Boundary estimates for the fractional spherical maximal function

Abstract: In this article, we study the fractional spherical maximal function and its lacunary counterpart. We study the necessary and sufficient conditions for $L<sup>p-L<sup>q$ boundedness of both maximal functions. In particular, we prove the restricted weak type estimate for both full and lacunary fractional spherical maximal functions at the boundary of the maximal $L<sup>p-L<sup>q$ bounded regions.

• The paper establishes restricted weak-type L^p-L^q estimates on the boundary of the boundedness regions for both full and lacunary fractional spherical maximal operators.
• It employs a synthesis of interpolation techniques, Littlewood-Paley decomposition, and precise kernel estimates to achieve the results.
• Key counterexamples demonstrate the sharp failure of strong-type bounds on certain boundary segments, highlighting the unique role of curvature.

Boundary Estimates for the Fractional Spherical Maximal Function

Introduction

The study of maximal functions associated with geometric averages remains central in harmonic analysis. This paper addresses the $L^p-L^q$ boundedness properties of the fractional spherical maximal function and its lacunary analogue—a question with fundamental connections to differentiation theory, partial differential equations, and oscillatory integrals. Special attention is given to restricted weak-type estimates on the boundary of boundedness regions, filling a key gap in the understanding initiated by works of Stein, Bourgain, Oberlin, and subsequent advancements. The results are established via an overview of harmonic analysis techniques, including complex interpolation and Bourgain’s interpolation principle.

Definitions and Background

Given $f \in \mathcal{S}(\mathbb{R}^n)$, the fractional spherical average at radius $t > 0$ with order $\alpha$ is defined by

$$A_t^\alpha f(x) = t^\alpha \int_{S^{n-1}} f(x - t y) \, d\sigma(y) = t^{\alpha} (f * \sigma_t)(x),$$

where $d\sigma$ is the normalized surface measure. The associated fractional spherical maximal operator is

$$A_{*}^\alpha f(x) = \sup_{t > 0} |A_t^\alpha f(x)|.$$

The lacunary analogue, restricting to dyadic dilates, is

$$A_{lac}^\alpha f(x) = \sup_{k \in \mathbb{Z}} |A_{2^k}^\alpha f(x)|.$$

For $\alpha = 0$, landmark works of Stein (for $n \geq 3$) and Bourgain ($f \in \mathcal{S}(\mathbb{R}^n)$0) identify the sharp range for strong-type $f \in \mathcal{S}(\mathbb{R}^n)$1-boundedness of $f \in \mathcal{S}(\mathbb{R}^n)$2. Oberlin extended the problem to $f \in \mathcal{S}(\mathbb{R}^n)$3 and characterized the region of admissible exponents via the point $f \in \mathcal{S}(\mathbb{R}^n)$4 lying in explicit convex regions (see Figure 1 in the paper).

Main Results

1. Boundary Restricted Weak-Type Estimates

The paper establishes the restricted weak-type $f \in \mathcal{S}(\mathbb{R}^n)$5 estimates for the fractional spherical maximal operator and its lacunary counterpart at the boundary of the known strong-type region, that were previously unsettled:

• For $f \in \mathcal{S}(\mathbb{R}^n)$6 and $f \in \mathcal{S}(\mathbb{R}^n)$7: If $f \in \mathcal{S}(\mathbb{R}^n)$8 is on the open segments $f \in \mathcal{S}(\mathbb{R}^n)$9, $t > 0$0, $t > 0$1 of the admissible region, with $t > 0$2, then

$t > 0$3

holds (Theorem 1).

• For $t > 0$4 and $t > 0$5: An analogous result holds on the segments $t > 0$6 and $t > 0$7.
• For the lacunary maximal operator: The same pattern holds on the corresponding boundaries, with explicit restricted weak-type estimates proved up to endpoints not ruled out by known counterexamples.

Notably, strong-type $t > 0$8 estimates fail on certain boundary segments (specifically, on the segment $t > 0$9 for the lacunary case), as is established via explicit counterexamples.

2. Characterization of the $\alpha$0 Bounded Regions

The range of exponents for which the $\alpha$1 estimates for the full and lacunary fractional maximal operators are available is completely characterized. For the fractional operator, boundedness holds in the interior of the convex region depicted in the figures in the paper and is extended to the boundary in restricted weak-type form except where sharp counterexamples preclude any improvement.

In the case of the classical fractional Hardy-Littlewood maximal function $\alpha$2, the boundedness region is substantially larger; the sharp distinctions in the boundedness range for the spherical case highlight the crucial influence of curvature and oscillation.

Methodology

The proof structure leverages several advanced tools:

• Littlewood-Paley Decomposition: Both maximal operators are decomposed into frequency-localized pieces through standard dyadic partition of unity. This enables the isolation of contributions at respective scales.
• Kernel Estimates and Oscillation Control: Careful pointwise bounds on the convolution kernels associated to the spherical averages are established, controlling both size and decay.
• Bourgain-Type Summability and Interpolation: Bourgain's interpolation principle is systematically applied to interpolate between estimates with different scale dependence, yielding the restricted weak-type bounds.
• Endpoint Analysis and Counterexamples: The paper verifies non-boundedness exactly at certain endpoints using explicit test functions, demonstrating that the strong-type result is unattainable and that restricted weak-type is optimal.
• Complex Interpolation: Endpoint estimates are bootstrapped to interior points via analytic family methods, including interpolation of restricted weak-type estimates as formalized by Sagher.

Numerical Estimates and Sharpness

The paper provides explicit decay rates for the operator norms of localized maximal pieces as functions of the frequency scale $\alpha$3. For example, for $\alpha$4 and $\alpha$5,

$\alpha$6

with $\alpha$7 precisely identified via kernel regularity and curvature computations. The sharpness of exponents is established, and the transition from strong-type to restricted weak-type bounds is explicitly described.

The lacunary case also receives a thorough analysis, with fine control of norm estimates as $\alpha$8 and the explicit failure of strong-type bounds on the segment $\alpha$9 is demonstrated via a covering argument and geometric measure computation.

Implications and Future Directions

This work closes a major chapter in the search for endpoint regularity and boundedness of (fractional) spherical maximal operators. The precise restricted weak-type estimates at the boundaries of the strong-type region are now established, with clear identification of failures and remaining open problems—such as the possibility of true weak-type boundedness strictly on the boundary and its implications for the boundedness of related local maximal functions, which are linked to deep questions about concentration and regularization for averages over curved hypersurfaces.

Practically, these estimates control the regularizing effect of spherical averaging along scales—a tool that is critical both for the analysis of PDEs with radial symmetry and for fine properties of Banach function spaces. The explicit counterexamples and norm-gain calculations may also inform the search for weighted norm inequalities and anisotropic generalizations in higher dimensions.

Theoretically, the comparative analysis with the (fractional) Hardy-Littlewood maximal function exposes essential differences arising from surface curvature—a demarcation point for endpoint phenomena in harmonic analysis.

Potential future directions include:

• Extending the endpoint theory to maximal operators associated to more general curved averages or submanifolds.
• Investigating the analogue of these results in the setting of metric measure spaces with non-Euclidean geometry.
• Further exploration of the weak-type endpoint, in particular, its precise status for the local spherical maximal function.

Conclusion

The results of this paper deliver a comprehensive solution to the question of $$A_t^\alpha f(x) = t^\alpha \int_{S^{n-1}} f(x - t y) \, d\sigma(y) = t^{\alpha} (f * \sigma_t)(x),$$0 boundedness—including strong, weak, and restricted weak-type—of the fractional spherical maximal function and its lacunary analogue at the boundary of the maximal region. Through a sophisticated synthesis of interpolation methods and fine-scale harmonic analysis, the authors identify the sharp transition to restricted weak-type at the endpoints, thus advancing both the technical understanding and the proper context for further developments in analysis.