On the existence problem of regular Gabor frames

Jaume de Dios Pont, Lukas Liehr, Mitchell A. Taylor · 2026

Read On the existence problem of regular Gabor frames by Jaume de Dios Pont, Lukas Liehr, Mitchell A. Taylor (2026) 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-2606-26052.txt.

Introduction

Jaume de Dios Pont, Lukas Liehr, and Mitchell Taylor answer a standing question about Gabor frames. A Gabor system shifts one window function across a lattice in time and frequency. Frames give stable, redundant representations of signals. The existence problem asks whether nice windows can generate frames on lattices of density above one in higher dimensions. Candidate windows live in respected spaces like the Schwartz space and the Feichtinger algebra.

Negative is the answer, under explicit criteria. For every dimension above one, the team lists lattice conditions under which a frame cannot be generated by any function with a continuous Zak transform. The result rests on characterizing when quasiperiodic functions share a common zero. That characterization may interest other problems too. A Lean 4 formalization of the main result accompanies the paper. Time-frequency analysis loses a hoped-for construction, but gains a sharp boundary for where to stop searching.

Catalog record

CREATORJaume de Dios Pont, Lukas Liehr, Mitchell A. Taylor
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE64 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2606.26052

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

On the existence problem of regular Gabor frames Jaume de Dios Pont, Lukas Liehr, Mitchell A. Taylor Abstract . For every dimension d > 1 d>1 , we establish explicit criteria on lattices Λ ⊂ ℝ 2 ​ d \Lambda\subset\mathbb{R}^{2d} with density D ⁡ ( Λ ) > 1 D(\Lambda)>1 such that no function with a continuous Zak transform generates a Gabor frame along Λ \Lambda . In particular, this gives a negative answer to the existence problem of Gabor frames with window functions in the Schwartz space, the Feichtinger algebra, and the Fourier-invariant Wiener space. Our result is based on a characterization of when a collection of quasiperiodic functions admits a common zero, which may be of independent interest. We also include a formalization of our main result in Lean 4. [ BoldFont = JuliaMono-Bold.ttf, ItalicFont = JuliaMono-RegularItalic.ttf, BoldItalicFont = JuliaMono-BoldItalic.ttf, Scale = MatchLowercase ] ean]lean4bgcolor=white eancode]lean4escapeinside=!!, breaklines, fontsize= 2020 Mathematics Subject Classification 42C15, 42C40, 55R40 1. Introduction and main results 1.1. For g\in L^{2}(\mathbb{R}^{d}) and \lambda=(x,\omega)\in{\mathbb{R}}^{2d} , let g_{\lambda}(t):=e^{2\pi i\omega\cdot t}g(t-x) denote the time-frequency shift of g with respect to \lambda . The Gabor system of g along a set \Lambda\subseteq{\mathbb{R}}^{2d} is the collection of time-frequency shifts of g with respect to the elements of \Lambda , i.e., \mathbf{G}(g,\Lambda):=\{g_{\lambda}:\lambda\in\Lambda\}. The function g is commonly referred to as the window function of \mathbf{G}(g,\Lambda) . A fundamental problem in time-frequency analysis is to determine for which g and \Lambda the system \mathbf{G}(g,\Lambda) forms a frame of L^{2}(\mathbb{R}^{d}) ; that is, when there exists constants A,B>0 such that for all f\in L^{2}({{\mathbb{R}}^{d}}) one has the inequalities A\|f\|^{2}\leq\sum_{\lambda\in\Lambda}|\langle f,g_{\lambda}\rangle|^{2}\leq B\|f\|^{2}. This question has been studied extensively under the structural assumption that \Lambda is a lattice, i.e., \Lambda=M\mathbb{Z}^{2d}=\{Mz:z\in{\mathbb{Z}}^{2d}\} for some invertible matrix M\in\mathbb{R}^{2d\times 2d} [ 21 , 5 , 23 , 19 , 20 ] . 1.2. In what follows, let \Lambda\subseteq{\mathbb{R}}^{2d} be a lattice and let D(\Lambda) be its density. It is well-known that if \mathbf{G}(g,\Lambda) is a frame, then \Lambda satisfies the necessary density condition D(\Lambda)\geq 1 [ 27 ] . A theorem of Bekka [ 4 ] , which is central to the study of Gabor frames, shows that the condition D(\Lambda)\geq 1 is also sufficient in an existential sense: whenever a lattice satisfies D(\Lambda)\geq 1 , then there exists some g\in{L^{2}({\mathbb{R}}^{d})} for which \mathbf{G}(g,\Lambda) is a frame for L^{2}(\mathbb{R}^{d}) . However, Bekka’s argument is non-constructive and it leaves open whether one may choose g with prescribed regularity or decay, such as belonging to the Schwartz space \mathcal{S}({{\mathbb{R}}^{d}}) . While the existence of a Gabor frame with window in \mathcal{S}({{\mathbb{R}}^{d}}) for lattices with critical density D(\Lambda)=1 is excluded by the Balian-Low theorem, the existence for lattices with supercritical density D(\Lambda)>1 constitutes a well-known open problem in dimensions d>1 (the existence of a Schwartz window in the case d=1 was settled affirmatively by Seip-Wallstén and Lyubarskii [ 42 , 43 , 37 ] ). Precisely, it is conjectured that for every lattice \Lambda with D(\Lambda)>1 there exists g\in\mathcal{S}({{\mathbb{R}}^{d}}) such that \mathbf{G}(g,\Lambda) is a frame [ 22 , p. 225] . Another formulation of the conjecture asks if one can choose g as an element of the Feichtinger algebra S_{0}({{\mathbb{R}}^{d}}) , which is the collection of all functions g\in{L^{2}({\mathbb{R}}^{d})} such that \int_{{\mathbb{R}}^{2d}}|\langle g,g_{z}\rangle|\,dz<\infty. This version of the existence problem of regular Gabor frames is recorded, for instance, in [ 29 ] . In fact, it is equivalent to the existence problem of Schwartz Gabor frames [ 18 , Proposition 4.4] . 1.3. In the present paper we consider the class of functions C_{Z}({{\mathbb{R}}^{d}}) which consists of all f\in{L^{2}({\mathbb{R}}^{d})} such that f has a continuous Zak transform, i.e., Zf(u,v):=\sum_{n\in\mathbb{Z}^{d}}f(u-n)e^{2\pi i\,n\cdot v} is continuous on {\mathbb{R}}^{d}\times{\mathbb{R}}^{d} . The space C_{Z}({{\mathbb{R}}^{d}}) is an enlargement of the Feichtinger algebra and contains the so-called Fourier invariant Wiener space W_{0}({{\mathbb{R}}^{d}}) , whose definition goes back to Wiener’s seminal work on Tauberian theorems [ 46 ] . A function f belongs to W_{0}({{\mathbb{R}}^{d}}) if \sum_{n\in{\mathbb{Z}}^{d}}\|f(\cdot+n)\|_{L^{\infty}([0,1]^{d})}<\infty,\quad\sum_{n\in{\mathbb{Z}}^{d}}\|\hat{f}(\cdot+n)\|_{L^{\infty}([0,1]^{d})}<\infty, where \hat{f} denotes the Fourier transform of f . Specifically, the aforementioned function spaces are related by the strict inclusions \mathcal{S}({{\mathbb{R}}^{d}})\subsetneq S_{0}({{\mathbb{R}}^{d}})\subsetneq W_{0}({{\mathbb{R}}^{d}})\subsetneq C_{Z}({{\mathbb{R}}^{d}}). (1.1) The fact that S_{0}({{\mathbb{R}}^{d}}) is a proper subspace of W_{0}({{\mathbb{R}}^{d}}) follows from a theorem of Losert [ 35 , Theorem 2] , who proved that when defining analogous spaces S_{0}(G) and W_{0}(G) on a locally compact abelian group G , the equality S_{0}(G)=W_{0}(G) occurs if and only if G is either compact or discrete. The remaining strict inclusions are discussed in Section 2.2 . 1.4. In contrast to what was expected from the literature and the series of positive results discussed below, we prove in every dimension d>1 that there is a general obstruction to the frame property of Gabor systems \mathbf{G}(g,\Lambda) . Our obstruction holds for window functions g belonging to the largest space in ( 1.1 ), i.e, g\in C_{Z}(\mathbb{R}^{d}) . As a consequence, this yields a negative resolution to the existence problem for Gabor frames in \mathcal{S}({{\mathbb{R}}^{d}}) and S_{0}({{\mathbb{R}}^{d}}) . Theorem 1.1 . For every d>1 there exists a lattice \Lambda\subset{\mathbb{R}}^{2d} with density D(\Lambda)>1 having the following property: there exists no g\in C_{Z}(\mathbb{R}^{d}) such that \mathbf{G}(g,\Lambda) is a frame for L^{2}(\mathbb{R}^{d}) . In combination with the theorems of Seip–Wallstén and Lyubarskii, we obtain that the property that every lattice in \mathbb{R}^{2d} with supercritical density admits a Gabor frame with window in C_{Z}(\mathbb{R}^{d}) is equivalent to the condition d=1 . This indicates a significant difference in the behavior of the existence problem for Gabor frames in higher dimensions compared to the univariate case. 1.5. Our theorem complements a series of positive results on the existence problem of regular Gabor frames. In dimension d=1 , Seip-Wallstén and independently Lyubarskii proved that the Gaussian Gabor system \mathbf{G}(e^{-\pi t^{2}},\Lambda) is a frame for every lattice \Lambda\subseteq\mathbb{R}^{2} satisfying D(\Lambda)>1 [ 42 , 43 , 37 ] . On the other hand, in dimensions d>1 , Gröchenig and Lyubarskii constructed lattices \Lambda\subseteq{\mathbb{R}}^{2d} with D(\Lambda)>1 for which the Gaussian Gabor system \mathbf{G}(e^{-\pi|t|^{2}},\Lambda) fails to be a frame [ 17 ] . This implies that the Gaussian cannot serve as a universal Schwartz window in all dimensions. The study of Gaussian Gabor frames in dimensions d>1 remains an active research field. See, for example, the recent contributions by Luef and Wang [ 36 ] and by Romero, Ulanovskii and Zlotnikov [ 41 ] . Pfander, Rashkov, and Wang demonstrated that in dimension d=2 and for separable lattices that satisfy specific geometrical conditions, one can construct smooth functions g with compact support such that \mathbf{G}(g,\Lambda) is a frame [ 40 ] . Their approach is based on an existence result of fundamental domains of pairs of lattices [ 25 ] (see also the recent contribution [ 16 ] on bounded fundamental domains). Bédos, Enstad, and van Velthoven showed in [ 3 ] that for every d\geq 1 and every lattice \Lambda\subseteq{\mathbb{R}}^{2d} satisfying Kleppner’s condition [ 32 ] , there exists a Schwartz function g such that \mathbf{G}(g,\Lambda) is a frame. Jakobsen and Luef proved the existence of a Schwartz function g generating a Gabor frame in the case of non-rational lattices [ 29 ] . Enstad, Thiel, and Vilalta obtained a partial positive result for rational lattices [ 12 ] (see also [ 13 ]

Read the complete text