Optimization of nonlocal time-delayed feedback controllers (Q276874): Difference between revisions

In this paper, the following optimal control problem is considered: \[ \partial_{t}u(x,t)-\Delta u(x,t)+R(u(x,t)) =\kappa\Big(\int_0^{T}g(\tau)u(x,t-\tau)d\tau-u(x,t)\Big ),\;(x,t)\in Q=\Omega\times (0,T),\tag{1} \] \[ u(x,s)=u_0(x,s),\;(x,s)\in \Omega\times [-T,0],\tag{2} \] \[ \partial_{n}u(x,t)=0,\; (x,t)\in \partial\Omega\times (0,T),\tag{3} \] \[ \frac{1}{2}\int_{Q}c(x,t)(u(x,t)-u_{d}(x,t))^2 dxdt +\frac{\nu}{2}\int_0^{T}g^2(t)dt\rightarrow \min_{g\in G}, \] where \(\Omega\subset \mathbb{R}^n\), \(n\leq 3\), is a bounded Lipschitz domain, \[ G=\Big \{g\in L^{\infty}(0,T):\;0\leq g(t) \leq \beta \;t\in [0,T],\int_0^{T}g(t)dt=1\Big\}, \] \(u_0,u_d \in L^{\infty}(0,T)\) are given functions and \(\nu\geq 0\) is a Tikhonov regularization parameter. For the initial-boundary value problem (1)--(3), the existence and uniqueness of a solution is proved. Necessary optimality conditions are established. Some numerical examples are presented.