On a fractional boundary version of Talenti's inequality in the unit ball

Tobias Weth; Yassin El Karrouchi

REVIEW 3 major objections 5 minor 10 references

On a fractional boundary version of Talenti's inequality in the unit ball

T0 review · 3 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read For fractional Laplacian Dirichlet–Poisson problems in the unit ball, the pointwise Talenti comparison fails universally for nonsymmetric radial sources, and the boundary inequality is reversed.

desk verdict A solid, novel symmetrization paper whose main theorem likely stands, but the key technical lemma needs a repair pass before publication. read the letter →

arxiv 2411.14534 v1 pith:RR73VGBM submitted 2024-11-21 math.AP

Yassin El Karrouchi , Tobias Weth This is my paper · ORCID

classification math.AP MSC 35R1135B0626D1535J25

keywords fractionalLaplacianTalenticomparisonprincipleSchwarzsymmetrizationnormalderivativeMartinkernelunitballradialfunctionsboundaryinequality

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper asks whether the classical pointwise Talenti comparison principle—symmetrizing the source cannot make the solution any smaller—survives for fractional Dirichlet–Poisson problems in the unit ball. It proves that for $s \in (0,1)$ it does not: for every radial nonnegative source $f$ that is not already Schwarz symmetric, the symmetrized solution $(u_f)^*$ is strictly larger than $u_{f^*}$ throughout a neighborhood of the boundary, while the boundary fractional normal derivatives satisfy the reverse inequality, with equality only for $f = f^*$. It also proves that for nonradial sources concentrated in a sufficiently small ball inside the ball, the usual boundary Talenti inequality does hold. This matters because it shows the fractional setting $s \in (0,1)$ behaves fundamentally differently from the classical Laplacian, where radial data give boundary equality automatically. The paper notes in Remark 5.4 that it leaves open whether the boundary Talenti inequality holds for arbitrary nonradial sources when $s \in (1,N]$, and it has no result for $s > N$.

What carries the argument

The load-bearing object is the boundary mean formula of Proposition 2.4: for a nonnegative $u$ with $u/\delta^s$ continuous near the boundary and strictly positive, the boundary value of its Schwarz symmetrization is $$\frac{u^*}{\delta^s} = \left(\frac{1}{\omega_{N-1}}\int_{$S^{{N-1}}$} \left(\frac{u}{\delta^s}(\vartheta)\right)^{-1/s} d\$\sigma$(\vartheta)\right)^{-s} \quad \text{on } \partial\$\Omega$.$$ For radial $f$, the solution $u_f$ is radial, so $(u_f)^* = u_f$, and Lemma 2.3 gives $u_f/\delta^s$ as a constant determined by the weighted integral $\int_\Omega f(y)(1-|y|^2)^{s-1}\,dy$. A layer-cake decomposition of $f - f^*$ then leaves the strictly monotone kernel $k_{N,s}(r) = (2\kappa_{N,s}\omega_{N-1}/s)(1-r^2)^{s-1}$, whose monotonicity for $s<1$ (and reversal for $s>1$) decides the direction of the boundary inequality; the nonradial case is handled by comparing the same harmonic mean against the Martin kernel bounds in Lemma 4.4.

What would settle it

Construct a non-radial positive function $u$ on $B_1(0)$ with $u/\delta^s$ continuous near the boundary and strictly positive, for example $u(x) = (1-|x|)^s \psi(x/|x|)$ with a nonconstant smooth $\psi$, and numerically compare the boundary value of $u^*/\delta^s$ with the right-hand side of formula (25); any disagreement beyond rounding error would disprove Proposition 2.4 and collapse the proof of Theorem 1.1.

Watch

Extended reading notes

Core claim

The central claim is Theorem 1.1: in $\Omega = B_1(0) \subset \mathbb{R}^N$ with $s \in (0,1)$, for every radial nonnegative $f \in L^\infty(\Omega)$ one has $(u_f)^*/\delta^s \geq u_{f^*}/\delta^s$ on $\partial\Omega$, with equality if and only if $f = f^*$; consequently, whenever $f$ is radial but not symmetric, $u_{f^*} < (u_f)^*$ in $\Omega \setminus B_r(0)$ for some $r \in (0,1)$, so the classical pointwise Talenti inequality fails. A companion result (Theorem 1.2) gives the opposite boundary inequality for nonradial sources supported in a small ball $B_\rho(\xi) \subset \Omega$, provided $\rho$ satisfies condition (13). The proofs show that all boundary comparisons reduce to comparing a single harmonic-mean quantity for the symmetrized profile with a weighted $L^1$ norm of the source, and the paper extends the radial comparison to $s > 1$, where the inequality reverses.

Load-bearing premise

The whole boundary comparison rests on Proposition 2.4: the claim that the boundary value of the symmetrized quotient $u^*/\delta^s$ is the $(-1/s)$-mean of the boundary values of $u/\delta^s$; if that formula fails for some positive $u$ with the stated regularity, the sign conclusions near the boundary do not follow.

Editorial extensions

If this is right

For $s \in (0,1)$ the pointwise Talenti inequality cannot hold for radial sources unless the source is Schwarz symmetric; symmetrization is not a valid pointwise comparison tool for fractional Poisson problems in balls.
The fractional normal derivative of the symmetrized-problem solution is always the smaller one for radial data, reversing the classical boundary-derivative order from the local case.
For nonradial sources concentrated in a sufficiently small ball inside $\Omega$, the boundary Talenti inequality (12) does hold strictly, so the direction of the boundary comparison depends on where the source is located.
For $s > 1$ the inequality reverses: for radial $f$, $(u_f)^*/\delta^s \leq u_{f^*}/\delta^s$ on the boundary, so $s = 1$ is the exact threshold between the two regimes.
A Green-function consequence is that $G(\cdot,\xi)^*/\delta^s < G(\cdot,0)/\delta^s$ on $\partial\Omega$ for every $\xi \neq 0$, making the boundary Talenti inequality true for point sources at any nonzero location.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The paper does not pursue the direction of the boundary inequality for general nonradial $f$ when $s \in (0,1)$; as a testable consequence of its mean formula, one would expect the boundary Talenti inequality (12) to hold only when the source is sufficiently concentrated, in the sense that the ratio of the harmonic mean to the weighted norm of the Martin kernel stays below 1.
The same layer-cake argument with the kernel $k_{N,s}$ should extend to other boundary-value problems with explicit Green functions in balls, such as fractional heat or Schrödinger operators, where the sign of $s-1$ in the monotonicity of the boundary kernel would again decide the direction of the comparison; this is an extension the paper does not mention.
The equality case $f = f^*$ in Theorem 1.1 suggests that rearrangement inequalities of Talenti type should be formulated at the level of boundary data rather than pointwise in the interior, since the boundary value of $u/\delta^s$ for radial solutions is not determined by the total mass of $f$ but by its rearrangement-invariant weighted integral.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Two things you should know. First, the main result is substantial and likely correct: for radial nonnegative data in the unit ball, the fractional boundary Talenti inequality fails in every dimension, and the reverse inequality holds, with equality only for f=f*. This is a genuinely new phenomenon, distinct from the local case and from the one-dimensional counterexamples in Ferone–Volzone. The s>1 reversal is also new and interesting. Second, the proof's load-bearing lemma, Proposition 2.4, has gaps that need fixing, though they look repairable.

What is good: the paper uses explicit kernel formulas and layer-cake arguments, with no circularity. Lemma 2.3, the boundary derivative formula for radial f, is clean and useful. The Martin kernel comparison in Proposition 4.3 is a nice byproduct. The theorems are stated carefully, with hypotheses that match the applications.

Soft spots: Proposition 2.4 is the weak point. As stated, its hypotheses do not imply the existence of a boundary trace ψ on S^{N-1}, but every application has u/δ^s ∈ C(Ω), so a trace exists—this is a hypothesis fix, not a change of result. The level-set formula for μ_h(t) is written without a positive part, so it is false for t larger than the essential infimum of ψ; the limit |x|→1 only needs small t, so the argument can be restricted to that range. And equation (27) has a typo: the factor should be N^s, not N; as written the harmonic-mean constant does not follow. None of these undermine Theorem 1.1, but the proof of the lemma is not fully correct as printed.

Bottom line: this is a good paper that deserves a serious referee. The referee should ask for a corrected and expanded proof of Proposition 2.4 and a careful pass for similar typos. After that revision, it should be published. I would bring it to a reading group and, once the revision is out, cite it.

Referee Report

3 major / 5 minor

Summary. The paper studies boundary versions of Talenti's comparison principle for fractional Dirichlet-Poisson problems in the unit ball Ω = B1(0) ⊂ R^N. For 0 < s < 1, it proves a reverse boundary Talenti inequality for radial nonnegative data: (u_f)^*/δ^s ≥ u_{f^*}/δ^s on ∂Ω, with equality only for f = f^*, and consequently that the classical pointwise inequality (u_f)^* ≤ u_{f^*} fails for every radial f that is not Schwarz symmetric. It also establishes a positive boundary Talenti inequality for nonnegative data supported in a sufficiently small ball away from the origin (Theorem 1.2), and, for s > 1, reverses the radial inequality (Theorem 5.2). The key structural tool is Proposition 2.4, an s-harmonic mean formula for the boundary value of the Schwarz symmetrization in terms of the fractional normal derivative, applied through Corollary 2.5 and in the proofs of Propositions 4.3 and Lemma 4.4. The arguments rely on explicit Green and Martin kernel formulas, layer-cake representations, and rearrangement properties.

Significance. If the central lemma is repaired, the results are significant: they show a universal failure of the pointwise Talenti inequality for radial nonsymmetric data in the fractional radial setting, in all dimensions, in contrast to the local case s = 1; they also provide sharp boundary comparison theorems in both directions depending on s. The paper exploits explicit formulas for the fractional Green and Martin kernels in the ball, and the layer-cake/rearrangement arguments are elegant and mostly rigorous. The main claims are falsifiable and well grounded in known results on fractional boundary regularity. However, the s-harmonic mean formula in Proposition 2.4, which is load-bearing for all later boundary comparisons, is not proved as stated; the proof has a missing hypothesis and two technical errors. These are repairable without changing the main theorems, because all applications satisfy the needed regularity and because the asymptotic range used for the boundary limit is exactly the range where the level-set formula is valid.

major comments (3)

[Section 2, Proposition 2.4] The stated hypotheses (u/δ^s ∈ C(Ω \ B_r(0)) and ess inf u/δ^s > 0) do not justify the proof's immediate use of the boundary traces ψ(ϑ) and the quantity max_{S^{N-1}} ψ; the function ψ is defined only on Ω, not on the boundary. The proposition should be restated with the additional assumption that u/δ^s extends to a continuous positive function on the closure Ω, or at least to a continuous function on an exterior collar with boundary values. All applications in Corollary 2.5, Lemma 4.4, and Proposition 4.3 satisfy this via fractional boundary regularity and Hopf's lemma, so the gap is local but the statement as written is not well-posed.
[Section 2, proof of Proposition 2.4, level-set formula] The formula μ_h(t) = (1/N) ∫_{S^{N-1}} (1 - (t/ψ(θ))^{1/s})^N dσ and the implicit equation (26) are only valid for t < essinf ψ; for t between essinf ψ and max ψ the integrand is negative on the set where ψ(θ) < t, and the displayed identity is false unless a positive part is included. Since the subsequent limit |x| → 1 only uses values of h^*(x) that tend to 0, the restriction to the range t < essinf ψ is sufficient, but the proof should state this qualification explicitly.
[Section 2, Eq. (27)] The displayed chain of equalities contains a factor error: the second equality should be N^s lim_{|x|→1} h^*(x)/(1-|x|^N)^s, not N lim. Restoring the exponent s, together with the Taylor expansion from (26), yields the stated final constant; without it the intermediate identity is inconsistent with the preceding line. This is a typographical error in a load-bearing displayed computation and must be corrected.

minor comments (5)

[Abstract] The word 'universial' should be 'universal'.
[Section 4, Lemma 4.4 proof] The final sentence of the proof says 'on Ω' where the claim (37) is on ∂Ω; this should read 'on ∂Ω'.
[Section 4, Lemma 4.4 proof] The proof refers to 'Lemma 2.4' when invoking the s-harmonic mean formula; the reference should be to Proposition 2.4.
[Section 5, Proposition 5.5 proof] The first sentence says 'The proof is exactly the same as the proof of Proposition 5.5'; this should refer to Proposition 4.3.
[Section 5, Theorem 5.3] In the statement of Theorem 5.3, the condition (43) contains '1-|x|' in the denominator of the right-hand side; this should be '1-|ξ|' as in the analogous condition (13) of Theorem 1.2. The variable x is otherwise not defined in the theorem statement.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular reasoning found: the boundary comparisons follow from explicit Green and Martin kernel formulas plus rearrangement theory, with no fitted parameters and no load-bearing self-citation chain.

full rationale

I walked the derivation chain. Theorem 1.1 uses Lemma 2.3, which evaluates u_f/delta^s for radial f by integrating the explicit Martin kernel (21), and Corollary 2.5/Proposition 2.4, which expresses the boundary value of the Schwarz-symmetrized ratio as a harmonic mean. That harmonic-mean formula is proved from the level-set definition of h^* and a Taylor expansion; it is not assumed, and the target inequalities do not enter the derivation. The strict comparison in Theorem 1.1 follows from the layer-cake representation of the radial weight k_{N,s}, whose monotonicity is explicit, so no fitted input is renamed as a prediction. Theorem 1.2 and Proposition 4.3 use the same derived formula together with elementary estimates such as T_{N,tau}(xi) > 1; the constants are explicit functions of s and N, not calibrated to data. The cited references [2], [4], [9], and [10] are external prior results (Green function and Martin kernel formulas, boundary regularity, and Talenti's theorem), and none of the load-bearing steps reduces to a self-citation by the present authors. I also note that the proof of Proposition 2.4 contains technical gaps concerning boundary trace assumptions, a positive-part issue in equation (26), and an apparent N versus N^s factor in equation (27); these are correctness defects in a derived lemma, not circularity, because the lemma is not equivalent to the paper's conclusions and no input is being relabeled as an output. Overall, the derivation is self-contained conditional on the cited kernel formulas, so the appropriate circularity score is 0.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The paper introduces no new physical or mathematical entities and fits no numerical parameters. It builds on explicit known kernel formulas and standard rearrangement facts; the constants kappa_{N,s} are fixed by the kernel, not adjusted to data.

assumptions (5)

domain assumption Fractional strong maximum principle and Hopf lemma: for nonnegative f, u_f > 0 in Omega and u_f/delta^s > 0 on the boundary.
Used to apply Proposition 2.4 and to ensure boundary ratios are finite and positive; taken from Ros-Oton's survey [8, Lemma 7.3].
domain assumption Explicit Green function and Martin kernel formulas for the fractional Laplacian in the unit ball.
Equations (17) and (21) are quoted from Blumenthal-Getoor-Ray [4] and Abatangelo-Jarohs-Saldana [2]; Lemma 2.3 and all estimates in Section 4 rely on them.
domain assumption Fractional elliptic regularity up to the boundary: u_f/delta^s extends to C^alpha(Omega).
Quoted from Ros-Oton-Serra [9]; needed for boundary evaluations and for the asymptotic formula in Proposition 2.4.
standard math Standard properties of Schwarz symmetrization: equimeasurability, order preservation, and commutation with truncation.
Used throughout Section 2.1 and in the layer-cake computations of Section 3.
domain assumption For s>1, the Green function representation and positivity preserving property in the ball.
Quoted from [2]; used in Lemma 5.1 and Theorems 5.2 and 5.3 because maximum principles may fail for higher-order fractional Laplacians.

how reviews work

0 comments

Cite this review

Pith. "Pith review of On a fractional boundary version of Talenti's inequality in the unit ball." pith.science (2026). https://pith.science/paper/RR73VGBM

@misc{pith2026241114534,
  author       = {Pith},
  title        = {Pith review of: On a fractional boundary version of Talenti's inequality in the unit ball},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RR73VGBM}},
  note         = {Machine review of arXiv:2411.14534}
}

abstract

Inspired by recent work of Ferone and Volzone arXiv:2007.13195, we derive sufficient conditions for the validity and non-validity of a boundary version of Talenti's comparison principle in the context of Dirichlet-Poisson problems for the fractional Laplacian $(-\Delta)^s$ in the unit ball $\Omega= B_1(0) \subset \mathbb{R}^N$. In particular, our results imply a universial failure of the classical pointwise Talenti inequality in the fractional radial context which sheds new light on the one-dimensional counterexamples given in arXiv:2007.13195.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

10 extracted references · 10 canonical work pages

[1]

On the loss of maximum principles for higher-order fractional Laplacians

Nicola Abatangelo, Sven Jarohs, and Alberto Saldaña. On the loss of maximum principles for higher-order fractional Laplacians. Proc. Amer. Math. Soc. , 146:4823–4835, 2018

work page 2018
[2]

Gr een function and Martin kernel for higher-order fractional Laplacians in balls

Nicola Abatangelo, Sven Jarohs, and Alberto Saldaña. Gr een function and Martin kernel for higher-order fractional Laplacians in balls. Nonlinear Analysis, 175:173–190, 2018

work page 2018
[3]

Alvino, G

A. Alvino, G. Trombetti, and P.-L. Lions. Comparison res ults for elliptic and parabolic equations via Schwarz symmetrization. Ann. Inst. H. Poincaré C Anal. Non Linéaire , 7:37– 65, 1990

work page 1990
[4]

R. M. Blumenthal, R. K. Getoor, and D B Ray. On the distribu tion of ﬁrst hits for the symmetric stable processes. Transactions of the American Mathematical Society , 99:540–554, 1961

work page 1961
[5]

Symmetrization for f ractional elliptic problems: A direct approach

Vincenzo Ferone and Bruno Volzone. Symmetrization for f ractional elliptic problems: A direct approach. Archive for Rational Mechanics and Analysis , 239:1733–1770, 2021

work page 2021
[6]

S. Kesavan. Symmetrization & Applications , volume 3 of Series in Analysis . World Scientiﬁc Publishing Co. Pte. Ltd., Hackensack, NJ, 2006

work page 2006
[7]

Lieb and Michael Loss

Elliott H. Lieb and Michael Loss. Analysis., volume 14 of Grad. Stud. Math. Providence, RI: American Mathematical Society (AMS), 2nd ed. edition, 2001

work page 2001
[8]

Nonlocal elliptic equations in bounde d domains: a survey

Xavier Ros-Oton. Nonlocal elliptic equations in bounde d domains: a survey. Publicacions Matemàtiques, 60:3–26, 2016

work page 2016

Show all 10 references

[9]

The Dirichlet proble m for the fractional Laplacian: Regularity up to the boundary

Xavier Ros-Oton and Joaquim Serra. The Dirichlet proble m for the fractional Laplacian: Regularity up to the boundary. Journal de Mathématiques Pures et Appliquées , 101:275–302, 2014

2014
[10]

Elliptic equations and Rearrangemen ts

Giorgio Talenti. Elliptic equations and Rearrangemen ts. Annali della Scuola Normale Supe- riore di Pisa - Classe di Scienze , Ser. 4, 3(4):697–718, 1976. 13 ON A FRACTIONAL BOUNDARY VERSION OF TALENTI’S INEQUALITY IN THE UNIT BALL Institut für Mathematik Goethe-Universität Fr...

1976

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [1]

On the loss of maximum principles for higher-order fractional Laplacians

Nicola Abatangelo, Sven Jarohs, and Alberto Saldaña. On the loss of maximum principles for higher-order fractional Laplacians. Proc. Amer. Math. Soc. , 146:4823–4835, 2018

work page 2018

[2] [2]

Gr een function and Martin kernel for higher-order fractional Laplacians in balls

Nicola Abatangelo, Sven Jarohs, and Alberto Saldaña. Gr een function and Martin kernel for higher-order fractional Laplacians in balls. Nonlinear Analysis, 175:173–190, 2018

work page 2018

[3] [3]

Alvino, G

A. Alvino, G. Trombetti, and P.-L. Lions. Comparison res ults for elliptic and parabolic equations via Schwarz symmetrization. Ann. Inst. H. Poincaré C Anal. Non Linéaire , 7:37– 65, 1990

work page 1990

[4] [4]

R. M. Blumenthal, R. K. Getoor, and D B Ray. On the distribu tion of ﬁrst hits for the symmetric stable processes. Transactions of the American Mathematical Society , 99:540–554, 1961

work page 1961

[5] [5]

Symmetrization for f ractional elliptic problems: A direct approach

Vincenzo Ferone and Bruno Volzone. Symmetrization for f ractional elliptic problems: A direct approach. Archive for Rational Mechanics and Analysis , 239:1733–1770, 2021

work page 2021

[6] [6]

S. Kesavan. Symmetrization & Applications , volume 3 of Series in Analysis . World Scientiﬁc Publishing Co. Pte. Ltd., Hackensack, NJ, 2006

work page 2006

[7] [7]

Lieb and Michael Loss

Elliott H. Lieb and Michael Loss. Analysis., volume 14 of Grad. Stud. Math. Providence, RI: American Mathematical Society (AMS), 2nd ed. edition, 2001

work page 2001

[8] [8]

Nonlocal elliptic equations in bounde d domains: a survey

Xavier Ros-Oton. Nonlocal elliptic equations in bounde d domains: a survey. Publicacions Matemàtiques, 60:3–26, 2016

work page 2016