Necessary optimality conditions in bilinear control problems (Q762757)

We shall derive necessary optimality conditions for distributed bilinear control problems of the following type: Minimize \(\int^{T}_{0}L(y(t),\quad u(t))dt+\phi (y(T)),\quad 0<T<\infty\) over the set of all functions \(u\in L^ 2(0,T)\), \(y\in C(0,T;H)\) subject to \(y'(t)=A\quad y(t)+u(t)B\quad y(t)\) f.a.a. \(t\in t]0,T[\), \(y(0)=y_ 0\). Here A denotes the infinitesimal generator of a \(C^ 0\) semigroup of bounded linear operators \(\{\) S(t); \(t\geq 0\}\) on a real Hilbert space H, \(B: H\to H\) is a linear bounded operator, L and \(\phi\) are lower semicontinuous convex functions defined on \(H\times R\) and H, respectively, with values in \(]-\infty,+\infty]\), not identically \(+\infty\). The necessary conditions for optimality are given in ''subdifferential'' form and do not require the differentiability of the cost functional. The main idea of our approach consists in approximating the control problem by a family of smooth problems for which the necessary optimality conditions are immediate and to tend to limit in the approximating equations.