A characterization of some sets of spectral synthesis (Q2657395)

For a closed subset \(E\) of \(\mathbb{R}\) one defines \(I(E)\) as the set of all \(f\in L^1(\mathbb{R})\) such that \({\widehat f}_{|E}=0\), and \(j(E)\) as the set of all \(f\in L^1(\mathbb{R})\) such that \(\widehat f\) vanishes in some neighborhood of \(E\). One defines \(I_0(E)=\overline{j(E)}\), where the closure is in \(L^1(\mathbb{R})\). By intersecting with \(H^1(\mathbb{R})\), we obtain the analogues of these sets for \(H^1(\mathbb{R})\). One calls \(E\) a set of spectral synthesis if \(I(E)=I_0(E)\). In a similar manner one may consider the question of spectral synthesis for closed subsets of \(\mathbb{R}^d\) \((d\ge 1)\). In this paper the authors give conditions under which a function belongs to \(I_0(E)\) or a set \(E\) is a set of spectral synthesis. For a closed set \(E\subset\mathbb{R}^d\), and \(\epsilon >0\), let \(E_\epsilon =\{t\colon 0<\rho (t,E)<\epsilon \}\) (\(\rho\) the Euclidean distance). In Theorem 3.1 it is shown that if \(g\in I(E)\cap L^1\) is such that \begin{enumerate} \item \(\liminf\limits_{\epsilon \to 0}\dfrac1{\epsilon^d}\displaystyle\int_{E_\epsilon}|\widehat g(t)|\, dt=0\), \item \(\widehat g(t)=0\) for all \(t\in E\), \end{enumerate} then \(g\in I_0(E)\). Theorem 4.1 shows that if \(g\in H^1\), and \[ \int_{E^c}\dfrac{|\widehat g(t)|}{\rho (t,E)^d}\, dt<\infty, \] and \(\widehat g(t)=0\), \(t\not\in K\), where \(K\subset\mathbb{R}^d\) is compact, then \[ \lim_{\epsilon \to 0}\frac1{\epsilon^d}\int_{E_\epsilon}|\widehat g(t)|\, dt=0, \] and \(g\in I_0(E)\). For a closed set \(E \subset\mathbb{R}\), let \[ Q(E)=\Biggl\{g\in H^1\cap I(E)\colon \int_{E^c}\dfrac{|\widehat g(t)|}{\rho (t,E)}\, dt<\infty \Biggr\}. \] Theorem 5.2 shows that \(\overline{Q(E)}=I_0(E)\), and hence \(E\) is a set of synthesis for \(L^1\) if and only if \(I(E)=\overline{Q(E)}\); and it is a set of synthesis for \(H^1\) if and only if \(H^1\cap I(E)=\overline{Q(E)}\) where the closure is in \(H^1\). Theorem 6.1 shows shows that \(Q(E)\cap H^1\) is a Banach space under the norm \(\|\cdot\|_1+\|\cdot\|_{H^1}\). A set \(E\) is called a \(Q\)-set if \(Q(E)\cap H^1\) is complete under \(\|\cdot\|_{H^1}\). Properties of such sets are studied. For example, if \(\partial E\) is a \(Q\)-set, so is \(E\), and if \(E\) and \(F\) are \(Q\)-sets, so is \(E\cup F\) (Corollaries 7.1,~7.2). The paper ends with some examples.