Analyse de sensibilité d'un problème de contrôle optimal bilinéaire (Q443802): Difference between revisions

The author is concerned with the sensitivity analysis of an optimal control problem of bilinear type. The perturbed optimal control problem consists of minimizing the cost functional \[ J(z,u)=\int_0^T \|z(t)\|^2_{L^2(\Omega)}dt+\int_0^T \|u(t)\|^2_{\mathbb{R}^n}dt, \] with a control \(u\) and the state \(z\) defined as a solution of the initial boundary-value problem \[ \begin{aligned} &\frac{\partial z}{\partial t}+\Delta^2 z=\sum_{i=1}^nu_i(t)\frac{\partial z}{\partial x_i}+f(t,x) \;\text{in}\;(0,T)\times\Omega,\\ &z(t,x)=\frac{\partial z}{\partial \nu}(t,x)=0\;\text{on}\;(0,T)\times\partial\Omega;\;z(0,x)=z^0(x)\;\text{in}\;\Omega, \end{aligned} \] where \(\Omega\) is a bounded \(C^1\) domain of \(\mathbb{R}^n\), \(n=2 \text{ or } 3\). Under conditions of smallness of the initial state \(z^0\) and the perturbation \(f\), the author exploits the properties of regularity and uniqueness of optimal control in order to demonstrate the stability of the optimal value function. The formula for directional derivatives at zero of the optimal value is explained.