On the zeros of the Fourier transforms of Cantor-Lebesgue measures (Q2565301)

Let us consider the Fourier transform \[ \widehat \mu(z)= \int^a_{-a} e^{izt} dF_\mu(t), \quad z\in C^1, \] of a measure \(\mu\) with its distribution function \(F_\mu(t)\) such that \(\text{supp} \mu \subset [-a,a]\) and the points \(\pm a\) are growth points of \(F_\mu (t)\). \(\mu\) is said to be in the Cantor-Lebesgue class if there exist \(\delta= \delta(\mu)\), \(0<\delta<a\), and \(\xi=\xi(\mu)\), \(0<\xi<\frac12\), such that \(F_\mu(t)\equiv L(t)\), \(t\in[-a,-a+\delta)\cup (a-\delta,a]\), where \(L(t)\) is the Lebesgue function constructed for the perfect set of the interval \([-a,a]\) with the constant ratio of dissection \(\xi\). The main result of the paper is the following. Theorem 1. Let the measure \(\mu\) be in the Cantor-Lebesgue class. Then all of the zeros of \(\widehat \mu(z)\) lie in the region \[ |\text{Im} z|<C_\mu \log^2 |\text{Re} z|, \quad C_\mu>0. \] The constant \(C_\mu\) depends only on \(\delta(\mu)\) and \(\xi(\mu)\).