Exponential Riesz bases, discrepancy of irrational rotations and BMO (Q650075)

Given an interval \(I=[a,b)\), let \(\Lambda= \Lambda(\alpha, I)= \{n \in\mathbb Z: a< n\alpha < b\}\) denote a simple quasicrystal, and set \[ E(\Lambda) = \left\{ e^{2 \pi i \lambda t}: \lambda \in \Lambda\right\}. \] Improving results of \textit{B. Matei} and \textit{Y. Meyer} [Rev. Mat. Iberoam. 25, No. 2, 669--692 (2009; Zbl 1184.42010)], in the paper under review the authors prove the following results: Theorem. Let \(|I| \in\mathbb Z\alpha +\mathbb Z\). Then the exponential system \(E(\Lambda(\alpha,I))\) is a Riesz basis in \(L^2(S)\) for every set \(S\subset\mathbb T\), \(\operatorname{mes} S = |I|\), which is the union of finitely many disjoint intervals whose lengths belong to \(\mathbb Z\alpha +\mathbb Z\). Theorem. Suppose that \(|I| \not\in\mathbb Z\alpha +\mathbb Z\). Then \(E(\Lambda(\alpha,I))\) is not a Riesz basis in \(L^2(S)\) for any set \(S \subset\mathbb T\) which is the union of finitely many intervals. For the proof, the authors utilize a generalized version of a classical result of \textit{H. Kesten} on the discrepancy of \(n \alpha\)-sequenes [Acta Arith. 12, 193--212 (1966; Zbl 0144.28902)], together with an idea of Meyer for which they coin the phrase ``Meyer's duality principle''. Further ingredients are characterizations of Riesz bases by Avdonin and Pavlov, see [\textit{S. A. Avdonin}, Vestn. Leningr. Univ. 1974, No. 13, Mat. Mekh. Astron. 1974, No. 3, 5--12 (1974; Zbl 0296.46033); Vestn. Leningr. Univ., Math. 7, 203--211 (1979; Zbl 0427.46016)] and [\textit{B. S. Pavlov}, Sov. Math., Dokl. 20, 655--659 (1979); translation from Dokl. Akad. Nauk SSSR 247, 37--40 (1979; Zbl 0429.30004)].