Hilbert cubes in arithmetic sets (Q2633994)

If \(a_0\neq 0, a_1,\dots,a_d\) are elements of an additive group, the set \[ H(a_0;a_1,\dots,a_d):=a_0+\{0,a_1\}+\cdots+\{0,a_d\}=\left\{a_0+\sum\limits_{i=1}^d\varepsilon_i a_i: \varepsilon_i\in \{0,1\}\right\} \] is called a Hilbert cube of dimension of \(d\). For \(a_0 = 0\) it is convenient to slightly amend the definition to \[ H(0;a_1,\dots,a_d):=\left\{\sum\limits_{i=1}^d\varepsilon_i a_i: \varepsilon_i\in \{0,1\}, \sum\limits_{i=1}^d\varepsilon_i>0\right\} \] excluding the empty sum. This special case of a homogeneous Hilbert cube is also often called the set of `subset sums'. The question of the maximal size \(d\) such that there is some Hilbert cube \(H(a_0; a_1,\dots, a_d)\) in a given set \(S\) of integers has been frequently studied. In this paper, the authors focus on questions where the set \(S\) has some interesting arithmetic meaning, such as the set of squares, or the set of powerful numbers. {\parindent=0.7cm\begin{itemize}\item[(a)] For the set of squares they obtain \(d = O(\log \log N)\). Using previously known methods this bound could have been achieved only conditionally subject to an unsolved problem of \textit{P. Erdős} and \textit{R. Rado} [J. Lond. Math. Soc. 35, 85--90 (1960; Zbl 0103.27901)]. \item[(b)] For the set \(W\) of powerful numbers they show \(d= O((\log N)^2)\). \item[(c)] For the set \(V\) of pure powers they also show \(d=O((\log N)^2)\), but for a homogeneous Hilbert cube, with \(a_0=0\), this can be improved to \(d=O((\log \log N)^3/ \log \log \log N)\), when the \(a_i\) are distinct, and \(d = O((\log \log N)^4/(\log \log \log N)^2)\), generally. This compares with a result of \(d=O((\log N)^3/(\log \log N)^{1/2})\) in the literature. \item[(d)] For the set \(V\) they also solve an open problem of \textit{N. Hegyvári} and \textit{A. Sárközy} [Ramanujan J. 3, No. 3, 303--314 (1999; Zbl 0990.11011)], namely they show that \(V\) does not contain an infinite Hilbert cube. \item[(e)] For a set without arithmetic progressions of length \(k\) they prove \(d=O_k(\log N)\), which is close to the true order of magnitude. \end{itemize}}