Dimension-free estimates for discrete maximal functions related to normalized gaussians
Mariusz Mirek, Tomasz Z. Szarek, Błażej Wróbel · 2025
Read Dimension-free estimates for discrete maximal functions related to normalized gaussians by Mariusz Mirek, Tomasz Z. Szarek, Błażej Wróbel (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-11259.txt.
Introduction
Three authors prove dimension free bounds for discrete maximal functions in this 2025 math paper. The functions are built from sampled Gaussian kernels, tied to the theta function, and the goal is estimates that do not worsen as the dimension grows.
The team covers maximal, jump, variational, and oscillation inequalities on sequence spaces, for the full range of exponents. Their case is the first discrete maximal function of this non-semigroup kind to get such bounds across the whole range. The proofs rest on robust Fourier methods joined with a fractional derivative tool, which had not been used on discrete problems of this type before.
Catalog record
| CREATOR | Mariusz Mirek, Tomasz Z. Szarek, Błażej Wróbel |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 99 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-11259
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Dimension-free estimates for discrete maximal functions related to normalized gaussians Thanks: Mariusz Mirek was partially supported by the NSF grant (DMS-2154712) and by the NSF CAREER grant (DMS-2236493). Tomasz Z. Szarek and Błażej Wróbel were supported by the National Science Centre, Poland, grant Sonata Bis 2022/46/E/ST1/00036. Mariusz Mirek, Tomasz Z. Szarek, Błażej Wróbel Abstract . In this paper, we investigate dimension-free estimates for maximal operators of convolutions with discrete normalized Gaussians (related to the Theta function) in the context of maximal, jump and r r -variational inequalities on ℓ p ( ℤ d ) \ell^{p}(\mathbb{Z}^{d}) spaces. This is the first instance of a discrete operator in the literature where ℓ p ( ℤ d ) \ell^{p}(\mathbb{Z}^{d}) bounds are provided for the entire range of 1 < p < ∞ 1<p<\infty . The methods of proof rely on developing robust Fourier methods, which are combined with the fractional derivative, a tool that has not been previously applied to studying similar questions in the discrete setting. 2020 Mathematics Subject Classification 42B25, 42B15 1. Introduction 1.1. Statement of the main results Let \mathbb{T}:=\mathbb{R}/\mathbb{Z} denote one-dimensional torus, which will be identified with [-1/2,1/2) . For t>0 and \zeta\in\mathbb{T} we define a one-dimensional Theta function by \displaystyle\theta_{t}(\zeta):=\sum_{k\in\mathbb{Z}}\exp(-\pi(k-\zeta)^{2}t). For \xi\in\mathbb{T}^{d} we also define its d -dimensional variant by (1.1) \Theta_{t}(\xi):=\prod_{j\in[d]}\theta_{t}(\xi_{j})=\sum_{k\in\mathbb{Z}^{d}}\exp(-\pi|k-\xi|^{2}t), where |\cdot| denotes the Euclidean norm. For t>0 the discrete normalized Gaussian kernel g_{t}\colon\mathbb{Z}^{d}\to\mathbb{R} is defined by \displaystyle g_{t}(n):=\frac{1}{\Theta_{1/t}(0)}\exp\bigg(-\frac{\pi|n|^{2}}{t}\bigg),\qquad n\in\mathbb{Z}^{d}, and the corresponding convolution operator on \mathbb{Z}^{d} with kernel g_{t} is defined by G_{t}f(x):=g_{t}*f(x)=\sum_{n\in\mathbb{Z}^{d}}g_{t}(n)f(x-n),\qquad x\in\mathbb{Z}^{d},\quad f\in\ell^{1}(\mathbb{Z}^{d}). We immediately see that G_{t} is a contraction on all \ell^{p}(\mathbb{Z}^{d}) spaces for all 1\leq p\leq\infty, that is (1.2) \sup_{t>0}\|G_{t}\|_{\ell^{p}\to\ell^{p}}\leq 1, which follows from Minkowski’s convolution inequality and the fact that \|g_{t}\|_{\ell^{1}}=1 for any t>0 . The main goal of this paper is to prove the following dimension-free maximal, jump, variational and oscillation estimates. See Section 2.5 for the definitions of these objects. Theorem 1.1 . For each p\in(1,\infty) there exists C_{p}>0 independent of the dimension d\in\mathbb{Z}_{+} such that (1.3) \displaystyle\sup_{\lambda>0}\big\|\lambda N_{\lambda}(G_{t}f:t>0)^{1/2}\big\|_{\ell^{p}}\leq C_{p}\|f\|_{\ell^{p}},\qquad f\in\ell^{p}(\mathbb{Z}^{d}). In particular, for each p\in(1,\infty) and r\in(2,\infty) there exists C_{p,r}>0 independent of the dimension d\in\mathbb{Z}_{+} such that \displaystyle\big\|V^{r}(G_{t}f:t>0)\big\|_{\ell^{p}}\leq C_{p,r}\|f\|_{\ell^{p}},\qquad f\in\ell^{p}(\mathbb{Z}^{d}). Further, for each p\in(1,\infty) there exists C_{p}>0 independent of the dimension d\in\mathbb{Z}_{+} such that (1.4) \displaystyle\sup_{J\in\mathbb{Z}_{+}}\sup_{I\in\mathfrak{S}_{J}((0,\infty))}\big\|O_{I,J}^{2}(G_{t}f:t>0)\big\|_{\ell^{p}}\leq C_{p}\|f\|_{\ell^{p}},\qquad f\in\ell^{p}(\mathbb{Z}^{d}). Furthermore, for each p\in(1,\infty) there exists C_{p}>0 independent of the dimension d\in\mathbb{Z}_{+} such that (1.5) \displaystyle\big\|\sup_{t>0}|G_{t}f|\big\|_{\ell^{p}}\leq C_{p}\|f\|_{\ell^{p}},\qquad f\in\ell^{p}(\mathbb{Z}^{d}). 1.2. Historical background and our motivations In order to explain why we are undertaking Theorem 1.1 , we need to briefly describe the historical background of dimension-free estimates in harmonic analysis. Let U be a closed, bounded, symmetric, and convex subset of \mathbb{R}^{d} with non-empty interior, which we simply call a symmetric convex body. We set U_{t}:=\{y\in\mathbb{R}^{d}:t^{-1}y\in U\} for every t>0 and, for every x\in\mathbb{R}^{d} and f\in L^{1}_{\rm loc}(\mathbb{R}^{d}) , we define the Hardy–Littlewood averaging operator by (1.6) \displaystyle M_{t}^{U}f(x):=\frac{1}{|U_{t}|}\int_{U_{t}}f(x-y)\,dy. A typical choice is U=B^{q}(d) for q\in[1,\infty] , where B^{q}(d) is the unit ball induced by the \ell^{q} norm in \mathbb{R}^{d} , see ( 2.1 ) and ( 2.2 ). Specifically, B^{2}(d) is the Euclidean unit ball and B^{\infty}(d) is the cube in \mathbb{R}^{d} . For \emptyset\neq\mathbb{I}\subseteq\mathbb{R}_{+} and p\in[1,\infty] , let 0<C(p,\mathbb{I},U)\leq\infty be the smallest constant for the following maximal inequality \displaystyle\big\|\sup_{t\in\mathbb{I}}|M_{t}^{U}f|\big\|_{L^{p}(\mathbb{R}^{d})}\leq C(p,\mathbb{I},U)\|f\|_{L^{p}(\mathbb{R}^{d})},\qquad f\in L^{p}(\mathbb{R}^{d}). Using a standard covering argument (Vitali covering lemma) for p=1 and interpolation with p=\infty , it is not hard to see that C(p,\mathbb{R}_{+},U)<\infty for every p\in(1,\infty] since C(\infty,\mathbb{R}_{+},U)=1 . Similarly, for every t>0 let U_{t}:=\{y\in\mathbb{R}^{d}\colon t^{-1}y\in U\} denote the dilate of a set U\subseteq\mathbb{R}^{d} . For x\in\mathbb{Z}^{d} , we define a discrete analogue of ( 1.6 ) by \displaystyle\mathcal{M}_{t}^{U}f(x):=\frac{1}{|U_{t}\cap\mathbb{Z}^{d}|}\sum_{y\in U_{t}\cap\mathbb{Z}^{d}}f(x-y),\qquad f\in\ell^{1}(\mathbb{Z}^{d}), where the right-hand side above is understood as 0 if U_{t}\cap\mathbb{Z}^{d}=\emptyset . The corresponding maximal operator is given by \mathcal{M}^{U}f(x):=\sup_{t>0}|\mathcal{M}_{t}^{U}f(x)| . For \emptyset\neq\mathbb{I}\subseteq\mathbb{R}_{+} , U\subseteq\mathbb{R}^{d} and p\in[1,\infty] , let 0<\mathcal{C}(p,\mathbb{I},U)\leq\infty be the smallest constant for the following maximal inequality \displaystyle\big\|\sup_{t\in\mathbb{I}}|\mathcal{M}^{U}_{t}f|\big\|_{\ell^{p}(\mathbb{Z}^{d})}\leq\mathcal{C}(p,\mathbb{I},U)\|f\|_{\ell^{p}(\mathbb{Z}^{d})},\qquad f\in\ell^{p}(\mathbb{Z}^{d}). If U\subseteq\mathbb{R}^{d} is a symmetric convex body, then using similar arguments as in the continuous setting yield \mathcal{C}(p,\mathbb{R}_{+},U)<\infty for every p\in(1,\infty] . A major open problem in this area since the late 1970’s arose from Stein’s work [ 26 ] and asserts that: Dimension-free conjecture . For every p\in(1,\infty] there is a constant C_{p}>0 such that (1.7) \displaystyle\sup_{d\in\mathbb{Z}_{+}}\sup_{U\in\mathfrak{B}(d)}C(p,\mathbb{R}_{+},U)\leq C_{p} where \mathfrak{B}(d) is the set of all convex symmetric bodies in \mathbb{R}^{d} . This problem has been studied for four decades by several authors. We now briefly list the current state of the art concerning ( 1.7 ). 1. Stein [ 26 ] (see also Stein and Strömberg [ 29 ] ) proved that there exists a constant C_{p}>0 depending only on p\in(1,\infty] such that \sup_{d\in\mathbb{N}}C(p,\mathbb{R}_{+},B^{2}(d))\leq C_{p} . The key idea from [ 26 , 29 ] is based on the method of rotations, which enables one to view high-dimensional spheres as an average of rotated low-dimensional ones. This in turn implies dimension-free estimates for the maximal spherical function. Unfortunately, this method is limited to the Euclidean balls. 2. Bourgain [ 3 , 4 ] and Carbery [ 10 ] independently proved ( 1.7 ) for p\in(3/2,\infty] . They also showed a dyadic variant of ( 1.7 ) for p\in(1,\infty] with \mathbb{D}=\{2^{n}:n\in\mathbb{Z}\} in place of \mathbb{R}_{+} . Although inequality ( 1.7 ) for p\in(1,3/2] still remains open, the case of q -balls is quite well understood. 3. For the q -balls B^{q}(d) , we know that for every p\in(1,\infty] and q\in[1,\infty] , there is a constant C_{p,q}>0 such that \sup_{d\in\mathbb{N}}C(p,\mathbb{R}_{+},B^{q}(d))\leq C_{p,q} . This was established by Müller in [ 21 ] (for q\in[1,\infty) ) and by Bourgain in [ 5 ] (for cubes q=\infty ). 4. The first and third authors with Bourgain and Stein [ 6 ] extended the results from [ 3 , 4 , 5 , 10 , 21 ] to r -variational estimates with bounds independent of the dimension. We also found new proofs of maximal theorems from [ 3 , 4 , 10 ] . The methods from [ 6 ] also shed new light on dimension-free estimates in the discrete setup, which is a new and interesting direction. Several years ago the first and third authors in a collaboration with Bourgain and Stein [ 7 , 8 , 9 ] initiated the study of dimension-free estimates in the discrete setting. This line of research was partially motivated by a question of E.M. Stein [ 28 ] from the mid 1990’s about the dimension-free estimates for the discrete Hardy–Littlewood maximal functions corresponding to the Euclidean balls B^{2}(d) . Stein’s question . Is it true that there is a universal constant C>0 such that (1.8) \displaystyle\sup_{d\in\mathbb{N}}\mathcal{C}(2,\mathbb{R}_{+},B^{2}(d))=\sup_{d\in\mathbb{N}}\mathcal{C}(2,\sqrt{\mathbb{N}_{0}},B^{2}(d))\leq C\,? We now briefly mention existing results in the discrete setting and their relation to ( 1.7 ) and ( 1.8 ). 1. In [ 8 ] , it was shown that ( 1.7 ) cannot be expected in the discrete setup. Surprisingly, in contrast to the continuous setup [ 3 , 4 , 5 , 6 , 10 , 21 , 26 , 25 ] , one can prove [ 8 ] that for every p\in(1,\infty) , there is a constant C_{p}>0 such that for certain ellipsoids E(d)\subset\mathbb{R}^{d} and all d\in\mathbb{Z}_{+} , one has (1.9) \displaystyle\mathcal{C}(p,\mathbb{R}_{+},E(d))\geq C_{p}(\log d)^{1/p}. On the other hand, for the cubes B^{\infty}(d) , it was also shown