On large values of Weyl sums (Q2189494)

For $d\geq 2$, let $\mathbf{x}=(x_1, \ldots, x_d) \in \mathsf{T}_d = (\mathbb{R}/ \mathbb{Z})^d$ be a vector in the $d$-dimensional unit torus. A Weyl sum is an exponential sum of the form $$ S_d(\mathbf{x}; N) = \sum_{n=1}^N \mathbf{e}\left( x_1 n + \ldots + x_d n^d \right) $$ in which $\mathbf{e}(x)= \exp(2\pi i x)$. The starting point of the paper is the upper bound on Weyl sums $$ |S_d(\mathbf{x}; N) | \leq N^{1/2} (\log N)^{3/2 + o(1)}, \quad \text{as $N \rightarrow \infty$}$$ which holds for almost all $\mathbf{x} \in \mathsf{T}_d$ and which follows from a special case of the so called Menshov-Rademacher Theorem. The main aim of the authors is to shed light on the structure of the \textit{exceptional sets of coefficients} $(x_1, \ldots, x_d)$ with large values of Weyl sums for infinitely many $N$. It is shown that these sets are big in terms of the Baire categories (Theorem 1.3) and Hausdorff dimension (Theorem 1.5). Similar results are obtained with a different technique for sums with just one monomial $xn^d$ (Theorem 1.6 and Theorem 1.7). The authors apply their results to infinite sequences of points in the $d$-dimensional unit torus generated from a fixed $\mathbf{x}=(x_1, \ldots, x_d)$ and defined as $$ (x_1 n + \ldots + x_d n^d)_{n \in \mathbb{N}}.$$ The irregularities of distribution of such an infinite sequence are usually quantified via the discrepancy, $D_d(\mathbf{x};N)$, of the first $N$ elements of the sequence. The authors showed [``New bounds on Weyl sums'', Int. Math. Res. Not, rnz293 (2019)] that $$ D_d(\mathbf{x};N) \leq N^{1/2 + o(1)}, \quad \text{as $N \rightarrow \infty$}, $$ for almost all $\mathbf{x} \in \mathsf{T}_d$. The corresponding exceptional sets can now be analysed using the same methods for Weyl sums; i.e. in Theorem 1.9--1.12 the authors provide results on the exceptional sets along the same lines as above. As the authors point out, these results are particularly interesting because \textit{a new type of problems of metric number theory} is considered ``where the vectors of real numbers are classified by the size of the corresponding Weyl sums (...) rather than by their Diophantine approximation properties as in the classical settings.'' The comprehensive paper concludes with a list of open problems and conjectures for future work.