Optimal control problems governed by semilinear parabolic equations with low regularity data (Q2434031): Difference between revisions

This very important and useful paper is concerned with optimal control problems for semilinear parabolic equations with low regularity data. The purpose of authors in this paper is to demonstrate the applicability of the results recently obtained, concerning optimal control problems with low regularity data. The nonlinear model problems where the controls enter linearly, is studied. Here, the state problem of the form \[ \begin{aligned}\partial_t y +{\mathcal A}y=f(x,t,y,\nabla y)+u_Q, &\quad x\in\Omega,\;t\in (0,T),\;y=0,\;x \in \Gamma_0,\;t\in[0,T],\\ \partial_\nu y = g(x,t,y)+u_\Sigma, &\quad x \in \Gamma_1,\;t\in[0,T], \\ y(\cdot,0)=y^0&\quad\text{in }\Omega , \end{aligned}\tag{1} \] is considered (\(\Omega \subset \mathbb{R}^n\) is an open bounded set, with smooth boundary \(\Gamma, n \geq 2, \Gamma = \Gamma_0 \cup \Gamma_1\) are open as well as closed in \(\Gamma\) and disjoint). The operator \(\mathcal A\) has the form: \(\mathcal A\) \(y=-\nabla \cdot (a\nabla y), \mathbf{a}=[a_{jk}] \in C^\infty (\bar{\Omega},\mathbb{R}^{n \times m})\) is symmetric and uniformly positive definite. The maps \(f\) and \(g\) are nonlinear Carathéodory functions, \(y^0\) is a bounded Radon measure in \(\Omega\), and the controls \(u_Q\) and \(u_\Sigma\) are bounded Radon measures in \( \Omega \times [0,T]\) and \(\Sigma_1 = \Gamma_1 \times [0,T]\) respectively. Furthermore the authors deal with optimal controls for problems governed by the system \[ \begin{aligned}\partial_t y_1 - \Delta y_1 = \kappa y_1 y_2 - by_1+u_1+v_1 y_1 &\quad\text{in }\Omega \times [0,T], \\ \partial_t y_2-d\Delta y_2 = \alpha y_1+u_2 &\quad\text{in }\Omega \times [0,T], \end{aligned}\tag{2} \] which is complemented by suitable boundary and initial conditions. Here \(d>0, a>0, \kappa, b \in \mathbb{R},\) and \(u_1, v_1, v_2\) are controls. Main result: The abstract existence theorem for optimal control problems of the form: minimize \(J(y(u),u)\) over \(u \in U^G _{\text{ad}}\) is proved. Here, the cost functional \(J(.\;u)\) is lower semicontinuous and coercive in a space \(Y\) which may depend on the nonlinearities \(f\) and \(g\). The set of admissible controls is defined as: \(U^G _{\text{ad}}=\{ u \in U_{\text{ad}}\): the solution \(y = y(u)\) of (1) exists globally \(\}\). In the monotone case (if \(f(x,t,\xi, \eta)\xi\leq 0\) and \(g(x,t,\xi)\xi\leq 0\) for all \([x,t,\xi, \eta]\)) all solutions of (1) are global and solutions with bounded data are uniformly bounded (without additional assumptions). For the case \(d=\kappa=1\) in (2) when at most one of the controls \([u_1, v_1, v_2]\) is non-zero provided suitable (low) regularity of controls and lower semicontinuity and coercivity of the corresponding cost functional, the existence of optimal controls is proved, too. Moreover, in the case \(u_2\neq0\) another control acting on the boundary \(\partial\Omega\) is admitted. This paper presents new results of the domain of optimal control problems governed by semilinear parabolic equations with low regularity data. The precise proofs of the abstract existence theorems are proposed.