Optimal distribution of the internal null control for the one-dimensional heat equation (Q616465): Difference between revisions

Given the interval \(\Omega=(0,1)\) and a final time \(T>0\), the authors consider the following optimal design problem: minimize, among all subsets \(\omega\subset\Omega\) with prescribed fixed measure, the \(L^2((0,T)\times\omega)\)-norm of the successful control function \(\varphi\) of the exact null controllability problem for the one-dimensional heat equation \[ u_t-u_{xx}=\chi_\omega\varphi\;\text{in } (0,T)\times\Omega, \] with null boundary condition and given initial condition. We recall that \(\chi_\omega\) is the characteristic function of the subset \(\omega\), and that the exact null controllability problem consists in finding \(\varphi\in L^2((0,T)\times\Omega)\) such that the corresponding solution \(u\) satisfies \(u(T,\cdot)=0\) in \(\Omega\). In order to have existence of a minimizer (still preserving the same infimum), the authors prove that it is sufficient to relax the problem by substituting the set of characteristic functions with fixed integral (corresponding to the class of subsets with fixed measure) by its convex envelope. Then they give a first-order optimality condition for the minimizers, and discuss some other particular problems. Finally, they also give some numerical experiments.