Spectrality of infinite Bernoulli convolutions (Q2355448)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Spectrality of infinite Bernoulli convolutions
scientific article

    Statements

    Spectrality of infinite Bernoulli convolutions (English)
    0 references
    0 references
    0 references
    0 references
    23 July 2015
    0 references
    Let \(\mu\) be a Borel probability measure on \(\mathbb R\) with compact support. Let \(\Lambda\) be a discrete set in \(\mathbb R\) and write \(e_\Lambda = \{e_\lambda: \lambda \in \Lambda\}\), where \(e_\lambda\) is the exponential function \(e^{-2\pi i\lambda x}\). The measure \(\mu\) is a \textit{spectral measure} if there exists \(\Lambda\) such that \(e_\Lambda\) is an orthonormal basis for \(L^2(\mu)\); in this case \(\Lambda\) is a spectrum of \(\mu\). In particular, if \(\mu\) is the normalized Lebesgue spectral measure supported on a Borel set \(\Omega\), then \(\Omega\) is called a spectral set. The authors study the spectrality of infinite Bernoulli convolutions. Let \(\{b_k - a_k\}_{k=1}^\infty\) be a sequence of positive integers with upper bound and let \(0 < \rho < 1\). Associated with them there is a probability measure \(\mu_{\rho,\{a_k,b_k\}}\) defined by \[ \mu_{\rho,\{a_k,b_k\}} = \delta_{\rho\{a_1,b_1\}}\ast \delta_{\rho^2\{a_1,b_1\}}\ast \dots \] where \(\delta_{\rho^k\{a_k,b_k\}}=\frac{1}{2}\delta_{\rho^k a_k}+\frac{1}{2}\delta_{\rho^k b_k}\) and \(\delta_a\) is the Dirac measure at \(a\), the sign \(\ast\) means the convolution and the convergence is in weak sense. If all \((a_k, b_k) = (0, 1)\), \(\mu_{\rho,\{a_k,b_k\}}\) is called \textit{infinite Bernoulli convolution}. Motivated by previous results as [\textit{X.-R. Dai}, Adv. Math. 231, No. 3--4, 1681--1693 (2012; Zbl 1266.42012); \textit{T.-Y. Hu} and \textit{K.-S. Lau}, Adv. Math. 219, No. 2, 554--567 (2008; Zbl 1268.42044)], the authors prove the following theorems: {Theorem 1.1.} Suppose that the infinite Bernoulli convolution \(\mu_{\rho,\{a_k,b_k\}}\) is a spectral measure. Then \(\rho=\frac{1}{2q}\) for some integer \(q \geq 1\). {Theorem 1.3.} Let \(b = 2^{l+1}q\) be an integer so that \(l>L\) if \(q = 1\) and \(l \geq L\) if the odd number \(q\) is bigger than \(1\). Then the infinite Bernoulli convolution \(\mu_{1/b,\{a_k,b_k\}}\) is a spectral measure. Some steps in the proofs are of independent interest. It should be noted that the Proposition~1.2, contained in the paper, gives an estimation for the infinite product of triangle functions.
    0 references
    0 references
    Borel probability measure
    0 references
    infinite Bernoulli convolution
    0 references
    orthonormal basis of exponential functions
    0 references
    Ramsey theorem
    0 references
    self-similar
    0 references
    spectral measure
    0 references
    spectrum
    0 references
    0 references
    0 references
    0 references
    0 references