Pointwise Fourier inversion and localisation in \(\mathbb{R}^n\) (Q1384986)

Let \(f\in L^2(\mathbb{R}^n)\), \(n\geq 2\), and \(\widehat f\) be its Fourier transform. The partial spherical inverse Fourier integrals are defined by \[ S_Rf(x)= \int_{| \xi| \leq R} \widehat f(\xi) e^{2\pi ix \cdot\xi} d\xi. \] In this paper, the authors study the localization problem: in what sense(s) does \(S_R f(x)\to 0\) as \(R\to \infty\) for \(x\in B\) if \(f\) is presumed to be identically zero on \(B=\{x\in \mathbb{R}^n: | x| <1\}\)? More precisely, the authors investigate the possible sets of divergence for the localization problem. Definition. \(E\subseteq B\) is a Set of Divergence for the Localization Problem (SDLP) if there exists an \(f\in L^2 (\mathbb{R}^n)\), with \(f\equiv 0\) on \(B\), such that \(S_Rf(x)\) diverges as \(R\to \infty\) for all \(x\in E\). It is very difficult to classify the SDLP's entirely, so the authors give some partial results. They establish several theorems, which provide some sufficient conditions to classify SDLP's. Two main results are Theorem 5 and Theorem 10. Theorem 5. Let \(n\geq 2\). If \(E\) is contained in a set of the form \([-1,1]\times E'\), where \(E'\subseteq [-1,1]^{n-1}\) has \((n-1)\) dimensional measure zero, then \(E\) is an SDLP. Theorem 10. Let \(n\geq 2\). Then the following conditions for a positive finite Borel measure \(\mu\) on \((0,1)\) are equivalent: (i) \(\sup_r \int^1_r s^{1/2} (s-r)^{-1/2} d\mu(s) <\infty\); (ii) For all \(f\in L^2(\mathbb{R}^n)\) with \(f \equiv 0\) on \(B\), \(\|\sup_{R\geq 1} | S_R f(x) | \|_{L^2 (d\widetilde\mu)} \leq C\| f\|_2\); (iii) For all \(f\in L^2 (\mathbb{R}^n)\) with \(f\equiv 0\) on \(B\), \(\sup_{R\geq 1} \| S_Rf \|_{L^2 (d\widetilde \mu)} \leq C\| f\|_2\), where \(\widetilde\mu\) is the radial version of \(\mu\).