Directional maximal function along the primes (Q2232025)

The directional maximal operator over a finite set of directions in the plane, \(V\subseteq \mathbb{S}^1\), is given by \[ M_V f(x) = \sup\limits_{v\in V}\sup\limits_{r>0} \left| \frac{1}{2r} \int_{-r}^{r} f(x - tv) dt \right| \] where \(f:\mathbb{R}^2 \rightarrow \mathbb{C}\). Part of its intrigue lies in its relation to the Kakeya maximal function in the plane. \textit{N. Katz} [Duke Math. J. 97, No. 1, 67--79 (1999; Zbl 0942.42009)] showed the following sharp bound for any finite set of directions \[ \|M_V f \|_{L^2(\mathbb{R}^2)} \leq C\log(|V|) \| f \|_{L^2(\mathbb{R}^2)} \] where \(C>0\) is some constant and \(|V|\) denotes the number of directions in \(V\). Motivated by the directional maximal operator the authors in the paper at hand study a discrete analogue where the dilates of the direction are taken over prime numbers. Let \(k\in\mathbb{N}\), \(V\subseteq\mathbb{Z}^2\) be a finite collection of lattice points, \(\phi\) a smooth, compactly supported function on the real line and denote by \(\mathbb{P}\) the set of all prime numbers. The operator the authors study is given by \[ A^{\mathbb{P}}_{V,k}f(x) = \sup\limits_{k\geq k_V} \sup\limits_{v\in V} \left| \sum\limits_{p\in\mathbb{P}} f(x - pv)2^{-k}\phi(2^{-k}p)\log p \right| \] where \(k_V\) is large, depending on \(|V|\), and \(f:\mathbb{Z}^2\rightarrow\mathbb{C}\) is a finitely supported function. The main result of the paper is a construction of a set of directions, that is robust enough to allow the study of all sufficiently large scales, where a bound holds, analogous to that obtained by Katz in the continuous setting. More precisely, given \(\epsilon > 0\) and \(N>0\) then there exist constants \(C_{\epsilon}\) and \(C_0\) such that for any \(A \geq N^{C_0}\) there exists a set of vectors \(V \subseteq \lbrace x\in\mathbb{Z}^2 : \frac{1}{100} A \leq |x| \leq 100 A \rbrace\) with \(|V| = N\) so that \[ \| \sup_{k\geq k_V} A^{\mathbb{P}}_{V,k}f \|_{\ell^2(\mathbb{Z}^2)} \leq C_{\epsilon}N^{\epsilon} \| f \|_{\ell^2(\mathbb{Z}^2)} \] provided \(k_V \geq N^{C_1}\) for a sufficiently large constant \(C_1\) depending on \(C_0\). \textit{L. Cladek} and \textit{B. Krause} [``Discrete analogues in harmonic analysis: directional maximal functions in \(\mathbb Z^2\)'', Preprint, \url{arXiv:1901.06070}] had previously obtained similar results for an analogous operator, where the set of dilations came from all integers. In that setting they also obtained some partial results for thinner sequences, such as only drawing dilations from the squares. Here the authors obtain the first full such result for a thin sequence. A key ingredient is using the high degree of randomness exhibited by the primes.