Hypersingular Cousin of Sparse Operators

• The paper unifies classical, fractional, and hypersingular integrals by extending sparse domination techniques with graded dyadic frameworks.
• It employs pointwise domination and dyadic averaging to control rough integral operators, establishing strong, weak, and endpoint mapping properties.
• Sharp Sobolev and weighted inequalities are derived using quantitative sparse bounds characterized by sparseness η and degree K_S in ℝⁿ.

A hypersingular cousin of sparse operators refers to a class of dyadic averaging operators and pointwise domination principles that generalize sparse bounds for classical Calderón–Zygmund and fractional integrals to hypersingular regimes, notably those more singular than the usual CZ case. These hypersingular operators arise naturally when studying rough integral operators $T_{\Omega, \alpha}$ with homogeneous kernels, where $0 < \alpha < n$, and more generally in models built from graded sparse families in $\mathbb R^n$. The sparsity notion is further refined by introducing geometric parameters such as sparseness $\eta$ and degree $K_{\mathcal S}$, which control admissible mapping properties on Lebesgue spaces. Hypersingular sparse operators unify the real-variable analysis of classical, fractional, and hypersingular singular integrals, with special attention to critical-line and endpoint mapping regimes (Hoang et al., 2024, Hu et al., 31 Dec 2025).

1. Rough Integral Operators and Hypersingular Regimes

Let $T_{\Omega, \alpha}$ denote the family of rough hypersingular operators defined by

$$T_{\Omega, \alpha} f(x) = \mathrm{p.v.} \int_{\mathbb R^n} \frac{\Omega(y/|y|)}{|y|^{n+1-\alpha}} f(x-y)\, dy, \quad 0 < \alpha < n,$$

where $\Omega$ is a measurable function on $S^{n-1}$ with $$\int_{S^{n-1}} \Omega(\theta) \, d\sigma(\theta) = 0$$. The principal value (p.v.) regularizes the singularity at $y=0$. Key cases include:

• $\alpha = 0$: classical rough singular integral.
• $0 < \alpha < 1$: rough hypersingular integral, more singular than CZ type.
• $1 < \alpha < n$: rough fractional integral.

The analysis requires assumptions on $\Omega$'s integrability:

• Critical: $\Omega\in L^{n,\infty}(S^{n-1})$,
• Subcritical: $1 < r < n$, $\Omega \in L^r(S^{n-1})$ (or Lorentz-refined $L^{r,r_*}(S^{n-1})$),
• Endpoint for $0 < \alpha < 1$: $\Omega\in L^1(S^{n-1};\log L)^\gamma, \gamma>1$.

2. Sparse Operators and Graded Sparse Families

A dyadic grid $\mathscr D$ in $\mathbb R^n$ comprises cubes whose side lengths are powers of two. A subcollection $\mathcal S \subset \mathscr D$ is $\eta$-sparse if for each $Q \in \mathcal S$, there exists a disjoint measurable set $E(Q) \subset Q$ with $|E(Q)| \geq \eta |Q|$. In the graded context, the degree $K_{\mathcal S}$ is defined so that

$$K_{\mathcal S} := \sup_{j \geq 0} \log_2 \frac{\mathfrak G_j}{\mathfrak G_{j+1}} < \infty,$$

where $\mathfrak G_j$ is the minimal sidelength in the $j$-th layer.

The associated hypersingular sparse operator is

$$A^{t}_{\mathcal S} f(x) := \sum_{Q \in \mathcal S} \frac{\mathbf{1}_Q(x)}{|Q|^t} \int_Q |f(y)| dy,$$

or equivalently

$$A^t_{\mathcal S} f(x) = \sum_{Q \in \mathcal S} \frac{\mathbf{1}_Q(x)}{|Q|^{t-1}} \langle |f| \rangle_Q .$$

3. Pointwise Domination and Sparse Potentials

For $T_{\Omega,\alpha}$ acting on compactly supported smooth $f \geq 0$, there exist finitely many sparse families $\mathcal S_1, ..., \mathcal S_N$ such that

$$|T_{\Omega, \alpha} f(x)| \leq C_{n, \alpha} \|\Omega\|_{L^r(S^{n-1})} \sum_{k=1}^N \mathcal S_{\alpha, r}(|\nabla f|)(x),$$

where the Riesz-potential-type sparse operator

$$\mathcal S_{\alpha, r} f(x) = \sum_{Q \in \mathcal S} |Q|^{\alpha/n} \left( \frac{1}{|Q|} \int_Q |f(y)|^r \, dy \right)^{1/r} \chi_Q(x).$$

This control principle extends to hypersingular regimes ($0 < \alpha < 1$) and reflects the deeper singularity by the appearance of the gradient $|\nabla f|$ rather than $f$ itself. The sparse domination mechanism relies on mean-zero cancellation, local Poincaré–Sobolev inequalities, and dyadic grid selection (Hoang et al., 2024).

4. $L^p \to L^q$ Mapping Properties and Critical Lines

Mapping properties for hypersingular sparse operators depend quantitatively on $\eta$ and $K_{\mathcal S}$. Let

$$\sigma := \frac{n K_{\mathcal S} (t-1)}{ -\log_2 (1-\eta) }.$$

For $A^{t}_{\mathcal S}$ built from graded $\eta$-sparse families:

• Strong-type: If $\frac{1}{q} - \frac{1}{p} > \sigma$, then $$\|A^t_{\mathcal S} f\|_{L^q} \lesssim_{n,t,\eta,K_{\mathcal S}} \|f\|_{L^p}$$.
• Weak-type: If $\frac{1}{q} - \frac{1}{p} = \sigma$, $1<q\leq\infty$, then $$\|A^t_{\mathcal S} f\|_{L^{q,\infty}} \lesssim_{n,t,\eta,K_{\mathcal S}} \|f\|_{L^p}$$.
• Restricted weak-type at $q=1$: For $$p_0 = \frac{-\log_2(1-\eta)}{ -\log_2(1-\eta) + n K_{\mathcal S}(1-t) }$$, $$\|A^t_{\mathcal S} f\|_{L^{1,\infty}} \lesssim \|f\|_{L^{p_0,1}}$$.

For classical dyadic Carleson boxes $(K_{\mathcal S}=1, \eta=1/2)$, $\sigma=2t-2$, concordant with the mapping theory for the hypersingular Bergman-type operator $K_{2t}$ (Hu et al., 31 Dec 2025).

5. Sobolev and Weighted Inequalities

Sparse domination by $\mathcal S_{\alpha, r}$ enables derivation of sharp Sobolev inequalities for $T_{\Omega, \alpha}$. If $1 < p < n/\alpha$ and $1/q = 1/p - \alpha/n$,

$$T_{\Omega, \alpha}: \dot W^{1,p}(\mathbb R^n) \longrightarrow L^q(\mathbb R^n), \qquad \|T_{\Omega, \alpha} f\|_{L^q} \lesssim \|\Omega\|_{L^r} \|\nabla f\|_{L^p}.$$

At $p=1$, the endpoint weak-type result is

$$T_{\Omega, \alpha}: \dot W^{1,1}(\mathbb R^n) \longrightarrow L^{n/(n-\alpha),\infty}(\mathbb R^n).$$

Hypersingular fractional operators thus inherit the unweighted and weighted mapping properties analogous to those for classical Riesz potentials (Hoang et al., 2024).

6. Comparison to Classical Sparse Domination

Calderón–Zygmund rough singular integrals ($\alpha=1$) admit sparse domination via $\mathcal S_{1,1}(|f|)$ [see references in (Hoang et al., 2024)]. Fractional integrals $I_\alpha f$ are dominated by $\mathcal S_{\alpha,1}(f)$. For hypersingular cases ($0 < \alpha < 1$), the domination $\mathcal S_{\alpha,r}(|\nabla f|)$ reflects the increased singularity and higher regularity required. This establishes $T_{\Omega, \alpha}$ as a hypersingular cousin of the classical Riesz potential, showing that the sparse domination principle spans rough singular, hypersingular, and fractional regimes under one analytic framework (Hoang et al., 2024, Hu et al., 31 Dec 2025).

7. Significance and Perspectives

The introduction of hypersingular cousins of sparse operators, quantified by sparseness $\eta$ and degree $K_{\mathcal S}$, offers a flexible and sharp machinery for analyzing operators beyond the Calderón–Zygmund scope, fully characterizing their mapping properties, including strong-type, weak-type, and restricted weak-type bounds along critical lines, as in Forelli–Rudin type and Bergman projection models (Hu et al., 31 Dec 2025). The real-variable, dyadic approach resolves previous inquiries regarding effective analytic tools for hypersingular regimes, and connects these models directly to sharp Sobolev inequalities essential in analysis and PDE theory.