On fine fractal properties of generalized infinite Bernoulli convolutions (Q2378595)

The paper is devoted to the investigation of generalized infinite Bernoulli convolutions, i.e., the distributions \(\mu=\mu_\xi\) of the following random variables: \[ \xi=\sum_{k=1}^\infty \xi_k a_k, \] where \(a_k\) are terms of positive convergent series, and \(\xi_k\) are independent random variables with \(P(\xi_k=0)=p_{0k}\), \(P(\xi_k=1)=p_{1k}=1-p_{0k}\). First, the authors study the topological and fractal properties of the spectrum (topological support) \(S_\mu\) of \(\mu\). It is shown, that if \((a_k)_{k\geq 0}\) is monotone non-decreasing then \(S_\mu\) is nowhere dense if and only if \(a_k>r_k:=\sum_{j=k+1}^\infty a_k\) holds for an infinite number of indices. Secondly, the upper bound for the Hausdorff-Bescovitch dimension of \(S_\mu\) is given, as well as the condition when this bound in achieved. Thirdly, under the condition \(a_k\geq r_k\) for all sufficiently large \(k\), the expression for the Hausdorff dimension for the measure \(\mu\) is found. Finally, the minimal dimensional supports of \(\mu\) are studied, and the conditions for \(S_\mu\) to be a minimal dimensional support are found. The results obtained in the paper are closely related to [\textit{M. J. Cooper}, Math. Proc. Camb. Philos. Soc. 124, No. 1, 135--149 (1998; Zbl 0912.28006)].