Weak-Type (1,1) Estimates in Harmonic Analysis

• Weak-type (1,1) estimates are defined by the bound μ({x: |Tf(x)| > λ}) ≤ C||f||₁/λ, offering endpoint control for singular and nonlinear operators.
• The theory underpins convergence theorems in harmonic analysis by providing sharper control than standard Lᵖ estimates, critical in singular settings.
• Methodologies such as dyadic decompositions and Calderón–Zygmund techniques, along with noncommutative extensions, enable precise endpoint bounds in PDEs and ergodic theory.

Weak-type (1,1) estimates constitute the foundational endpoint theory in classical and modern harmonic analysis, ergodic theory, and partial differential equations. These estimates assert control over the distribution of large values of nonlinear or singular integral operators applied to $L^1$ data, using a scale-invariant quasinorm—specifically, for an operator $T$, the measure of $\{x:|Tf(x)|>\lambda\}$ is bounded by $C\|f\|_{L^1}/\lambda$. Such endpoint control is sharper than $L^p$-boundedness for $p>1$, is critical in singular scenarios, and underpins many convergence theorems in analysis.

1. Definitions and Prototypical Results

Let $T$ be a sublinear operator on functions over a measure space $(X,\mu)$. $T$ is said to be of weak type $(1,1)$ if for some $T$0, for all $T$1 and all $T$2,

$T$3

Equivalently, $T$4, the weak $T$5 (quasi-)norm space.

This property holds for the Hardy-Littlewood maximal operator, Calderón–Zygmund singular integrals with standard kernels, and forms the endpoint for a classical interpolation scale. Notably, it often fails for “rougher” operators unless additional cancellation or integrability conditions are imposed.

2. Weak-type (1,1) for Discrete Maximal Functions and Ergodic Averages Along Thin Sets

Discrete Averages along Thin Arithmetic Sets

Let $T$6 denote a “thin” sequence of integers defined via inverse-regularly-varying functions $T$7 with window width function $T$8: $T$9 so that $\{x:|Tf(x)|>\lambda\}$0 has Banach density zero but contains arbitrarily long blocks.

Given $\{x:|Tf(x)|>\lambda\}$1, define the maximal averaging operator: $\{x:|Tf(x)|>\lambda\}$2 where $\{x:|Tf(x)|>\lambda\}$3 and $\{x:|Tf(x)|>\lambda\}$4. The maximal ergodic average on measure-preserving systems transfers to the discrete case via Calderón's principle.

Main Theorem (Theorem 1.5): For $\{x:|Tf(x)|>\lambda\}$5 and compatible $\{x:|Tf(x)|>\lambda\}$6, there exists $\{x:|Tf(x)|>\lambda\}$7 so that for all $\{x:|Tf(x)|>\lambda\}$8 and $\{x:|Tf(x)|>\lambda\}$9,

$C\|f\|_{L^1}/\lambda$0

By interpolation, $C\|f\|_{L^1}/\lambda$1 extends to $C\|f\|_{L^1}/\lambda$2 for all $C\|f\|_{L^1}/\lambda$3. This endpoint theorem facilitates $C\|f\|_{L^1}/\lambda$4 pointwise convergence of ergodic averages along $C\|f\|_{L^1}/\lambda$5 and disproves a conjecture of Rosenblatt–Wierdl for certain zero Banach density sets.

Methodology Summary:

• Reduce to a dyadic-smoothed maximal operator $C\|f\|_{L^1}/\lambda$6.
• Prove a decomposition $C\|f\|_{L^1}/\lambda$7 with $C\|f\|_{L^1}/\lambda$8 the main piece (size $C\|f\|_{L^1}/\lambda$9, regularity $L^p$0), $L^p$1 negligible ($L^p$2).
• Apply a multiscale Calderón–Zygmund decomposition for $L^p$3, refining “bad” parts into “very bad”, “bad”, and “locally constant” components.
• For each piece, exploit mean-zero, regularity, and fine combinatorics to show the level set has measure $L^p$4.
• Abstract summations package these bounds into a general theorem paralleling previous work (e.g. [Mirek 2015]) (Daskalakis, 2023).

Maximal Functions along 1-Regular Sequences

For $L^p$5 (threshold growth rate), the maximal function $L^p$6 defined analogously admits a weak-type (1,1) estimate: $L^p$7 (Trojan, 2020).

A key step is the precise control of the self-convolution of the normalized counting kernel, showing $L^p$8, which is the unweighted “Calderón–Zygmund kernel” property essential to all subsequent arguments.

The result is sharp: if the sequence grows either too slowly or too rapidly, the maximal operator ceases to be of weak-type (1,1).

3. Discrete Rough Singular Integrals and Their Inverses

Maximal, strongly oscillatory, or rough discrete singular integrals, e.g. the discrete rough Hilbert transform,

$L^p$9

with $p>1$0, admit invertibility and weak-type (1,1) estimates for their inverses under fine regularity and spectrum assumptions. The critical algebra is to realize all operators and their inverses as convolution operators within a Banach algebra $p>1$1 with cancellation, $p>1$2 decay, and Hölder regularity at multiple scales (Paluszynski et al., 2017).

Key points:

• The norm $p>1$3 mixes $p>1$4 and regularity components.
• Invertibility in $p>1$5 ensures endpoint bounds, with logarithmic growth in $p>1$6 matching an explicit block construction example.
• This framework extends (with modifications) to continuous rough operators and multi-frequency contexts.

4. Calderón–Zygmund Operators: Paradigm and Generalizations

For a Calderón–Zygmund operator $p>1$7 (principal value integral with standard kernel; see size and smoothness conditions below), classical theory ensures weak-type (1,1) (Stockdale, 2018): $p>1$8

Key Structure:

• Size: $p>1$9.
• Smoothness: $T$0 for $T$1, and similarly in $T$2.
• $T$3 boundedness: $T$4.

The modern proofs often rely on explicit “good–bad” decompositions via Whitney covering, mean-zero atom constructions, and explicit cancellation sets, mapping the problem to estimating interactions between “good parts” (controlled in $T$5) and “bad parts” (localized, mean-zero, controlled outside slightly enlarged supports).

This approach extends to the weighted case ($T$6 weights) and to non-doubling measures under additional controls.

5. Weak-type (1,1) and Dimension: Riesz Transforms and Quantitative Rates

A specific and central case is the Riesz transform in $T$7. Here, Spector–Stockdale (Spector et al., 2020) established the best-known bound: $T$8 where $T$9 is absolute and $(X,\mu)$0 is the ambient dimension.

• The reduction principle: to control the general $(X,\mu)$1 function case, it suffices to treat discrete measures (linear combinations of Dirac deltas).
• For such measures, Janakiraman’s result gives the $(X,\mu)$2 bound.
• It is an open question, attributed to Stein, whether dimension-free weak-type (1,1) bounds are possible for Riesz transforms—this remains unresolved for both the Riesz transforms and general Calderón–Zygmund operators (Spector et al., 2020).

6. Endpoint Weak-type Theory in PDEs and Ergodic Theory

Linear Elliptic Equations with Dini-regularity

For second-order linear elliptic operators in divergence or non-divergence form with uniformly elliptic, Dini-mean-oscillation coefficients, weak-type (1,1) estimates for the gradient (divergence form) or Hessian (non-divergence form) hold up to the boundary: $(X,\mu)$3 given appropriate regularity and boundary conditions (Dong et al., 2016, Dong et al., 2017).

Methodology involves blockwise CZ decomposition and duality to control off-support integrals by leveraging local Dini-assumptions—crucially, Dini oscillation is sharp, as weaker conditions do not yield unweighted weak-(1,1).

Noncommutative Weak-type (1,1) Theory

For noncommutative $(X,\mu)$4 spaces (von Neumann algebras with a trace), endpoint weak-type (1,1) holds for certain ergodic square functions, maximal averages, and now for Calderón–Zygmund operators under $(X,\mu)$5-mean Hörmander (i.e., $(X,\mu)$6-average regularity) hypotheses (Lai et al., 7 Dec 2025, &&&10&&&). The key tool is a refined noncommutative Calderón–Zygmund decomposition using Cuculescu's projections, smoothing of “bad” parts (with operator-valued mollifiers), and a careful analysis of off-diagonal operator interactions.

7. Extensions: Strongly Singular Operators, Fourier Integral Operators, Bergman Projections, and More

• For maximal truncated rough singular integral operators with kernel $(X,\mu)$7 and minimal angular integrability $(X,\mu)$8, sharp endpoint weak-type (1,1) holds, closing a long-standing open problem (Lai, 25 Aug 2025).
• Endpoint theory has been extended to strongly oscillatory and strongly singular operators (e.g., with kernels behaving like $(X,\mu)$9 with minimal regularity) (Folch-Gabayet et al., 2018).
• For the harmonic Bergman projection on the unit ball (with Calderón–Zygmund kernel in the Bergman metric), weak-type (1,1) can be proved using adaptations of the dyadic decomposition and CZ methodology to the metric-measure space involved (Zhang, 5 Jan 2026).
• For Fourier integral operators of critical scaling (order $T$0), a local-to-global criterion converts locally uniform weak-(1,1) estimates and off-diagonal decay into a global endpoint bound (Cardona et al., 2021).

8. Notable Techniques and Modern Directions

• Dyadic and Whitney Decompositions: Adaptive coverings and partitioning to localize badness and exploit cancellations.
• Bellman Function Methods: Compute sharp constants and extremizers, e.g., for positive dyadic shifts, yielding optimal factor (2 in the local weak-(1,1) for such Lerner operators) (Rey et al., 2013).
• Probabilistic and Interpolation Tools: Noncommutative Khintchine, Cotlar-Stein almost-orthogonality, and Rademacher–Menshov techniques in maximal function and oscillatory sum control.
• Microlocal Analysis: Directional and frequency decomposition in commutator or rough operator treatment (Seeger, 2012, Lai, 25 Aug 2025).
• Weak-type Theory for Non-Translation Invariant, Non-doubling, and Metric Spaces: Adaptation to general doubling, spaces of homogeneous type, and beyond (Stockdale, 2018).

9. Tables for Major Model Results

Operator Class	Main Weak-Type (1,1) Bound	Key Reference
CZ (standard kernel, Euclidean)	$T$1	(Stockdale, 2018)
Max. avg. along 1-regular/“thin” sets	$T$2	(Daskalakis, 2023Trojan, 2020)
Riesz transform, $T$3	$T$4	(Spector et al., 2020)
Maximal truncated rough singular integral	$T$5 (endpoint)	(Lai, 25 Aug 2025)
Dyadic positive shifts (local sharp const)	$T$6	(Rey et al., 2013)
Noncommutative CZ, ergodic square fxn	$T$7	(Lai et al., 7 Dec 2025Hong et al., 2019)

10. Open Problems and Future Directions

• Dimension-free or improved dependence for higher-dimensional weak-type (1,1) bounds (Riesz transforms, CZ class operators).
• Endpoint behavior under relaxed ($T$8-mean) Hölder/Hörmander kernel conditions, particularly in the noncommutative setting.
• Precise constants, extremizers, and potential Orlicz improvements near $T$9.
• Robust endpoint theory for maximal and variational truncations beyond $(1,1)$0, $(1,1)$1 (variation norms, jump inequalities).
• Structured, metric-measure settings (e.g., spaces of homogeneous type, Bergman or other integral projections).

Weak-type (1,1) theory thus provides the optimal nonlinear summability and maximal control at the $(1,1)$2 endpoint for a vast class of operators, linking deep regularity, oscillation, and cancellation mechanisms in analysis (Daskalakis, 2023, Trojan, 2020, Paluszynski et al., 2017, Stockdale, 2018, Spector et al., 2020, Lai et al., 7 Dec 2025, Lai, 25 Aug 2025, Dong et al., 2016, Dong et al., 2017).