Quantitative unique continuation for Schrödinger operators (Q2182591)

The author makes important contributions to the (strong) unique continuation property problem regarding solutions to second-order elliptic equation of the form \[ \Delta u+Vu=0 \] in some open, connected subset of \(\mathbb{R}^n\), where \(n\geq3\). Specifically, the author primarily proves the following results: Theorem 1. Assume that for some \(t\in(\frac{n}{2},\infty]\), one has \(\|V\|_{L^t(B_{R_0})}\leq M\). Let \(u\) be a solution in \(B_{10}\). Suppose further that \(u\) is bounded and normalized in the sense that \(\|u\|_{L^{\infty}(B_6)}\leq \hat{C}\) and \(\|u\|_{L^{\infty}(B_1)}\geq1\). Then, for each \(x_0\in B_1\) and all sufficiently small \(\varepsilon,r>0\), it follows that \[ \|u\|_{L^{\infty}(B_r(x_0))}\geq cr^{CM^\mu}, \] where \(C=C(n,t,\hat{C},\varepsilon)\), \(c=c(n,t,\hat{C},\varepsilon)\), \(\tilde{\varepsilon}=C(n,t,\varepsilon)\varepsilon\) and \[ \mu= \begin{cases} \frac{4t}{6t-3n+2}&t\in(n,\infty],\\ \frac{4nt}{(2t-n)(3n+2)}+\tilde{\varepsilon}&t\in(\frac{n}{2},n]. \end{cases} \] In combination with the scaling argument from [\textit{J. Bourgain} and \textit{C. E. Kenig}, Invent. Math. 161, No. 2, 389--426 (2005; Zbl 1084.82005)], the author can prove the following quantitative unique continuation at infinity. Theorem 2. Assume that \(\|V\|_{L^{t}\left(\mathbb{R}^{n}\right)} \leq A_{0}\) for some \(t \in\left(\frac{n}{2}, \infty\right]\). Let \(u\) be a solution in \(\mathbb{R}^{n}\) for which \(\|u\|_{L^{\infty}\left(\mathbb{R}^{n}\right)} \leq C_{0}\) and \(|u(0)| \geq 1 .\) Then for any \(\varepsilon \in\left(0, \varepsilon_{0}\right)\) and any \(R\) sufficiently large \[ \inf _{\left|x_{0}\right|=R}\|u\|_{L^{\infty}\left(B_{1}\left(x_{0}\right)\right)} \geq \exp \left(-C R^{\Pi} \log R\right), \] where \(\tilde{\varepsilon}=C(n, t, \varepsilon) \varepsilon\), \(C=C\left(n, t, A_{0}, C_{0}, \varepsilon\right)\) and \[ \Pi=\begin{cases} \frac{4(2 t-n)}{6 t-3 n+2} & t \in(n, \infty], \\ \frac{4 n}{3 n+2}+\tilde{\varepsilon} & t \in\left(\frac{n}{2}, n\right]. \end{cases} \] The proof depends on a new version of the \(L^{p}\)-\(L^{q}\) Carleman estimate for \(\Delta\) for all values of \(t>\frac{n}{2} \). This along with a three-ball inequality allows the author to prove a propagation of smallness result that leads to the stated estimate. The results also give an improvement to some of the results presented in [the author and \textit{J. Zhu}, Calc. Var. Partial Differ. Equ. 57, No. 3, Paper No. 92, 1--27 (2018; Zbl 1395.35081)] for smaller values of \(t\).