The range of Rademacher series in \({\mathbb R}^d\) (Q2421712)

For a sequence $(a_n)$ in $\mathbb R^d$ with $\lim_{n \to \infty} a_n = 0$ and $\sum_{n \in \mathbb N} \|a_n\| = \infty$ the author studies the Rademacher range of $(a_n)$, that is the set $\mathcal R(a_n)$ of those $x \in \mathbb R^d$ that have a representation of the form $x = \sum_{n=1}^\infty t_n a_n$ with $t_n = \pm 1$. A vector $v \in \mathbb R^d$ is called a direction of $(a_n)$ if there exist a subsequence $n_1 < n_2 < \ldots$ and an index $j \in \{1, \ldots, d\}$ such that $\sum_{k \in \mathbb N} \|a_{n_k}\| = \infty$ and $\lim_{k \to \infty}\left(a_{n_k}^j\right)^{-1} a_n = v$, where $a_{n_k}^j$ denotes the $j$th coordinate of the vector $a_{n_k}$. The main result of the paper (Theorem 1.6) says that, for $d \ge 2$ and $(a_n)$ as above, the Hausdorff dimension of the Rademacher range is in $[1, d]$. Moreover, for each $s \in [1, d]$, there exists a sequence $(a_n)$ with only one direction such that $\mathcal R(a_n)$ has Hausdorff dimension $s$, is dense in $\mathbb R^d$ but does not coincide with $\mathbb R^d$. Remark, that Rademacher ranges are closely connected with the achievement sets \[ A(b_n) =\left\{\sum_{n=1}^\infty s_n b_n : (s_n) \in \{0, 1\}^{\mathbb N}\right\}. \] Indeed, the condition $\lim_{n \to \infty} a_n = 0$ implies by the Dvoretzky-Hanani theorem (see Theorem 2.2.1 of the book [\textit{M. I. Kadets} and \textit{V. M. Kadets}, Series in Banach spaces: conditional and unconditional convergence. Transl. from the Russian by Andrei Iacob. Basel: Birkhäuser (1997; Zbl 0876.46009)]) the existence of $\theta_n = \pm 1$ such that $\sum_{n=1}^\infty \theta_n a_n$ converges. Taking into account that for every $t_n = \pm 1$ either $\theta_n + t_n = 0$, or $\theta_n + t_n = 2 \theta_n$, one obtains that \[ \mathcal R(a_n) + \sum_{n=1}^\infty \theta_n a_n = 2A(\theta_n a_n). \] This means that $\mathcal R(a_n)$ and $A(\theta_n a_n)$ have the same topological properties. It looks like that the people dealing with $A(a_n)$ [\textit{A. Bartoszewicz} et al., Colloq. Math. 152, No. 2, 235--254 (2018; Zbl 1408.40001); \textit{S. Gołąb} and \textit{J. Marchwicki}, J. Math. Anal. Appl. 459, No. 1, 476--489 (2018; Zbl 1385.40001)] and with $\mathcal R(a_n)$ ([\textit{C. Liu}, Acta Math. Sin., Chin. Ser. 58, No. 5, 705--716 (2015; Zbl 1340.28004)], [\textit{X.-G. He} and \textit{C.-T. Liu}, Ann. Acad. Sci. Fenn., Math. 40, No. 1, 135--148 (2015; Zbl 1332.28009)] and the paper under review do not know that they study essentially the same object.