TL;DR
This paper tackles a fundamental question in harmonic analysis: when can a set in one dimension be both a "spectral set" (allowing for a special type of orthonormal basis) and a "tile" (meaning it can perfectly fill space without overlap)? The authors prove a significant result: for one-dimensional spectral sets, the associated "spectrum" (a set of frequencies) must be rational. This resolves a long-standing conjecture and establishes a crucial link between spectral sets on the real line and their counterparts on finite cyclic groups.
What Problem Does This Paper Address?
The paper delves into the fascinating world of spectral sets and tiling problems, particularly in one dimension. A set $\Omega \subset \mathbb{R}^d$ is called a spectral set if there exists a countable set of frequencies $\Lambda$ such that the complex exponentials $\{e^{2\pi i \langle \lambda, x \rangle} : \lambda \in \Lambda\}$ form an orthonormal basis for the space of square-integrable functions on $\Omega$, denoted $L^2(\Omega)$. Such a set $\Lambda$ is called a spectrum of $\Omega$.
Simultaneously, a set $\Omega$ is a translational tile if there exists a countable set of translations $T$ such that $\Omega$ perfectly covers $\mathbb{R}^d$ without overlap, i.e., $\mathbb{R}^d = \bigcup_{t \in T} (\Omega + t)$ and $(\Omega + t) \cap (\Omega + t') = \emptyset$ for $t \neq t'$.
The Fuglede Conjecture (1974) posits that for any set $\Omega \subset \mathbb{R}^d$, it is a spectral set if and only if it is a translational tile. While this conjecture has been disproven in higher dimensions ($d \ge 3$), it remains an open and actively researched problem in dimensions one and two.
This paper focuses on a specific aspect of this problem in one dimension ($d=1$). It addresses the question of the nature of the spectrum $\Lambda$ when $\Omega$ is a spectral set. Specifically, it investigates whether the spectrum $\Lambda$ must be "rational" (meaning all its elements are rational numbers) under certain natural conditions. The authors aim to prove that for a one-dimensional spectral set $\Omega$ with Lebesgue measure 1 and containing 0 in its spectrum, the spectrum $\Lambda$ must indeed be rational. This result has significant implications, as it connects the Fuglede conjecture on the real line to its equivalent on finite cyclic groups.
Methods
The paper employs a multi-pronged approach, building upon existing results and introducing novel connections.
- Weak Tiling Property: The proof is motivated by a "weak tiling" property of spectral sets, observed by Lev and Matolcsi. This property relates the auto-correlation measures of the spectrum to a positive measure $\gamma$. This measure $\gamma$ is shown to be translation-bounded and has specific properties related to the spectrum's support.
- Weighted Tiling Problem: The rationality problem is transformed into a "weighted tiling problem." This involves finding non-negative weights $v_n$ such that a weighted sum of indicator functions of the set $\Omega$ (shifted by integers) equals 1. The core of the problem then becomes proving that these weights $v_n$ must be periodic.
- Fourier Analytic Criterion: Fourier analysis is a central tool. The paper utilizes properties of the Fourier transform of measures and functions. Specifically, it leverages results that relate the support of the Fourier transform of a measure to the zero set of a function.
- Kronecker Theorem and Linear Independence: A crucial lemma (Lemma 4.1) is proven, which states that if a linear combination of sines and cosines with irrational frequencies is zero for all integers, then the coefficients must be zero. This lemma relies on the density property of irrational numbers under translation (Kronecker's theorem) and the linear independence of trigonometric functions.
- Structure of Spectra: The paper builds on the known result that spectra of one-dimensional spectral sets must be periodic (Theorem 3.1). This periodicity is then used to decompose the spectrum and weights, leading to the application of Lemma 4.1.
- Proposition 4.3 (from prior work): The authors reference a result by Wang and Zhou (Proposition 4.3) which establishes that the support of a certain measure (related to the weights) is a finite union of translates of a lattice, even without assuming the set is a spectral set. This proposition is instrumental in establishing the structure of the weights.
Datasets
Not discussed in the paper. The paper deals with theoretical mathematical objects (sets, measures, spectra) rather than empirical data.
Key Insights and Ideas
- Rationality of Spectra in 1D: The most significant insight is the definitive proof that for one-dimensional spectral sets (with unit measure and 0 in the spectrum), the spectrum must be rational. This resolves a long-standing conjecture in this specific dimension.
- Equivalence of Conjectures: The paper establishes a powerful equivalence: the Fuglede conjecture for spectral sets on $\mathbb{R}^1$ is equivalent to the Fuglede conjecture on all finite cyclic groups $\mathbb{Z}_n$. This means proving the conjecture for $\mathbb{R}^1$ automatically implies its truth for these discrete structures.
- Transformation to Weighted Tiling: The clever reduction of the rationality problem to a weighted tiling problem is a key methodological contribution. This rephrasing allows for the application of tools from Fourier analysis and number theory.
- The Role of Irrational Frequencies: Lemma 4.1 is a cornerstone, demonstrating that if a trigonometric polynomial with irrational frequencies vanishes for all integer inputs, its coefficients must be zero. This is a powerful tool for proving rationality.
- Structure of Weights: The paper shows that the weights derived from the spectral set have a specific periodic structure, which is crucial for applying Lemma 4.1.
Key Results
- Theorem 1.1 (Main Theorem): Let $\Omega \subset \mathbb{R}$ be a bounded spectral set with $|\Omega|=1$ and spectrum $\Lambda$ such that $0 \in \Lambda$. Then $\Lambda$ is rational.
- Theorem 1.2: This theorem formalizes the equivalence between the Fuglede conjecture on $\mathbb{R}^1$ and cyclic groups $\mathbb{Z}_n$.
- (1) Bounded tiles on $\mathbb{R}^1$ are spectral sets if and only if all tiles in $\mathbb{Z}_n$ are spectral sets in $\mathbb{Z}_n$ for all $n \ge 1$.
- (2) Bounded spectral sets on $\mathbb{R}^1$ are translational tiles if and only if all spectral sets in $\mathbb{Z}_n$ are tiles in $\mathbb{Z}_n$ for all $n \ge 1$.
Limitations and Future Work
- Dimension One Focus: The results are specifically for dimension one. While the paper mentions the Fuglede conjecture is open for $\mathbb{R}^2$, the methods used here are tailored to the one-dimensional case and may not directly extend.
- Equivalence to Cyclic Groups: The paper establishes the equivalence of the $\mathbb{R}^1$ conjecture to its $\mathbb{Z}_n$ counterparts. Future work could focus on further exploring and potentially resolving the conjecture on these cyclic groups, which has seen significant progress recently.
- Generalization of Methods: Investigating whether the techniques used to prove rationality in 1D can be adapted or extended to higher dimensions, perhaps under additional assumptions, would be a natural next step.
Summary updated