Weighted Riesz Potential Inequalities

Updated 30 June 2026

Weighted Riesz Potential Inequalities are a class of results that establish boundedness and sharp structural properties for fractional integral operators in Muckenhoupt weighted spaces.
They use advanced techniques such as sparse domination and Corona decompositions to derive two-weight, mixed, and variable exponent estimates.
The inequalities provide sharp constants and optimal weight conditions, underpinning precise embedding results in Sobolev, Morrey, and matrix-weighted frameworks.

Weighted Riesz potential inequalities concern the boundedness properties, structural sharpness, and embedding consequences of the Riesz potential (fractional integral) operator in spaces equipped with Muckenhoupt-type weights, often under extension to two-weight, Lorentz, Morrey, variable-exponent, or vector/matrix-valued frameworks. The subject integrates classical potential theory, weighted harmonic analysis, and nonlinear embedding theorems. The form of the inequalities, the optimal weight conditions, and the precise constants are of central importance in analysis and PDEs.

1. Riesz Potentials and Weighted Spaces

The fractional Riesz potential operator of order $0<\alpha<n$ is

$I_\alpha f(x) = \int_{\mathbb{R}^n} \frac{f(y)}{|x-y|^{n-\alpha}}\,dy,$

naturally mapping $L^p(\mathbb{R}^n)$ into $L^q(\mathbb{R}^n)$ for exponents satisfying $1/p - 1/q = \alpha/n$ , $1 < p < n/\alpha$ , $q < \infty$ . The classical weighted analysis concerns characterizing weights $w$ for which the mapping

$I_\alpha: L^p(w^p) \to L^q(w^q)$

is bounded, with quantitative dependence on $w$ .

The central class is the Muckenhoupt–Wheeden $I_\alpha f(x) = \int_{\mathbb{R}^n} \frac{f(y)}{|x-y|^{n-\alpha}}\,dy,$ 0 condition:

$I_\alpha f(x) = \int_{\mathbb{R}^n} \frac{f(y)}{|x-y|^{n-\alpha}}\,dy,$ 1

for $I_\alpha f(x) = \int_{\mathbb{R}^n} \frac{f(y)}{|x-y|^{n-\alpha}}\,dy,$ 2. The operator-theoretic formulation often involves two-weight pairs $I_\alpha f(x) = \int_{\mathbb{R}^n} \frac{f(y)}{|x-y|^{n-\alpha}}\,dy,$ 3, with $I_\alpha f(x) = \int_{\mathbb{R}^n} \frac{f(y)}{|x-y|^{n-\alpha}}\,dy,$ 4, $I_\alpha f(x) = \int_{\mathbb{R}^n} \frac{f(y)}{|x-y|^{n-\alpha}}\,dy,$ 5, translating mapping properties of $I_\alpha f(x) = \int_{\mathbb{R}^n} \frac{f(y)}{|x-y|^{n-\alpha}}\,dy,$ 6 into testing conditions on $I_\alpha f(x) = \int_{\mathbb{R}^n} \frac{f(y)}{|x-y|^{n-\alpha}}\,dy,$ 7 and $I_\alpha f(x) = \int_{\mathbb{R}^n} \frac{f(y)}{|x-y|^{n-\alpha}}\,dy,$ 8 (Cruz-Uribe et al., 2012).

2. Mixed and Two-Weight Inequalities

Sharp forms of the weighted Riesz potential inequalities include mixed $I_\alpha f(x) = \int_{\mathbb{R}^n} \frac{f(y)}{|x-y|^{n-\alpha}}\,dy,$ 9– $L^p(\mathbb{R}^n)$ 0 estimates: $L^p(\mathbb{R}^n)$ 1 where $L^p(\mathbb{R}^n)$ 2, and $L^p(\mathbb{R}^n)$ 3 are alternative $L^p(\mathbb{R}^n)$ 4 constants (Cruz-Uribe et al., 2012).

Two-weight inequalities enter through testing

$L^p(\mathbb{R}^n)$ 5

and can be fully characterized by (restricted) testing on balls or equal-measure sets under mild rearrangement regularity conditions on $L^p(\mathbb{R}^n)$ 6 and $L^p(\mathbb{R}^n)$ 7. Necessity and sufficiency criteria can often be verified via functional rearrangement inequalities (Nursultanov et al., 2013).

Log-bump and Orlicz-bump extensions describe further regularity and near-endpoint improvement, where conditions involve Orlicz norms and variants (e.g., $L^p(\mathbb{R}^n)$ 8), and corona decompositions adapted to sparse potential averages are used to prove such results (Cruz-Uribe et al., 2012).

3. Pointwise and Subrepresentation Inequalities

Subrepresentation (pointwise control) results provide bounds of the form

$L^p(\mathbb{R}^n)$ 9

where $L^q(\mathbb{R}^n)$ 0 is a weighted Riesz-type operator

$L^q(\mathbb{R}^n)$ 1

For appropriate kernel, weight classes, and parameter balance, these inequalities extend subrepresentation to rough kernels, weighted and Morrey targets, and more general settings (Chamorro et al., 18 Aug 2025).

These pointwise forms underlie further embedding theorems, such as weighted Sobolev or Morrey-type inequalities, which relate the $L^q(\mathbb{R}^n)$ 2 or Morrey norms of operators $L^q(\mathbb{R}^n)$ 3 to $L^q(\mathbb{R}^n)$ 4 (or Morrey) norms of $L^q(\mathbb{R}^n)$ 5, with parameters determined by sharp homogeneity and integrability constraints.

4. Extensions: Variable Exponents, Morrey, Radial, and Matrix-Weighted Cases

Variable Exponent and Vector-Valued Settings

In the variable-exponent regime, weighted Riesz potential inequalities are established in spaces $L^q(\mathbb{R}^n)$ 6, with mapping into $L^q(\mathbb{R}^n)$ 7, for $L^q(\mathbb{R}^n)$ 8. The theory requires atomic decomposition, vector-valued inequalities, and off-diagonal Fefferman–Stein bounds. Notably, the "off-diagonal cube-norm" hypothesis can be eliminated, with the main technical tools being generalized vector-valued estimates (Rocha, 2 Nov 2025, Rocha, 2022).

Morrey-Type Spaces

Weighted bounds for the (generalized) Riesz potential in local and global Morrey-type spaces rest on suitable Muckenhoupt–Wheeden classes ( $L^q(\mathbb{R}^n)$ 9) and Zygmund-type or doubling conditions for auxiliary functions controlling the scale behavior. Both Spanne-type (norm-to-norm) and Adams-type (potential-to-maximal) estimates are available, with best constants captured by one-dimensional Hardy-type integrals (Kucukaslan, 2021).

Radial Functions and Weighted Lorentz Spaces

For radial functions, weighted Riesz inequalities in Lebesgue and Lorentz spaces are admissible in a wider range, with parameters determined by the "weight-defect" $1/p - 1/q = \alpha/n$ 0 and reflecting extra angular cancellation. Duality, interpolation, and improved Young–O'Neil-type convolution estimates for radial functions yield strong and endpoint estimates not available in the non-radial case. These have applications to weighted Besov embeddings for radial spaces (Nápoli et al., 2012).

Matrix-Valued Weights

For matrix-valued weights $1/p - 1/q = \alpha/n$ 1, a matrix $1/p - 1/q = \alpha/n$ 2 theory is formulated by reducing to the action of "reducing matrices" $1/p - 1/q = \alpha/n$ 3 controlling the $1/p - 1/q = \alpha/n$ 4 and $1/p - 1/q = \alpha/n$ 5 behavior of $1/p - 1/q = \alpha/n$ 6 on cubes. The Riesz potential and associated fractional maximal operators are bounded between $1/p - 1/q = \alpha/n$ 7 and $1/p - 1/q = \alpha/n$ 8 with sharp exponents depending on $1/p - 1/q = \alpha/n$ 9 (specifically, $1 < p < n/\alpha$ 0 for $1 < p < n/\alpha$ 1), thus generalizing the sharp classical Muckenhoupt–Wheeden theory. The proofs employ sparse/Corona decomposition and vector-valued maximal/Calderón–Zygmund techniques (Isralowitz et al., 2016).

5. Weighted Poincaré–Sobolev and Fractional/Non-Euclidean Generalizations

The weighted Riesz inequalities provide the precise constants and exponents governing the best possible Poincaré–Sobolev and fractional Gagliardo–Nirenberg inequalities in weighted settings, with the sharp $1 < p < n/\alpha$ 2 dependence and full resolution of conjectures on the role of the $1 < p < n/\alpha$ 3 and $1 < p < n/\alpha$ 4 constants (Claros, 3 Dec 2025). Extensions include high-order derivatives, Lorentz space refinements, and fractional seminorms with BBM-type gains.

On homogeneous Lie groups and structures with non-trivial dilation, anisotropic versions (fractional Caffarelli–Kohn–Nirenberg and Lyapunov-type inequalities) are available. These are characterized by precise homogeneity balances and weighted scaling relationships, and apply both to scalar and system settings (notably, fractional $1 < p < n/\alpha$ 5-sub-Laplacian systems and their eigenvalue spectra) (Kassymov et al., 2018).

For the Dunkl setting, which generalizes classical PDE by incorporating root systems and reflection symmetries, the weighted $1 < p < n/\alpha$ 6 boundedness of the Dunkl-Riesz potentials is available under conditions directly generalizing the Muckenhoupt–Stein theory, with sharp rearrangement and Hardy estimates determining the admissible weights (Abdelkefi et al., 2013).

6. Proof Techniques and Corona Decomposition

Modern proofs rely heavily on sparse domination and Corona decomposition techniques, particularly utilizing dyadic grid sparse forms and stopping-time arguments. The reduction of operators to sparse "averages" enables precise quantitative control in both strong and weak-type bounds, as well as in endpoint and bump Orlicz norm settings. This strategy parallels that used in the $1 < p < n/\alpha$ 7 conjecture for singular integral operators, and supports the extension to multidimensional, vector-valued, and variable-exponent settings (Cruz-Uribe et al., 2012, Isralowitz et al., 2016, Rocha, 2 Nov 2025).

Auxiliary lemmas underpinning these results include:

Fractional maximal operator bounds in the weighted setting,
Fractional Carleson embedding theorems,
Exponential decay estimates (John–Nirenberg for sparse forms),
Orlicz–Hölder and interpolation-type estimates for bump conditions.

These enable generalization and unification across a diverse range of weighted potential theory scenarios.

7. Optimality, Limiting Cases, and Applications

Sharpness of exponent dependence on weights is critical, as evidenced by counterexamples involving power-law weights and the failure of inequalities outside precise parameter ranges. Endpoint cases (weak-type, limiting $1 < p < n/\alpha$ 8 or $1 < p < n/\alpha$ 9), log–log bump extensions, and two-weight testing conditions provide both necessary and sufficient criteria for boundedness.

Applications include sharp Sobolev embeddings, eigenvalue bounds for fractional and degenerate PDEs, local regularity and existence in systems with degenerate or anisotropic weights, and embeddings in spaces with nonstandard growth or structure.

The subject remains a cornerstone of quantitative analysis in both classical and weighted harmonic analysis, geometric PDE, and analysis on groups and metric measure spaces.