Existence, Uniqueness and Convergence of Simultaneous Distributed-Boundary Optimal Control Problems

TL;DR

This paper investigates the existence, uniqueness, and convergence of optimal controls in a steady-state heat conduction problem with mixed boundary conditions, analyzing how solutions behave as the heat transfer coefficient varies.

Contribution

It establishes the existence and uniqueness of optimal controls for both the original and parameterized problems, and proves their strong convergence as the parameter approaches infinity.

Findings

Existence and uniqueness of optimal controls for all parameter values.

Strong convergence of controls and states as the heat transfer coefficient tends to infinity.

Estimates relating solutions of vectorial and scalar optimal control problems.

Abstract

We consider a steady-state heat conduction problem $P$ for the Poisson equation with mixed boundary conditions in a bounded multidimensional domain $\Omega$. We also consider a family of problems $P_{\alpha}$ for the same Poisson equation with mixed boundary conditions being $\alpha>0$ the heat transfer coefficient defined on a portion $\Gamma_{1}$ of the boundary. We formulate simultaneous \emph{distributed and Neumann boundary} optimal control problems on the internal energy $g$ within $\Omega$ and the heat flux $q$, defined on the complementary portion $\Gamma_{2}$ of the boundary of $\Omega$ for quadratic cost functional. Here the control variable is the vector $(g,q)$. We prove existence and uniqueness of the optimal control $(\overline{\overline{g}},\overline{\overline{q}})$ for the system state of $P$, and $(\overline{\overline{g}}_{\alpha},\overline{\overline{q}}_{\alpha})$ for the system state of $P_{\alpha}$, for each $\alpha>0$, and we give the corresponding optimality conditions. We prove strong convergence, in suitable Sobolev spaces, of the vectorial optimal controls, system and adjoint states governed by the problems $P_{\alpha}$ to the corresponding vectorial optimal control, system and adjoint states governed by the problem $P$, when the parameter $\alpha$ goes to infinity. We also obtain estimations between the solutions of these vectorial optimal control problems and the solution of two scalar optimal control problems characterized by fixed $g$ (with boundary optimal control $\overline{q}$) and fixed $q$ (with distributed optimal control $\overline{g}$), respectively, for both cases $\alpha>0$ and $\alpha=\infty$.