REVIEW 4 minor
A $p = 2$ dichotomy for uniform Riesz transform bounds on stratified Lie groups
T0 review · 0 major / 4 minor · reviewed 2026-08-27 · deepseek-v4-flash
Pith's one-line read Theorem 1.1: the full horizontal Riesz transform on any stratified Lie group is weak type (1,1) with constant at most 2, uniformly in all group parameters.
desk verdict Strong, important result: uniform weak-type (1,1) bound with constant 2 for the full horizontal Riesz transform on every stratified Lie group; the only genuine defect, a false range equality in Prop 2.1, is easily repaired and not load-bearing. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The load-bearing object is the fractional obstacle problem on the cone $K_\alpha = \{v \in L^1 \cap V_\alpha : v \ge 0\}$ for the energy $J_\lambda^\alpha(v) = \tfrac12 E_\alpha(v) - \int (f-\lambda)v$, where $E_\alpha(v) = \|\mathcal{L}^{\alpha/4}v\|_2^2$ is the fractional Dirichlet form. Its unique minimizer $u$ satisfies $f = \mu + \mathcal{L}^{\alpha/2}u$ with $\mu = \lambda - \eta$, where $\eta$ is the Lewy-Stampacchia density $0 \le \eta \le (\lambda-f)_+$; the proof of that inequality is what turns the abstract variational inequality into the explicit complementarity relation $\mu = \lambda$ on $\{u>0\}$. The heat semigroup enters through the two-sided Gaussian kernel bound, which yields the fractional Nash inequality $\|v\|_2^2 \le C_{G,\alpha} \|v\|_1^{2\alpha/(Q+\alpha)} E_\alpha(v)^{Q/(Q+\alpha)}$ used to make the functional coercive. The final ingredient is the $L^2$ isometry identity for the full horizontal Riesz transform, which replaces the Fourier analytic cancellation of the Euclidean argument.
What would settle it
Compute on a concrete stratified group, for instance the Heisenberg group, the ratio $\lambda |\{x : |Rf(x)|>\lambda\}| / \|f\|_1$ for a sequence of $L^1$ functions approximating a point mass; if any limit exceeded $2$, Theorem 1.1 would be wrong. Alternatively, verify the proof's premise: if a stratified group's heat kernel failed the two-sided Gaussian estimate (2.7), the fractional Nash inequality would fail and the obstacle minimizer would not be guaranteed to exist.
Extended reading notes
Core claim
The paper's central result is Theorem 1.1: the operator $R = \nabla_H \mathcal{L}^{-1/2}$, initially defined on $L^1 \cap L^2$, extends uniquely to a continuous linear map from real-valued $L^1(G)$ to $L^{1,\infty}(G;\mathbb{R}^m)$, and $\|Rf\|_{L^{1,\infty}} \le 2\|f\|_1$ for every $f$. The proof is carried by an obstacle decomposition: every nonnegative $f \in L^1 \cap L^2$ is written as $f = \mu + \mathcal{L}^{\alpha/2}u$ with $0 \le \mu \le \lambda$, total mass preserved, $\mu = \lambda$ on the set where $u > 0$, and $\lambda |\{u>0\}| \le \|f\|_1$; for $\alpha \ge 1$ the horizontal gradient $\nabla_H u$ vanishes outside that set. The weak-type estimate follows by applying this at $\alpha = 1$ to $f^+$ and $f^-$, then splitting $Rf = R\mu + \nabla_H u$. Chebyshev's inequality and the exact $L^2$ identity $\|Rg\|_2 = \|g\|_2$ bound the $\mu$ contribution by $\|f\|_1$, while the measure bound on the support of $\nabla_H u$ gives the matching contribution, summing to constant $2$.
Load-bearing premise
The argument stands on the imported two-sided Gaussian bound for the heat kernel of the sub-Laplacian, Eq. (2.7): the bound must hold on every stratified group with finite constants, because it is what makes the fractional Nash inequality (2.11) hold and therefore guarantees that a minimizer $u$ exists.
Editorial extensions
If this is right
- Theorem 1.1 gives a weak $(1,1)$ bound with constant $2$ for each component $X_j\mathcal{L}^{-1/2}$, since $|X_j\mathcal{L}^{-1/2}f| \le |Rf|$ componentwise.
- Specializing to $G = \mathbb{R}^m$, Theorem 1.1 recovers the dimension-free Euclidean result of [OSS26].
- The result applies uniformly to every stratified group, including Heisenberg-type groups, Engel groups, and free Carnot groups of arbitrary rank and step, with no dependence on the homogeneous dimension or horizontal dimension.
- For complex-valued functions, the same argument yields a weak-type bound with constant $4$; the paper makes no claim that $4$ is optimal.
- Theorem 1.2 supplies a fractional balayage decomposition valid for every $0<\alpha<2$, which may be of independent use in endpoint estimates for other fractional differential operators.
Reading between the lines
- Because the proof uses only the heat semigroup, Dirichlet form, and a Nash inequality, the same constant-$2$ mechanism is likely to extend to any noncommutative or infinite-dimensional setting whose sub-Laplacian has a heat kernel satisfying two-sided Gaussian bounds, such as complete Riemannian manifolds with doubling and Poincaré inequalities.
- The exact mass conservation and complementarity suggest a 'sharp function' interpretation of $Rf$ via balayage; one could test whether the weak-type constant $2$ is asymptotically achieved by concentrating $f$ near the identity on groups with large step.
- A testable extension would be to replace the full horizontal gradient by a single component and see whether the obstacle decomposition can be adapted, since the $L^2$ isometry for a single component fails.
- The parameter $\alpha$ in Theorem 1.2 is not used in the endpoint proof; pushing the same decomposition to $\alpha > 1$ might yield weak-type estimates for higher-order Riesz-like operators built from $\mathcal{L}^{-\alpha/2}$.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper proves that on any stratified Lie group G, the full horizontal Riesz transform R = ∇_H L^{-1/2}, initially an L2 isometry, extends to real-valued L1(G) and satisfies the weak type (1,1) bound ||Rf||_{L^{1,∞}} ≤ 2||f||_1, with a constant independent of the horizontal dimension, homogeneous dimension, step, and group structure. The proof is variational: after splitting a real function into positive and negative parts, it constructs decompositions f± = μ± + L^{1/2}u± through a fractional obstacle problem (Theorem 1.2), with mass conservation, the bounds 0 ≤ μ± ≤ λ, and ∇_H u± supported on a set of measure at most λ^{-1}||f±||_1. The level-set estimate combines the measure of this support with Chebyshev's inequality applied to Rμ±, using the L2 isometry. The remainder of the paper establishes the obstacle decomposition for all 0 < α < 2 using heat-kernel Gaussian bounds, a fractional Nash inequality, spectral interpolation, and a Lewy–Stampacchia estimate. The Euclidean case G = R^m recovers the recent dimension-free theorem of Ouyang–Spector–Stockdale.
Significance. If correct, this is a substantial result: it gives a universal weak-type constant 2 for the full horizontal Riesz transform on every stratified group, including Heisenberg, Heisenberg-type, Engel, and free Carnot groups, with no dependence on any dimension or step. The proof is self-contained after standard external tools (spectral theory, Folland's subelliptic Sobolev spaces, and the VSC92 two-sided heat-kernel bounds), the constant emerges cleanly from an exact L2 isometry and a mass-preserving obstacle decomposition, and there is no circularity or free parameter. The paper is clearly written and the main line from the variational decomposition to the endpoint estimate is checkable. My reading found no load-bearing flaw; the mathematical inaccuracies are local and repairable.
minor comments (4)
- [Section 2.1, proof of Proposition 2.1] The displayed chain Ran(L^{1/2}) = (ker L^{1/2})^\perp = L^2(G) is false for the unbounded self-adjoint operator L: its range is dense in (ker L)^\perp, but it is not equal to it. This does not affect the conclusion, because the isometry U is defined on the dense subspace Ran(L^{1/2}) and therefore extends uniquely to L^2, but the proof should be rewritten using density rather than equality.
- [Section 2.2, Eq. (2.4)] For complex-valued v,w ∈ Vα the kernel representation should have a complex conjugate on the second factor, namely (v(x)-v(xz)) overline{(w(x)-w(xz))} Jα(z) dz dx; as written the identity is not the correct sesquilinear form. All later applications are to real functions, so this is a local correction rather than a substantive gap.
- [Section 3, proof of Theorem 1.1] In the Borel–Cantelli step, the displayed bound on the sum gives at most ∑_{r≥1} 2^{-r} = 1, so the strict inequality '< 1' is not literally correct; it should be '≤ 1' or, more simply, '< ∞'. The argument only needs finiteness of the sum.
- [Section 2.2, Eq. (2.7)] The two-sided Gaussian estimate (2.7) and the resulting Nash inequality (2.11) are imported from VSC92 with constants that may depend on the group. The paper should state this explicitly so that the universality claim is clearly understood as applying only to the final weak-type constant 2, not to the auxiliary constants used in the existence proof.
Circularity Check
No significant circularity: the endpoint estimate is self-contained after standard external tools; one repairable, non-circular range-equality slip.
full rationale
The derivation is self-contained. The load-bearing exogenous input is the two-sided Gaussian heat-kernel bound (2.7), quoted from VSC92 Theorem VIII.2.9 with no author overlap, which is a standard external theorem; from it (or from the dilation scaling (2.3) plus Schwartz kernel bounds) the paper derives the Nash inequality (2.11) used only to prove coercivity of the obstacle functional. Proposition 4.2 gives the minimizer; Propositions 4.3–4.5 yield the Lewy–Stampacchia bound, the identity L^{α/2}u = f − μ, and mass conservation; Lemma 4.6 gives complementarity μ = λ on {u > 0}. Theorem 1.2 then assembles these independent facts. In Section 3, the weak-type estimate uses only the level-set bound λ|Ω| ≤ ‖f‖_1, the L2 isometry of Proposition 2.1, and Chebyshev; the constant 2 is the sum of two explicitly bounded contributions, not an input. OSS26 is used only as Euclidean precedent and for recovery in the abelian case, and no load-bearing self-citation or fitted parameter appears. A non-circular technical flaw should be flagged: in Proposition 2.1 the assertion 'Ran(L^{1/2}) = (ker L^{1/2})^⊥ = (ker L)^⊥ = L^2(G)' is false (the range is dense in L2, not equal); this does not undermine the argument because the isometry extends uniquely by density, and the later uses of Proposition 2.1 only need the existence of that extension. Consequently, no circular step is identified.
Assumptions & free parameters
assumptions (4)
- domain assumption Two-sided Gaussian heat kernel estimates (2.7) hold for the sub-Laplacian on every stratified Lie group.
- standard math C_c∞(G) is dense in the fractional energy space Vα with respect to the form norm (Eq. 2.8).
- standard math The heat semigroup has a symmetric convolution kernel p_t of mass one with dilation scaling and Schwartz p1; positive distributions are represented by Radon measures.
- standard math For a nonnegative self-adjoint operator with zero kernel, the range of L^{1/2} is dense in L2 (the paper states this as equality, which is not generally true).
Cite this review
Pith. "Pith review of A $p = 2$ dichotomy for uniform Riesz transform bounds on stratified Lie groups." pith.science (2026). https://pith.science/paper/OMBPFQVV
@misc{pith2026260820267,
author = {Pith},
title = {Pith review of: A $p = 2$ dichotomy for uniform Riesz transform bounds on stratified Lie groups},
year = {2026},
howpublished = {\url{https://pith.science/paper/OMBPFQVV}},
note = {Machine review of arXiv:2608.20267}
}
abstract
Let $\mathbb{G}$ be a stratified Lie group and $\mathcal L$ be its sub-Laplacian. We prove that the full horizontal Riesz transform $\nabla_{H} \mathcal L^{-1/2}$ is of weak type $(1,1)$ on real-valued functions, with constant at most $2$. In particular, the constant is independent of the horizontal dimension, the homogeneous dimension, the step, and the underlying group structure of $\mathbb{G}$. Our result provides a noncommutative generalization of the dimension-free Euclidean theorem of Ouyang, Spector, and Stockdale arXiv:2608.18068, with the same universal constant. Our proof relies upon a fractional obstacle problem adapted to stratified Lie groups by using the functional calculus of $\mathcal L$ instead of the Fourier transform. As a consequence, by interpolation we obtain uniform $L^p$ bounds for the full horizontal Riesz transform $\nabla_{H} \mathcal L^{-1/2}$ for $p \in (1,2]$. By contrast, for every $p > 2$, we construct a sequence of stratified Lie groups with fixed horizontal dimension $5$ and steps tending to infinity for which the $L^p$ norms of the horizontal Riesz transforms diverge.
Reviewed August 27, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.