Approximate controllability of a bilinear wave equation and minimum time
Karine Beauchard, Thomas Perrin, Eugenio Pozzoli
TL;DR
This work establishes global approximate controllability results for a bilinear Klein-Gordon/wave equation on the torus, with controls supported on only the first (2d+1) Fourier modes. The authors develop a two-pronged strategy: (i) small-time control of the velocity via Lie-bracket approximations that generate new velocity increments while preserving the profile, and (ii) a subsequent transfer of velocity into the profile using dual flows, all within the STAR framework to handle infinite-dimensional dynamics. In low dimensions (d ∈ {1,2}) they prove Tmin = r(W0) for nonzero initial data, while in higher dimensions (d ≥ 3) Tmin can vanish provided the zero-set of the initial state has positive-measure gaps, and GAC is achievable in large time from any W0 ≠ 0. The results synthesize Agrachev–Sarychev Lie-bracket techniques with propagation of well-prepared positive states, yielding both small-time and large-time controllability statements and clarifying the role of the zero-set geometry in controllability times.
Abstract
We study the global approximate controllability (GAC) of a Klein-Gordon wave equation, posed on the torus $\mathbb{T}^d$ of arbitrary dimension $d\in \mathbb{N}^*$, with bilinear control potentials supported on the first $(2d+1)$-Fourier modes. Let $Z(W_0)\subset \mathbb{T}^d$ be the set of essential zeroes of the initial state $W_0\in H^1\times L^2(\mathbb{T}^d)$, and $r(W_0)\geq 0$ be the maximum radius of a ball of $\mathbb{T}^d$ contained in $Z(W_0)$. Due to finite speed of propagation, the minimum control time starting from $W_0$ is necessarily larger than or equal to $r(W_0)$. We prove the following three facts. In low dimensions $d \in \{1,2\}$: the minimum time for GAC from $W_0 \neq 0$ is equal to $r(W_0)$. In any dimensions $d\geq 3$: the minimum time for GAC from $W_0$ is zero if $Z(W_0)$ has zero Lebesgue measure; and the GAC in sufficiently large time from all $W_0\neq 0$. The proof strategy consists in combining Lie bracket techniques \emph{à la Agrachev-Sarychev} with the propagation of well-prepared positive states.
