The theory of one-dimensional spectral measures

Xin-Rong Dai, Shui Yu, Xing‐Gang HeScientia 2025 · DOI

TL;DR

This paper provides a comprehensive review of the theory of one-dimensional spectral measures, exploring the conditions under which a measure admits an orthonormal basis of exponential functions (a Fourier basis). It bridges harmonic analysis, fractal geometry, and tiling theory, summarizing decades of progress on famous puzzles like Fuglede’s Conjecture and the spectrality of Cantor-type fractal measures.

What Problem Does This Paper Address?

In classical Fourier analysis, we know that the space $L^2([0, 1])$ has a natural orthonormal basis of exponentials: $\{e^{2\pi i n x}\}_{n \in \mathbb{Z}}$. The core question this paper addresses is: For which other measures $\mu$ does the Hilbert space $L^2(\mu)$ possess an orthonormal basis of exponential functions $\{e^{2\pi i \lambda x}\}_{\lambda \in \Lambda}$?

If such a set $\Lambda$ (called a "spectrum") exists, the measure $\mu$ is called a spectral measure. This problem is deceptively simple but deeply complex. It asks whether we can perform "standard" harmonic analysis on non-standard spaces, such as fractal sets (like the Cantor set) or discrete measures. The challenge lies in the fact that the geometry of the support of the measure must "fit" the oscillations of the exponential functions perfectly to maintain orthogonality and completeness.

Methods

As a review paper, the authors synthesize several key mathematical frameworks used to tackle spectrality:

  • The Jorgensen-Pedersen Criterion: A fundamental tool using the Fourier transform of the measure to check for the existence of an orthogonal set of exponentials.
  • Iterated Function Systems (IFS): Used to define self-similar measures (like the Cantor measure) via scaling factors ($\rho$) and digit sets ($D$).
  • Tiling Theory: Exploring the link between a set's ability to tile space by translation and its ability to serve as a support for a spectral measure (Fuglede’s Conjecture).
  • Tree-Mapping and Symbolic Dynamics: A newer approach that represents spectra as paths through a tree, allowing researchers to "count" or "construct" spectra for fractal measures.
  • Ramsey Theory: Recently applied to study "Moran measures," which are more general than self-similar measures because their scaling factors can change at each step of the construction.

Datasets

Not discussed in the paper. As a theoretical mathematics review, this work relies on formal proofs and logical constructions rather than empirical datasets or benchmarks.

Key Insights and Ideas

  • The "1/4 vs. 1/3" Paradox: One of the most famous results in the field is that the standard middle-third Cantor measure is not spectral, but the middle-half (1/4) Cantor measure is. This highlights how sensitive spectrality is to the specific scaling factors and digit sets used to build a fractal.
  • Fuglede’s Conjecture: The paper discusses the long-standing hypothesis that a set is "spectral" if and only if it can "tile" the space. While disproven in higher dimensions ($n \ge 3$), the conjecture remains a major open problem in 1D and 2D.
  • Spectral Eigenvalues: A fascinating recent development is the study of "spectral eigenvalues"—scaling factors $t$ such that if $\Lambda$ is a spectrum for $\mu$, then $t\Lambda$ is also a spectrum. This reveals a hidden multiplicative structure within fractal spectra.
  • Beurling Dimension: The authors highlight that the "density" of a spectrum (measured by Beurling dimension) is often tied to the Hausdorff dimension of the measure's support, suggesting a deep link between the measure's geometry and its harmonic properties.
  • Convergence Anomalies: Fourier series on fractal measures behave differently than on the real line. For some fractal measures, the Fourier series of a continuous function might diverge at certain points, a phenomenon that doesn't happen in the same way with the classical Lebesgue measure.

Key Results

  • Classification of Self-Similar Measures: The paper summarizes the "Laba-Wang Conjecture," which aims to categorize all self-similar spectral measures. Significant progress has been made for small digit sets (e.g., $\#D = 2, 3, 4$).
  • Moran Measure Progress: The authors present recent results showing that even if the scaling factors vary, one can still find "infinite orthogonal sets," though proving they form a complete basis remains difficult.
  • Strichartz's Legacy: The paper details solutions to problems posed by Robert Strichartz regarding the convergence of "mock" Fourier series on Cantor sets, proving that under certain conditions, these series converge almost everywhere.

Limitations and Future Work

  • The General Laba-Wang Conjecture: While many cases are solved, a universal proof for all self-similar measures remains elusive.
  • Higher Dimensional 1D Measures: Most results focus on measures supported on the real line. Extending these to "one-dimensional" curves in higher-dimensional space is an active frontier.
  • Riesz Bases vs. Orthonormal Bases: The paper notes that many measures that aren't "spectral" (don't have an orthonormal basis) might still have a "Riesz basis" (a basis that is "almost" orthogonal). Characterizing which measures admit Riesz bases is a major open challenge (the Strichartz Problem).
  • Applications: While the theory is robust, the authors suggest that the application of these fractal Fourier bases in fields like signal processing or "fractal uncertainty principles" is still in its infancy.

Summary updated

BibTeX
@inproceedings{dai2025theory,
  author = {Xin-Rong Dai and Shui Yu and Xing‐Gang He},
  title = {{The theory of one-dimensional spectral measures}},
  booktitle = {Scientia},
  year = {2025},
  doi = {10.1360/ssm-2025-0083},
}
Figure