Dyadic Hypersingular Maximal Operator

• The dyadic hypersingular maximal operator is defined on the unit disc using dyadic arcs and Carleson boxes to capture endpoint estimates.
• It utilizes hypersingular sparse domination and dyadic modeling alongside Bourgain interpolation to establish strong, weak, and restricted weak-type L^p–L^q bounds.
• The framework extends to weighted settings and higher dimensions, providing a unified approach to analyze critical-line behavior in singular integral operators.

The dyadic hypersingular maximal operator is a fundamental object in the real-variable harmonic analysis of singular integral operators, particularly in regimes where standard strong-type $L^p\to L^q$ estimates break down at critical exponents. Recent work has constructed a complete $L^p$–$L^q$ mapping theory for this operator, leveraging a combination of dyadic modeling, hypersingular sparse domination, and Bourgain's interpolation method. These tools provide a precise framework for endpoint and critical-line estimates, notably in settings such as the unit disc, and extend to new classes of hypersingular sparse operators in higher dimensions (Hu et al., 31 Dec 2025).

1. Definition and Dyadic Construction

The dyadic hypersingular maximal operator, denoted $\mathcal{M}_t^{\mathcal{D}}$, is defined on the unit disc $\mathbb{D}$ for a hypersingular index $1 < t < 3/2$ as follows. One fixes a dyadic system $\mathcal{D}$ of arcs on the unit circle $\mathbb{T}$ (typically via a 2-adic subdivision with possible shifts). For each dyadic arc $I\in\mathcal{D}$, the associated Carleson box is

$$Q_I = \{ z \in \mathbb{D} : z/|z| \in I,\ 1-|I| \leq |z| < 1 \}.$$

The operator acts by

$$\mathcal{M}_t^{\mathcal{D}} f(z) = \sup_{I\in\mathcal{D},\ z\in Q_I} |Q_I|^{-t} \int_{Q_I} |f(w)|\,dA(w).$$

By the "1/3–trick," the non-dyadic maximal operator is pointwise comparable to a finite sum of dyadic versions, yielding equivalent $L^p$–$L^q$ bounds.

2. $L^p \to L^q$ Theory and Critical Line Behavior

The $L^p\to L^q$ mapping theory for $\mathcal{M}_t^{\mathcal{D}}$ is fully characterized by the fractional parameter $\alpha = 2t-2\in(0,1)$. The operator mimics the behavior of the fractional maximal operator of order $\alpha$, with the following regimes:

• Strong-Type Off the Critical Line: For $1/q-1/p>\alpha$ ($1\leq p,q\leq\infty$), the operator admits strong-type boundedness:

$$\|\mathcal{M}_t^{\mathcal{D}}f\|_{L^q(\mathbb{D})} \leq C\|f\|_{L^p(\mathbb{D})}.$$

• Weak-Type on the Critical Line: Along $1/q - 1/p = \alpha$, strong-type fails, but weak-type holds for $1<q\leq\infty$:

$$\mathcal{M}_t^{\mathcal{D}}: L^p(\mathbb{D})\to L^{q,\infty}(\mathbb{D}),$$

at $(p,q)$ satisfying $1/q-1/p=2t-2$.

• Restricted Weak-Type Endpoint: At the left endpoint $(p,q)=(1/(3-2t),1)$, only restricted weak-type bounds are possible.

This critical-line phenomenon is visualized as an admissible region in $(1/p,1/q)$, with strong-type below the line, weak-type on it, and failure beyond the left endpoint.

3. Weighted Endpoints and Radial Criteria

Endpoint behavior for $\mathcal{M}_t^{\mathcal{D}}$ admits a complete characterization in the setting of radial weights $\omega(z)=\omega(|z|)$. For boundary annuli $D_k = \{1-2^{-(k+1)} \leq |z|<1-2^{-k}\}$ and integrals

$$I_k = \int_{1-2^{-(k+1)}}^{1-2^{-k}} \omega(r)^{-\frac{3-2t}{2t-2}}\,dr,$$

the bounding conditions are:

• Weak (Restricted) Type: $$\mathcal{M}_t^{\mathcal{D}}: L^{1/(3-2t)}(\mathbb{D},\omega)\to L^{1,\infty}(\mathbb{D})$$ is bounded if and only if

$\sup_k 2^k I_k <\infty.$

• Strong Type (if $\omega$ in Bekollé–Bonami class): Boundedness of $$\mathcal{M}_t^{\mathcal{D}}: L^{1/(3-2t)}(\mathbb{D},\omega)\to L^1(\mathbb{D})$$ is equivalent to $\sum_{k\geq0} 2^k I_k <\infty$.

In the unweighted case, $I_k\approx 2^{-k}$ always holds, so the weak and strong endpoint criteria are automatically satisfied in the radial context.

4. Hypersingular Sparse Domination and Graded Sparse Families

A central methodological advance is the "hypersingular sparse domination" principle. For both the maximal and Bergman-type hypersingular operators, there exists a sparse collection $\mathcal{S}\subset\mathcal{D}$ of Carleson boxes such that

$$\mathcal{M}_t^{\mathcal{D}}f(z) \leq A^t_{\mathcal{S}}|f|(z):= \sum_{I\in\mathcal{S}} 1_{Q_I}(z) |Q_I|^{-t} \int_{Q_I}|f|.$$

Sparseness is quantified by the existence of disjoint subsets $E(I)\subset Q_I$ with $|E(I)|\geq\eta|Q_I|$. For Carleson-boxes in the disc, $\eta=1/2$ suffices.

The notion of graded sparse families introduces a structural parameter—the degree $K_{\mathcal{S}}$—encoding combinatorial generation depth, with $\mathcal{S}$ partitioned into layers where sidelengths decrease by at most $2^{K_{\mathcal{S}}}$ per step. In $\mathbb{D}$, this yields $(\eta, K_{\mathcal{S}})=(1/2,1)$. Extension to $\mathbb{R}^n$ leads to general graded sparse operators, governed by $(n, t, \eta, K_{\mathcal{S}})$.

For $1<t<1-(\log_2(1-\eta))/(nK)$, the strong/weak/restricted weak-type bounds for $A^t_{\mathcal{S}}$ are as follows:

Bound Type	Exponent Condition	Mapping Property
Strong-Type	$1/q-1/p>nK(t-1)/(-\log_2(1-\eta))$	$A^t_\mathcal{S}: L^p \to L^q$
Weak-Type (crit.)	$1/q-1/p=nK(t-1)/(-\log_2(1-\eta)),\ q>1$	$A^t_\mathcal{S}: L^p \to L^{q,\infty}$
Restricted Weak-Type	$p=-\log_2(1-\eta)/[-\log_2(1-\eta)+nK(1-t)],\ q=1$	$A^t_\mathcal{S}: L^{p,1}\to L^{1,\infty}$

In the Carleson box case, this recovers the disc critical line $1/q-1/p=2t-2$.

5. Bourgain Interpolation and Endpoint Analysis

Endpoint restricted weak-type estimates are established using a scale-sliced decomposition. Writing $A^t_\mathcal{S} = \sum_{j\geq0} A_j$ with operators $A_j$ corresponding to boxes of fixed sidelength, one observes differing $L^1\to L^1$ growth and $L^\infty\to L^1$ decay exponents: $$\|A_j\|_{L^1\to L^1} \lesssim 2^{(t-1)n j},\quad \|A_j\|_{L^\infty\to L^1} \lesssim 2^{-(3-2t)j} \quad (n=2\ \text{in}\ \mathbb{D}).$$ Bourgain's interpolation lemma, applied at

$\theta = \frac{3-2t}{(t-1)n + (3-2t)},$

produces precisely the restricted weak-type exponent $p=1/(3-2t),\ q=1$. Marcinkiewicz interpolation theory then yields all weak-type and strong-type results off and on the critical line.

6. Relationship to Hypersingular Bergman Projections and Broader Applications

The dyadic and sparse domination framework, including the Bourgain interpolation scheme, applies both to $\mathcal{M}_t^{\mathcal{D}}$ and to the positive Bergman-kernel operator

$$f\mapsto \int_{\mathbb{D}}|1-z\overline{w}|^{-2t}f(w)\,dA(w),$$

with critical line and endpoint behaviors matching those in the maximal operator case. This unified approach leads to strong, weak, and restricted weak-type mapping theories for other hypersingular operators, including those of Forelli–Rudin type and their analogs in real and complex spaces.

The new real-variable methodology directly addresses the need for effective analysis in the hypersingular regime $t>1$, as posed by Cheng–Fang–Wang–Yu, and extends the toolkit available for the study of critical-line and endpoint regularity in both maximal and integral operator settings (Hu et al., 31 Dec 2025).