An optimal uncertainty principle in twelve dimensions via modular forms (Q2316826): Difference between revisions

The authors study an uncertainty relation -- a statement that a function \(f\) and its Fourier transform \(\hat{f}\) cannot both be `pinned down' simultaneously -- in the following setting: \begin{itemize} \item[-] a function \(f:\mathbb{R}^d\rightarrow\mathbb{R}\) is \textit{eventually nonegative} (resp. \textit{nonpositive}) if \(f(x)\geq 0\) (resp. \(f(x)\leq 0\)) for all sufficiently large \(|x|\), \item[-] then the radius of its last sign change is \[ r(f)=\inf\{R\geq 0: f(x)\text{ has the same sign for }|x|\geq R\}, \] \item[-] its Fourier transform is given by \[ \hat{f}(\xi)=\int_{\mathbb{R}^d}\,f(x)e^{-2\pi i\langle x,\xi\rangle}dx, \] \item[-] \(\mathcal{A}_{+}(d)\) denotes the set of functions \(f:\mathbb{R}^d\rightarrow\mathbb{R}\) such that \begin{itemize} \item[a.] \(f, \hat{f}\in L^1(\mathbb{R}^d)\) and \(\hat{f}\) is real-valued (i.e. \(f\) is even), \item[b.] \(f\) is eventually nonnegative while \(\hat{f}(0)\leq 0\), \item[c.] \(\hat{f}\) is eventually nonnegative while \(f(0)\leq 0\). \end{itemize} \item[-] \(\mathcal{A}_{-}(d)\) denotes the set of functions \(f:\mathbb{R}^d\rightarrow\mathbb{R}\) such that \begin{itemize} \item[a.] \(f, \hat{f}\in L^1(\mathbb{R}^d)\) and \(\hat{f}\) is real-valued (i.e. \(f\) is even), \item[b.] \(f\) is eventually nonnegative while \(\hat{f}(0)\leq 0\), \item[c.] \(\hat{f}\) is eventually nonpositive while \(f(0)\geq 0\). \end{itemize} \item[-] the uncertainty principle of Bourgain, Clozel and Kahane [\textit{J. Bourgain} et al., Ann. Inst. Fourier 60, No. 4, 1215--1232 (2010; Zbl 1298.11105)] concerns \[ A_{\pm}(d)=\inf_{f\in\mathcal{A}_{\pm}(d)\setminus\{0\}}\,\sqrt{r(f)r(\hat{f})}>0. \] \item[-] the minimality problem is Problem 1.1 (\(+1\) eigenfunction principle). Minimize \(r(g)\) over all \(g:\mathbb{R}^d\rightarrow\mathbb{R}\) with \begin{itemize} \item[a.] \(g\in L^{1}(\mathbb{R}^d)\setminus\{0\}\) and \(g=\hat{g}\), \item[b.] \(g(0)=0\) and \(g\) is eventually nonnegative. \end{itemize} \end{itemize} Then \(A_{+}(d)\) turns out to be the minimal value of \(r(g)\) with \(r(g)=r(\hat{g})=A_{+}(d)\). Problem 1.3 (\(-1\) eigenfunction principle). Minimize \(r(g)\) over all \(g:\mathbb{R}^d\rightarrow\mathbb{R}\) with \begin{itemize} \item \(g\in L^{1}(\mathbb{R}^d)\setminus\{0\}\) and \(\hat{g}=-g\), \item \(g(0)=0\) and \(g\) is eventually nonnegative. \end{itemize} Although upper and lower bounds for \(A_{+}\) are known, the authors are the first to determine the exact value in any dimension, given in Theorem 1.2. We have \(A_{+}(12)=\sqrt{2}\). In particular, there exists a radial Schwartz function \(f:\mathbb{R}^{12}\rightarrow\mathbb{R}\) that is eventually nonnegative and satisfies \(\hat{f}=f,\,f(0)=0\) and \(r(f)=r(\hat{f})=\sqrt{2}\). Moreover, as a radial function \(f\) has a double root at \(|x|=0\), a single root at \(|x|=\sqrt{2}\), and double roots at \(|x|=\sqrt{2j}\) for integers \(j\geq 2\). Finally, the authors prove a result on the relative growth of \(A_{\pm}(d)\): Theorem 1.4. Let \(s\in\{\pm\}\). Then there exist positive constants \(c\) and \(C\) with \[ c\leq\frac{A_s(d)}{\sqrt{d}}\leq C, \] for all \(d\). Moreover, for each \(d\) there exists a radial function \(f\in\mathcal{A}_{\pm}(d)\setminus\{0\}\) with \(\hat{f}=sf,\,f((0)=0\), and \(r(f)=A_s(d).\) The layout of the paper is as follows \S1. Introduction \S2. The \(+1\) eigenfunction uncertainty principle in dimension \(12\) (9 pages) -- contains the proof of Theorem 1.2 \S3. The \(-1\) eigenfunction uncertainty principle (6 pages) \S4. Numerical evidence (4 pages) \S5. Summation formulas (6 pages) References (19 items)