Observability and unique continuation inequalities for the Schrödinger equation (Q2009218)

In this interesting paper the authors have established several observability and unique continuation inequalities (at either two points in time or one point in time) for the following free Schrodinger equation: \begin{align*} \left\{\begin{array}{cc}i\partial_t u(x,t)+\Delta u(x,t)=0,\quad (x,t)\in \mathbb{R}^n \times (0,\infty),\\ \hspace{-5cm}u(x,0)\in L^2(\mathbb{R}^n; \mathbb{C}).\end{array}\right.\tag{1} \end{align*} From the perspective of applications to these inequalities correspond different controllability properties for the Schrodinger equation. The free Schrodinger equation (1) describes the evolution of the wave function for a particle without external field. Though the Schrodinger equation with a potential \(V\neq 0\) is more attractive, the free Schrodinger equation is also important and has been extensively studied. There are three main theorems in this paper and the first one gives an observability inequality at two points in time for the equation (1). Theorem. Given \(x^{\prime}, x^{\prime\prime}\in \mathbb{R}^n, r_1, r_2>0\) and \(T>S\geq 0\), there is a positive constant \(C=C(n)\) such that for all \(u_0 \in L^2 (\mathbb{R}^n;\mathbb{C})\), \begin{align*} \int_{\mathbb{R}^n}^{} |u_0(x)|^2\mathrm{d}x\leq Ce^{Cr_1r_2\frac{1}{T-s}}\bigg(\int_{B_{r_1}^c (x^{\prime})}^{}|u(x,S;u_0)|^2\mathrm{d}x+\int_{B_{r_2}^c (x^{\prime\prime})}^{}|u(x,T;u_0)|^2\mathrm{d}x\bigg)\tag{2} \end{align*} The second theorem gives a unique continuation inequality at one time point for a class of solutions to (1). (This class consists of solutions whose initial data have exponential decay at infinity). Theorem. The following conclusions are true for all \(r, a, T>0:\) \begin{itemize} \item [(i)] There are \(C=C(n)>0\) and \( \theta=\theta(n)\in (0,1)\), depending only on \(n\), such that for all \(u_0 \in C_0^\infty(\mathbb{R}^n;\mathbb{C})\), \begin{align*} \int_{\mathbb{R}^n}^{} |u_0(x)|^2\mathrm{d}x\leq C\bigg(1+\frac{r^n}{(aT)^n}\bigg)\bigg(\int_{B_{r}^c (0)}^{}|u(x,T;u_0)|^2\mathrm{d}x\bigg)^{\theta^{1+\frac{r}{aT}}}\nonumber\\\times \bigg(\int_{\mathbb{R}^n}^{} e^{a|x|}|u_0(x)|^2\mathrm{d}x\bigg)^{1-\theta^{1+\frac{r}{(aT)}}}\tag{3} \end{align*} \item [(ii)] There is \(C=C(n)>0\) such that for any \(\beta>1\) and \(\gamma \in(0,1)\) and all \(u_0 \in C_0^\infty(\mathbb{R}^n;\mathbb{C}),\) \begin{align*} \int_{\mathbb{R}^n}^{} |u_0(x)|^2\mathrm{d}x\leq Ce^{\frac{c^\beta r^\beta}{a(1-\gamma)^{T^\beta}}^{\frac{1}{(\beta-1)}}}\bigg(\int_{B_{r}^c (0)}^{}|u(x,T;u_0)|^2\mathrm{d}x\bigg)^{\gamma}\nonumber\\\times \bigg(\int_{\mathbb{R}^n}^{} e^{a|x|^\gamma}|u_0(x)|^2\mathrm{d}x\bigg)^{1-\gamma}.\tag{4} \end{align*} \item [(iii)] Let \(\alpha(s), s \in \mathbb{R}^{+}\), be an increasing function with \(\lim_{s\rightarrow \infty}\alpha(s)/s =0\). Then for each \(\gamma \in (0,1)\), there is no positive constant \(C\) such that for all \(u_0\in C_0^\infty (\mathbb{R}^n;\mathbb{C}),\) \begin{align*} \int_{\mathbb{R}^n}^{} |u_0(x)|^2\mathrm{d}x\leq C\bigg(\int_{B_{r}^c (0)}^{}|u(x,T;u_0)|^2\mathrm{d}x\bigg)^{\gamma} \bigg(\int_{\mathbb{R}^n}^{} e^{a\alpha|x|}|u_0(x)|^2\mathrm{d}x\bigg)^{1-\gamma}.\tag{5} \end{align*} \end{itemize} The last main theorem gives another kind of unique continuation inequality at one point for a class of solutions to (1). Theorem. Given \(x^{\prime}, x^{\prime\prime}\in \mathbb{R}^n\) and \(r_1, r_2,a, T>0\), the following estimate holds for all \(u_0 \in L^2 (\mathbb{R}^n;\mathbb{C})\), \begin{align*} \int_{B_{r_2}(x^{\prime\prime})}^{} |u(x,T;u_0)|^2\mathrm{d}x\leq Cr_2^n((aT)\wedge r_1)^{-n}\bigg(\int_{B_{r_1 (x^{\prime})}}^{}|u(x,T;u_0)|^2\mathrm{d}x\bigg)^{\theta^p}\nonumber\\\times \bigg(\int_{\mathbb{R}^n}^{} e^{a|x|}|u_0(x)|^2\mathrm{d}x\bigg)^{1-\theta^p}.\tag{6} \end{align*} where \(C=C(n)>0, \theta=\theta(n) \in (0,1)\) and \begin{align*} p:=1+\frac{|x^{\prime}-x^{\prime\prime}|+r_1+r_2}{(aT)\wedge r_1}\tag{7} \end{align*} There are several remarks given about these theorems. Also the authors presented three consequence of the above main theorems.