On the Fourier dimension of fractional Brownian graphs

Basic Information

• Paper ID: 2510.09818
• Title: On the Fourier dimension of fractional Brownian graphs
• Authors: Cheuk Yin Lee, Samy Tindel
• Classification: math.PR (Probability Theory)
• Publication Date: October 10, 2025 (arXiv preprint)
• Paper Link: https://arxiv.org/abs/2510.09818

Abstract

This paper proves that the Fourier dimension of the graph $G(B)$ of fractional Brownian motion $B$ with Hurst parameter $H \in (0,1/2)$ equals 1 almost surely. This completely resolves a conjecture posed by Fraser and Sahlsten, while providing an exact formula for the gap between the Hausdorff dimension and Fourier dimension of $G(B)$: $\dim_{\rm H}(G(B)) - \dim_{\rm F}(G(B))$. The proof is based on intricate combinatorial procedures involving multiple integrals related to the covariance function of fractional Brownian motion.

Research Background and Motivation

Problem Background

1. Significance of Fourier Dimension: The Fourier dimension is a fundamental concept in geometric measure theory that characterizes the geometric structure of sets through the decay properties of the Fourier transform of measures. For a set $A \subset \mathbb{R}^n$, the Fourier dimension is defined as: $\dim_F A = \sup\{\beta \in [0,n] : \exists \mu \in P(A), |\hat{\mu}(\xi)| \lesssim |\xi|^{-\beta/2}\}$
2. Concept of Salem Sets: When a set's Hausdorff dimension equals its Fourier dimension, the set is called a Salem set. This is an important geometric property, but most graphs of random processes are not Salem sets.
3. Fraser-Sahlsten Conjecture: Fraser and Sahlsten proved in 2018 that the Fourier dimension of the graph of standard Brownian motion equals 1 almost surely, and conjectured that the Fourier dimension of fractional Brownian motion graphs also equals 1.

Research Motivation

1. Theoretical Completeness: This conjecture is an important open problem in the study of geometric properties of fractional Brownian motion
2. Precise Characterization of Dimension Gap: Once the Fourier dimension is determined, one can precisely compute the gap between Hausdorff and Fourier dimensions
3. Technical Challenges: The case $H < 1/2$ is technically more challenging, requiring handling of more complex singularities

Core Contributions

1. Complete Resolution of Fraser-Sahlsten Conjecture: Proves that for all $H \in (0,1)$, the Fourier dimension of the fractional Brownian motion graph $G(B)$ equals 1 almost surely
2. Exact Formula for Dimension Gap: $\dim_H(G(B)) - \dim_F(G(B)) = (2-H) - 1 = 1-H$
3. Development of New Technical Framework: Establishes a complex combinatorial integral method for handling the case $H < 1/2$
4. Completion of Theoretical Picture: Combined with known results for $H \geq 1/2$, provides a complete theory of the Fourier dimension of fractional Brownian motion graphs

Methodology Details

Problem Setup

Prove that for fractional Brownian motion $B = \{B_t, t \geq 0\}$ with Hurst parameter $H \in (0,1/2)$, its graph $G(B) = \{(t, B_t) : t \in [0,1]\}$ has Fourier dimension equal to 1 almost surely.

Core Technical Framework

1. Fourier Transform of Graph Measures

Define the graph measure $\mu_G$: $\mu_G(dt, dx) = \mathbf{1}_{[0,1]}(t) \delta_x(B_t) dt dx$

Its Fourier transform is: $\hat{\mu}_G(\xi) = \int_0^1 e^{-2\pi i(\xi_1 t + \xi_2 B_t)} dt$

2. Reduction via Moment Estimates

Through Proposition 2.2, lower bounds for Fourier dimension can be obtained via the following moment estimates:

• Vertical Bound: $E[|\hat{\nu}(\xi_2)|^{2q}] \leq C|\xi_2|^{-\gamma_2 q}$
• Horizontal Bound: $E[|\hat{\mu}_G(\xi_1, \xi_2)|^{2q}] \leq C|\xi_1|^{-\gamma_1 q}$

The goal is to prove $\gamma_1 = 1$.

3. Combinatorial Decomposition

Using Kahane decomposition and scaling properties of fractional Brownian motion, the key formula is obtained: $E[|\hat{\mu}_G(\xi_1, \xi_2)|^{2q}] = \frac{(q!)^2}{|\xi_2|^{2q/H}} \sum_{\varepsilon \in A_{2q}} I_T^\lambda[\varepsilon, G_\varepsilon]$

where:

• $\lambda = \xi_1/|\xi_2|^{1/H}$, $T = |\xi_2|^{1/H}$
• $A_{2q} = \{\varepsilon \in \{-1,1\}^{2q} : \sum_{j=1}^{2q} \varepsilon_j = 0\}$
• $G_\varepsilon(u) = e^{-\pi \text{Var}(\sum_{i=1}^{2q} \varepsilon_i B_{u_i})}$

Technical Innovations

1. Integration by Parts on Simplices

Develops systematic integration by parts formulas (Lemma 2.6): $I[\varepsilon, G_\varepsilon] = \left(\frac{1}{2\pi i\lambda}\right)^q \prod_{j=1}^q \frac{1}{\varepsilon_{2j-1}} \sum_{\sigma \in \Sigma(q)} I[\sigma; \varepsilon, G_\varepsilon]$

2. Singularity Analysis

For $H < 1/2$, the derivatives of the variance function exhibit singularities:

• First derivative: $|\partial_{s_i} g_a(s)| \lesssim \|a\|_\infty^2 I[(s_i - s_{i-1})^{2H-1} + (s_{i+1} - s_i)^{2H-1}]$
• Second derivative: $|\partial_{s_i}\partial_{s_j} g_a(s)| \lesssim \|a\|_\infty^2 (s_j - s_i)^{2H-2}$

3. Recursive Integration Strategy

Systematically handles exponential decay and singularities through stepwise variable integration: $U(I, P, J^*, \theta) \leq C^q T^{I-\ell}$

Experimental Setup

This is a pure theoretical mathematics paper with no numerical experiments. The proof is entirely based on rigorous mathematical analysis.

Experimental Results

Main Theoretical Results

Theorem 1.1: For fractional Brownian motion with Hurst parameter $H \in (0,1)$, the Fourier dimension of its graph $G(B)$ equals 1 almost surely.

Corollary: The exact formula for the dimension gap is: $\dim_H(G(B)) - \dim_F(G(B)) = 1 - H$

Technical Achievements

Proves the key moment estimate (inequality (80)): $E[|\hat{\mu}_G(\xi_1, \xi_2)|^{2q}] \leq C_q |\xi_1|^{-q}$

This provides the required horizontal bound, which combined with known vertical bounds, completes the proof of the Fourier dimension.

Related Work

Historical Development

1. Kahane (1985): Established fundamental methods for computing Fourier dimension via moment estimates
2. Fraser-Sahlsten (2018): Proved the case for standard Brownian motion graphs and proposed the conjecture for fractional Brownian motion
3. Lai-Lee (2025): Resolved the case $H \geq 1/2$
4. This Paper: Completes the most difficult case $H < 1/2$

Technical Comparison

• $H \geq 1/2$: Singularities in derivatives of variance functions are milder, techniques are relatively straightforward
• $H < 1/2$: Requires handling strong singularities of type $(s_j - s_i)^{2H-2}$, necessitating refined combinatorial analysis

Conclusions and Discussion

Main Conclusions

1. Completely resolves the Fraser-Sahlsten conjecture, proving that the Fourier dimension of fractional Brownian motion graphs is always 1
2. Establishes new technical frameworks for handling strongly singular integrals
3. Provides exact formulas for the gap between Hausdorff and Fourier dimensions

Technical Significance

1. Methodological Contribution: The developed combinatorial integral techniques may be applicable to other random processes
2. Theoretical Completeness: Completes the theoretical picture of geometric properties of fractional Brownian motion

Future Directions

1. Extend techniques to multidimensional fractional Brownian motion
2. Study Fourier dimensions of graphs of other Gaussian processes
3. Explore applications of Fourier dimension in random geometry

In-Depth Evaluation

Strengths

1. Theoretical Importance: Resolves an important open problem in the field
2. Technical Innovation: Develops new methods for handling strongly singular integrals
3. Rigorous Proof: Mathematical arguments are complete and technically refined
4. Complete Results: Provides unified results for all values of $H$

Technical Challenges

1. Complexity: The proof involves extremely intricate combinatorial calculations
2. Technical Threshold: Requires deep background in harmonic analysis and stochastic process theory
3. Computational Volume: Estimation of multiple integrals requires handling numerous case classifications

Impact Assessment

1. Academic Value: Completes an important theoretical conjecture with milestone significance
2. Technical Impact: Developed methods may inspire research on related problems
3. Theoretical Completeness: Provides a complete picture for the geometric theory of fractional Brownian motion

Application Scenarios

This research is primarily of theoretical value, contributing to:

1. Development of foundations in random geometry theory
2. New perspectives on applications of Fourier analysis in probability theory
3. Complete framework for studying geometric properties of fractional Brownian motion

References

The paper cites 21 important references, primarily including:

• Fraser & Sahlsten (2018): Proposed the original conjecture
• Lai & Lee (2025): Resolved the case $H \geq 1/2$
• Kahane (1985): Foundational theory of Fourier dimension
• Adler (1977): Classical results on Hausdorff dimension of fractional Brownian motion

Overall Assessment: This is a high-quality theoretical mathematics paper that completely resolves an important open problem in the field. Although technically extremely complex, the proof is rigorous and complete, with significant theoretical value and methodological contributions.