Multiplicatively badly approximable numbers and generalised Cantor sets (Q645221)

The authors discuss certain sets related to the mixed Littlewood conjecture of \textit{B. de Mathan} and \textit{O. Teulié} [Monatsh. Math. 143, No. 3, 229--245 (2004; Zbl 1162.11361)]. This conjecture states that for any \(\alpha \in {\mathbb R}\), \[ \liminf_{q \rightarrow \infty} q | q |_{\mathcal D} \| q \alpha \| = 0, \] where \(q\) runs over the natural numbers, \(\| x \|\) denotes the distance from \(x\) to the nearest integer, and \(| q |_{\mathcal D}\) denotes the pseudo-absolute value of \(q\) associated with the sequence \({\mathcal D}\). The latter is an extension of the notion of a \(p\)-adic absolute value and is defined via a sequence \({\mathcal D} = (d_k)\) of integers greater than or equal to \(2\). Letting \(D_0 = 1\) and \(D_k = \prod_{i=1}^k d_i\), we set \[ | q |_{\mathcal D} = \inf \{D_n^{-1} : q \in D_n {\mathbb Z}\}. \] If \({\mathcal D}\) is the constant sequence \(p\) with \(p\) a prime, this is the usual \(p\)-adic absolute value. The conjecture has strong ties with the classical Littlewood conjecture. The authors prove that the mixed Littlewood conjecture cannot be improved by much. For general sequences \({\mathcal D}\), it is shown that the set of \(\alpha\) for which \[ \liminf_{q \rightarrow \infty} q \log q \log\log q | q |_{\mathcal D} \| q \alpha \| > 0, \] has Hausdorff dimension \(1\). It is also shown that there is an uncountable number of unbounded sequences \({\mathcal D}\) such that for one of these sequences the set of \(\alpha\) for which \[ \liminf_{q \rightarrow \infty} q \log\log q \log\log\log q | q |_{\mathcal D} \| q \alpha \| > 0, \] again has Hausdorff dimension \(1\). In other words, the mixed Littlewood conjecture cannot be improved by the factor \(\log q \log\log q\) for any sequence \({\mathcal D}\). For a large number of sequence it is not even possible to improve the conjecture by \(\log\log q \log\log\log q\). The proof depends on a general result on the Hausdorff dimension of a rather general class of Cantor sets, which is of major interest in its own right. Several interesting open problems are listed at the end of the paper.