Singular Fractal Measures

• Singular fractal measures are probability measures supported on Lebesgue-null fractal sets that exhibit non-integer dimensions and self-similar structures.
• They are constructed via iterated function systems and digit-restriction methods, leading to explicit scaling laws and dimension formulas.
• Their analysis employs multifractal formalism, harmonic analysis, and advanced numerical schemes to uncover intricate geometric, spectral, and dynamical properties.

A singular fractal measure is a probability or finite Borel measure supported on a set of Lebesgue measure zero, typically a fractal, that fails to be absolutely continuous with respect to the ambient Lebesgue measure. These measures encode complex geometric, spectral, and probabilistic properties driven by the often self-similar, non-integer dimensional nature of their support. Singular fractal measures arise naturally across harmonic analysis, dynamical systems, geometric measure theory, probability, and statistical physics—exemplified by classical constructions (e.g. Cantor–type, self-similar Moran sets), Riesz products (e.g. Thue–Morse), Salem-type singular distributions, and modern stochastic fractals (e.g. SLE/CLE pivotal and carpet/gasket measures).

1. Constructive Definitions and Algebraic Frameworks

A key paradigm for singular fractal measures is the class of self-similar or affine invariant measures under iterated function systems (IFS). Given contractive maps $s_m(x)=\rho_m A_m x+\delta_m$ for $m=1,\dots,M$, a compact attractor $\Gamma$ is defined by Hutchinson’s equation $\Gamma = \bigcup_{m=1}^M s_m(\Gamma)$. Assigning probability weights $\{p_m\}$, one constructs a self-similar measure $\mu$ by the fixed-point equation $\mu(E) = \sum_{m=1}^M p_m\,\mu(s_m^{-1}(E))$ for Borel sets $E$ (Gibbs et al., 2023). In the case that the Hausdorff dimension $d< n$ for $\Gamma\subset \mathbb{R}^n$, the measure $\mu$ becomes singular.

Variants include measures induced by digit-restriction expansions (e.g. binary, ternary, or nega-$P$ representations) combined with singular Salem-type distribution functions such as $F_s(x)$ and $G_s(x)$ (Serbenyuk, 2020). The support of such measures (e.g., generalized Cantor or alternating Moran sets) yields Hausdorff dimension $\alpha$ via solutions to $\sum_i p_i^\alpha = 1$ or corresponding infinite-product Moran equations.

Measures arising from random fractals—such as SLE/CLE pivotal, cut-point, or gasket measures—are constructed through scaling limits of discrete models, quantum natural parametrizations, and dequantization procedures via Liouville quantum gravity (LQG) (Cai et al., 2022). These are singular and conformally covariant.

2. Geometric Properties, Dimension, and Content

The intrinsic “fractal dimension” of the support is fundamental. For classic self-similar sets under the open set condition (OSC), the Hausdorff dimension $d$ is determined by the similarity ratios via $\sum_{m=1}^M \rho_m^d = 1$ (Gibbs et al., 2023). For singular continuous measures associated to Cantor-like sets, the dimension is $d = \log M/\log N$, e.g., $d = \log2/\log3$ for the middle-third Cantor set.

Generalized Minkowski dimension and content provide refined local scaling descriptions. For a Borel measure $\mu$ and Cantor-type $K\subset \chi$, the $\epsilon$-neighborhood $N_\epsilon(K)$ scales as $\mu(N_\epsilon(K))\sim A_M \epsilon^{d_M}$, yielding a Minkowski codimension $d_M$ and a content $A_M$ if the corresponding limits exist. For extremely rare sets, the scaling is non-standard, e.g., $\mu(N_\epsilon(K))\sim B\,\exp(-D \epsilon^{-q})$, and the corresponding constants enter extreme value statistics (Mantica et al., 2015).

In random fractal contexts (SLE/CLE), the Hausdorff dimension is determined by model parameters (e.g., $d= 3-3\kappa'/8$ for SLE cut points), and the natural measures are singular but σ-finite, supported on sets of non-integer dimension (Cai et al., 2022). Dimension formulae for alternating or non-homogeneous singular distributions require sophisticated infinite-product equations and Moran theory (Serbenyuk, 2020).

3. Harmonic and Spectral Analysis

Singular fractal measures possess unique harmonic properties, often reflected in their Fourier analytic behavior. The dichotomy of Hutchinson and Jessen–Wintner establishes that invariant self-similar measures are either purely singular or absolutely continuous (Jorgensen et al., 2007). For Cantor-like measures, the Fourier transform decays at infinity at a rate governed by the dimension, $|\widehat\mu(\xi)| = O(|\xi|^{-d})$ (Jorgensen et al., 2007). In contrast, for Pisot-type parameters, the Fourier transform remains bounded below along prescribed sequences, indicating singularity and “chaos” under perturbation.

Spectral theory distinguishes spectral measures—admitting orthonormal exponential bases—from non-spectral, frame-spectral measures. Non-spectral singular fractal measures can nonetheless admit Fourier frames (almost-tight decompositions), as demonstrated by almost-Parseval-frame tower constructions (Lai et al., 2015). In higher dimensions, slice-singular, fibered singular measures support non-orthogonal 2D Fourier expansions, with explicit duality and algorithmic coefficient calculability—extending harmonic analysis to planar singular sets, such as Bedford-McMullen carpets and Sierpinski carpets (Berner et al., 2024).

Moreover, Kuznecov-type formulae precisely link the asymptotics of Laplace eigenfunction sums to the Ahlfors regularity and averaged density content of fractal measures. For $\mu$ $s$-Ahlfors regular on a manifold $M$ of dimension $n$, $$N_\mu(\lambda)\sim (2\pi)^{-(n-s)}\,\text{vol}\,(B^{n-s})\,A_\mu\,\lambda^{n-s}$$ as $\lambda\to\infty$, recovering the dimension and total mass, with sharp remainder estimates (Xi, 20 Dec 2025).

4. Scaling, Multifractality, and Thermodynamic Formalism

Singular fractal measures frequently exhibit multifractal scaling properties, captured by the local dimension spectrum and thermodynamic formalism. For the Thue-Morse measure—arising as an infinite Riesz product—the local dimension $d_\mu(x)$ satisfies $d_\mu(x)<1$ almost everywhere, and the measure is purely singular continuous (Baake et al., 2018). The multifractal spectrum $f(\alpha)$ is computed via Legendre transforms of the pressure function $p(q)$ associated to unbounded potentials, with technical challenges resolved by subshift truncations and exhaustion principles.

These multifractal spectra quantify the Hausdorff dimensions of level sets where local dimension equals $\alpha$ and reveal deep connections between ergodic theory, statistical mechanics, and geometric measure theory. Random fractal measures (SLE/CLE) also display similar multifractal and scaling behaviors, with exponents matched to KPZ scaling relations via LQG coupling (Cai et al., 2022).

5. Numerical Methods and Integrals over Singular Sets

Computational analysis of singular integrals with respect to pairs of self-similar measures on non-disjoint fractal sets necessitates advanced numerical schemes. The key is to utilize self-similar decomposition, splitting singular bilinear forms into sums of fundamental regular and singular sub-integrals, configuring a finite linear system for practical computation. Regular sub-integrals admit high-accuracy quadrature (composite barycentre, chaos-game Monte Carlo, or Gauss methods), while singular sub-integrals are managed by recursive subdivision, with formulae established for classic fractals such as the Sierpinski triangle and Koch snowflake (Gibbs et al., 2023).

The practical stability of these schemes, error bounds, and convergence rates depend on the spectral properties and scaling structure of the measures involved, with Ahlfors regularity or open-set condition ensuring tractability.

6. Special Constructions and Applications

Salem-type singular measures constructed via digit-restriction expansions and their singular distribution functions $F_s$, $G_s$ generalize classic Cantor measures and introduce complex new fractal supports carrying singular measures of arbitrary (possibly non-homogeneous) dimension. These constructions offer flexibility for multifractal analysis, Diophantine approximation, and functional transformations via pushforwards, with the dimension determined by explicit algebraic or infinite-product equations (Serbenyuk, 2020).

Random fractal measures arising in statistical mechanics and probability (e.g., SLE/CLE pivotal/cutpoint/carpet/gasket measures) play critical roles in encoding critical phenomena, scaling limits, and conformal invariance. Their robust construction via coupling to LQG provides existence, uniqueness, and concrete identification with discrete models, such as percolation cluster area laws (Cai et al., 2022).

7. Dynamical and Extreme Value Structures

In dynamical systems, intensity observables with uncountable sets of singularities (e.g., fractal landscapes) induce stationary processes whose extremal statistics depend explicitly on the fractal geometry of the singular set. Generalized Minkowski dimension, content, and non-standard Minkowski constants precisely control the scaling of exceedance probabilities and the limit laws for maxima. Both standard (power-law) and non-standard (double-exponential) scaling regimes admit rigorous extreme value laws, verified by numerical experiments across singular continuous measures and Lebesgue null fractal sets (Mantica et al., 2015).

This reveals the direct impact of fractal geometry and measure singularity on rare event statistics, with significant implications for ergodic theory, statistical physics, and probability.

---

Singular fractal measures form a central object of study interlinking harmonic analysis, spectral theory, dynamical systems, geometric measure theory, and statistical physics. Their construction, geometric properties, spectral behavior, multifractal structure, and dynamical implications have been elucidated through rigorous analytic, probabilistic, and computational frameworks, drawing connections across classic self-similar sets, Salem and Riesz products, modern planar and random fractals, and energy-based spectral asymptotics. These objects continue to motivate deep inquiries into the foundations of measure, dimension, functional analysis, and stochastic processes.