(y)\in{\mathcal C}\subseteq C(\overline\Omega)\). The cost functional to be minimized is \[ J(y,u,v)= \int_\Omega F(x,y,u)dx+ \int_\Gamma H(x,y,v) d\sigma. \] Under suitable assumptions on the domain and on the various functions and coefficients, the authors obtain a version of (the steady state) Pontryagin's maximum principle. The adjoint state \(p\) satisfies a parabolic initial value problem \[ \begin{aligned} A^*p+ ap & =\mu_\Omega\quad\text{in }\Omega,\\ {\partial p\over\partial n_{A^*}}+ bp & = \mu_\Gamma\quad\text{in }\Gamma,\end{aligned} \] where \(\mu_\Omega\) (resp. \(\mu_\Gamma\)) is a Radon measure in \(\Omega\) (resp. \(\Gamma\)).