Optimization Problems With Concentration And Oscillation Effects: Relaxation Theory And Numerical Approximation (Q4265779)

The authors consider optimal control problems of the form: \[ \min J(y,u)=\int_0^T (a(t,y,u)+ b(t,u(t))) dt+ f(y(T)), \] when the pair \((y,u)\), the state and control variables, satisfies the differential equation \(y'=c(t,y,u)+d(t,u)\), \(y(0)=y_0\), \(y\in W^{1,1}(0,T;\mathbb{R}^n)\), and \(u\in L^p(0,T;\mathbb{R}^m)\). Due to concentrations in the control \(u\), the response \(y\) may fall out \(W^{1,1}(0,T;\mathbb{R}^n)\), keeping its variation bounded. Instead of extending the problem in the space of functions with bounded variation \(\text{BV}(0,T; R^n)\), the authors consider a bit finer extension \(W^{1,1}_{\mu}(0,T;\mathbb{R}^n)\) of \(W^{1,1}(0,T;\mathbb{R}^n)\) introduced by \textit{J. Souček} [Čas. Pěst. Mat. 97, 10-46 (1972; Zbl 0247.46051)]. Using DiPerna-Majda's measures and Souček's extension the authors introduce a relaxed problem. A relaxation theorem is proved. Approximations of DiPerna-Majda measures are studied and numerical approximations are performed on different examples.