Control Norms for Large Control Times

The control system considered is \[ y(t) + Ay(t) = Bu(t) \] where \(y(t)\) takes values in a Hilbert space \(H,\) \(A\) is a self-adjoint operator in \(H\) with positive pure point spectrum and \(B : U \to H\) is linear and bounded; \(U\) is a second Hilbert space. The author assumes that every initial state \((y_0, y_1)\) can be driven to \((0, 0)\) in time \(T > T_0 > 0\) and studies the problem of obtaining bounds on the norm of the minimum norm control \(u_T\) for \(T > T_0\). Assuming that the initial state is a finite linear combination of eigenmodes, two of these bounds are \[ {c \over \sqrt T} \leq \| u_T\| _{L^2(0, T; U)} \leq {C \over \sqrt T}, \qquad c \leq \| u_T\| _{L^1(0, T; U)} \leq C \, . \] The first was shown in a particular case by Krabs, Leugering and Seidman in 1985; the second was experimentally observed by various authors (see references in the paper). Approaches such as the Hilbert Uniqueness Method are not precise enough for this task; the author uses instead nonharmonic Fourier analysis (in particular biorthogonal systems and Riesz bases). When the results are applied to the wave equation, the necessary controllability results are obtained from the geometric conditions of Bardos, Lebeau and Rauch. The author notes that a similar problem can be posed for parabolic equations; since, for these equations, controllability, when available, can be achieved in arbitrarily small time \(T > 0\), one can try to obtain bounds on \(\| u_T\| \) not only for large \(T\) but also as \(T \to 0.\) To the references in this paper on the latter problem add \textit{E. Güichal}, J. Math. Anal. Appl. 110, 519--527 (1985; Zbl 0579.49029), who derived the first known bound from below for small \(T.\)

Recommendations

Cites work

Cited in

(2)

This page was built for publication: Control Norms for Large Control Times

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4256523)