Boundary regularity for nonlocal elliptic equations over Reifenberg flat domains

Adriano Prade

REVIEW 2 major objections 4 minor 14 references

Boundary regularity for nonlocal elliptic equations over Reifenberg flat domains

T0 review · 2 major / 4 minor · reviewed 2026-08-08 · deepseek-v4-flash

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

Pith's one-line read This paper proves that weak solutions to nonlocal elliptic equations with kernels comparable to the fractional Laplacian are $C^{s-\varepsilon}$ up to the boundary on Reifenberg flat domains whose flatness parameter is small enough.

desk verdict A genuine, carefully written generalization of nonlocal boundary regularity to Reifenberg flat domains; the main barrier argument is sound, but the compactness upgrade leans on an unverified black-box Liouville theorem and a few constants need cleaning. read the letter →

arxiv 2502.04107 v2 pith:PZ7Y2P3M submitted 2025-02-06 math.AP

Adriano Prade This is my paper · ORCID

classification math.AP MSC 35R1135B6535D30

keywords fractionalLaplaciannonlocalellipticequationsboundaryregularityReifenbergflatdomainsHölderbarriermethodcompactnessargumentweakcomparisonprinciple

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

This paper proves that weak solutions to nonlocal elliptic equations driven by operators comparable to the fractional Laplacian are Hölder continuous up to the boundary with the near-optimal exponent $s-\varepsilon$, provided the domain is Reifenberg flat with small flatness parameter and the solution vanishes outside the domain. This matters because Reifenberg flat sets include boundaries with fractal behavior, much less smooth than the Lipschitz or $C^{1,\alpha}$ domains covered before, and the conclusion matches the sharp regularity known for the fractional Laplacian itself. The proof first establishes a small Hölder exponent $\sigma$ by an induction with explicit barriers, then upgrades to $C^{s-\varepsilon}$ through a compactness argument and a half-space Liouville theorem. The estimate is linear in the $L^\infty$ norm of the right-hand side, so the result is quantitative and applies with arbitrary $L^\infty$ data.

What carries the argument

The argument is carried by two mechanisms. First, a family of barriers $v_k$ built from powers of the regularized distance $\delta^\varepsilon$ to the complement of a shrinking ball: Lemma 3.4 shows $L\delta^\varepsilon \geq c\,d^{\varepsilon-2s}$ near the boundary for the full class of comparable kernels, and an induction using the weak comparison principle turns this supersolution growth into a $C^\sigma$ estimate at every scale, exploiting Reifenberg flatness only to align the barrier with an approximating hyperplane. Second, Theorem 4.3 upgrades $\sigma$ to $s-\gamma$ by contradiction: rescaling around a boundary point where the claimed bound fails produces a blow-up sequence that converges, via stability of distributional solutions, to a solution of a limiting nonlocal equation in a half-space; the half-space Liouville theorem then forces the limit to vanish while the normalization forces it to be nonzero. The regularized distance bounds (2.1)-(2.2), the measure estimate of Lemma 3.1, and the scaling property of the operator class are the auxiliary identities that make both steps quantitative.

What would settle it

Solve, numerically or analytically, the Dirichlet problem for an operator of the form (1.1)-(1.2) with a non-homogeneous kernel, for example $K(y)=a(y/|y|)|y|^{-n-2s}$ with a nonconstant angular factor, on a Reifenberg-flat domain built with a self-similar boundary, and measure the boundary growth of $u$. If for some arbitrarily small flatness $\eta$ the maximum of $u$ in $B_r$ decays slower than $r^{s-\varepsilon}$ for some $\varepsilon>0$, say like $r^s\log(1/r)$, then Theorem 1.2 is false. Reproducing the predicted $C^{s-\varepsilon}$ bound at several scales would confirm it.

Watch

Extended reading notes

Core claim

The central assertion is Theorem 1.2: for any $s\in(0,1)$, any $\varepsilon\in(0,s)$, and any operator $L$ with kernel $K$ satisfying $\lambda|y|^{-n-2s}\leq K(y)\leq \Lambda|y|^{-n-2s}$, if $\Omega$ is $(\eta,r_0)$-Reifenberg flat with $\eta$ no larger than a constant $\eta_0(n,s,\lambda,\Lambda,\varepsilon)$, then the weak solution of $Lu=f$ in $\Omega$, $u=0$ outside $\Omega$, belongs to $C^{s-\varepsilon}(\Omega)$ with norm at most $C(n,s,\lambda,\Lambda,\varepsilon)\lVert f\rVert_{L^\infty(\Omega)}$. The regularity is sharp in the sense that for kernels of order $2s$ one cannot expect more than $C^{s-\varepsilon}$ in this generality, matching the $\delta$-Lipschitz result in [14] that this paper extends to the less regular Reifenberg-flat framework. A distinguishing feature is that the condition on $\Omega$ is purely geometric and scale-invariant: at every boundary point and every small scale the boundary is uniformly close to a hyperplane, so the conclusion holds uniformly across possibly fractal boundaries.

Load-bearing premise

The argument's load-bearing premise is that the half-space Liouville theorem imported from [14]—that the only distributional solution of a limiting comparable nonlocal equation in a half-space with zero exterior data and growth at most $(1+|x|)^{s-\gamma}$ is zero—holds for the full class of kernels allowed by (1.2); together with the stability and blow-up selection tools from [4], it is what turns the compactness limit into a contradiction.

Editorial extensions

If this is right

The $C^{s-\varepsilon}$ boundary estimate holds for all operators comparable to the fractional Laplacian, not only for the fractional Laplacian itself, and for arbitrary $L^\infty$ right-hand sides.
The flatness threshold $\eta_0$ depends only on $n$, $s$, $\lambda$, $\Lambda$, and $\varepsilon$, so the regularity is uniform over the whole Reifenberg-flat class once $\eta\leq\eta_0$.
Combining the local boundary estimate with interior $C^{2s}$ estimates gives the global Hölder norm on $\Omega$, and the same conclusion extends to continuous distributional solutions by Remark 4.4.
The result is optimal in the exponent scale: $C^{s-\varepsilon}$ for every $\varepsilon>0$ is the best one can expect in this setting, consistent with the known optimality for $\delta$-Lipschitz domains.
The theorem reduces the regularity theory over Reifenberg-flat sets to the same quantitative statement known for much smoother boundaries, so existing applications relying on boundary Hölder bounds can be transferred to fractal domains.

Reading between the lines

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

A natural next step would be to test whether the flatness threshold can be made quantitative in $r_0$ or whether $\eta_0$ has an explicit power-type dependence on $\varepsilon$; the paper only states the existence of $\eta_0$ and $C$.
The same barrier-and-compactness scheme may adapt to operators with kernels that are only comparable at small scales or to systems, since the barriers use only the pointwise ellipticity bounds and scaling.
Because the blow-up step uses the half-space Liouville theorem as a black box, replacing it with a self-contained proof for the full kernel class would remove the most fragile imported ingredient and might yield the same conclusion for more general translation-invariant limits.
One could look for a constructive example at the threshold: a Reifenberg-flat domain with flatness exactly $\eta_0$ where the boundary behavior is no better than $|x|^{s-\varepsilon}$, which would locate the sharpness of the condition.

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

Short version: this is a real result and it deserves a serious referee. The paper proves sharp C^{s-epsilon} boundary regularity for operators comparable to the fractional Laplacian over Reifenberg flat domains with L^infinity right-hand side, which genuinely extends earlier work: [14] covered delta-Lipschitz domains, and [13] was restricted to the fractional Laplacian with continuous RHS. That is a natural and worthwhile completion of the picture.

The best part of the paper is Section 3 and Theorem 4.1. The barrier construction via the regularized distance, the measure estimate, and the induction argument are adapted carefully from [4] and [12]-[13], with the right modifications for the weaker Reifenberg geometry. The proof that the barrier is admissible in the weak form (the finite-energy check) is done conscientiously. I read the main chain as likely correct.

The soft spots are the ones the reader flagged, and they are real but not fatal. First, in the proof of Theorem 4.1, the passage from the dyadic estimate (4.7) to the pointwise Holder bound (4.6) as written gives constant 1 instead of the usual constant rho^{-sigma}. That is an easy fix, but the text should make it. Second, the role of eta is compressed at the end of the induction: setting eta = rho after having chosen rho small is legitimate only if rho is chosen after eta0 is fixed, and the exposition should say so explicitly. Both are minor.

The more serious concern is Theorem 4.3. The compactness upgrade is standard in outline, but it relies on the half-space Liouville theorem [14, Theorem 6.2] without reproducing its hypotheses or checking them against the blow-up limit v. The blow-up objects are distributional solutions with polynomial growth and the limiting operator is only known to lie in L_n^s(lambda, Lambda); if [14, Theorem 6.2] requires energy solutions, extra kernel regularity, or a different growth bound, the contradiction step collapses. This is a load-bearing external dependency, and citing it by number is not enough in a paper whose main novelty is precisely this upgrade. I would not desk-reject over this, but I would insist that the referee or the author verify the matching.

Who is this for? Researchers working on boundary regularity for nonlocal operators. It is a solid contribution, not a breakthrough, and the proof is mostly a competent adaptation of known techniques. The citation pattern is fine; self-citation is limited and appropriate.

Recommendation: send it to peer review. A good referee should check [14, Thm. 6.2] against the blow-up objects and ask for the small constant/eta clean-ups. If the Liouville theorem applies, the paper should be accepted after minor revision.

Referee Report

2 major / 4 minor

Summary. This paper proves boundary C^{s-ε} regularity for solutions of Lu = f in Ω, u = 0 outside Ω, where L is a symmetric nonlocal operator of order 2s with kernel comparable to the fractional Laplacian and Ω is an (η,r0)-Reifenberg flat domain with η sufficiently small (Theorem 1.2). The proof is in two stages. Section 3 constructs supersolutions δ^ε via the regularized distance and a measure-theoretic density lemma, yielding the barrier estimate Lδ^ε ≥ c d^{ε-2s} (Lemma 3.4). Theorem 4.1 then uses an induction with these barriers and the comparison principle to obtain C^σ boundary regularity. Theorem 4.3 upgrades this to C^{s-γ} by a compactness argument: assuming the sharp estimate fails, it produces a normalized blow-up sequence, passes to a limiting solution v in a half-space, and invokes a half-space Liouville theorem from [14] to rule out nonzero limits.

Significance. If the proof is completed, this is a natural and valuable result: it extends optimal boundary regularity for operators comparable to the fractional Laplacian from C^1/Lipschitz boundaries to Reifenberg flat domains, with the sharp s-ε exponent and L∞ right-hand sides. The barrier construction is largely self-contained and the dyadic induction in Theorem 4.1 is transparent and does not rely on the main theorem circularly. The main weakness is the final compactness step, which outsources the only exclusion of nontrivial blow-up profiles to an external Liouville theorem whose hypotheses are not verified against the objects produced by the blow-up argument.

major comments (2)

[Theorem 4.3, after Eq. (4.21)] The statement 'By Liouville theorem in the half-space [14, Theorem 6.2] and the second property in (4.21) we have v = 0' is the only step that excludes a nonzero blow-up profile; estimates (4.16) and (4.15) both depend on it. The hypotheses of [14, Theorem 6.2] are not reproduced, and the manuscript does not check that the objects produced here satisfy them. In particular, [4, Proposition 2.2.36] yields L∞ only as a limit in the measurable class L_n^s(λ,Λ), and v is only constructed as a distributional solution with growth |v| ≤ 2(1+|x|^{s-γ}); it is not shown that [14, Theorem 6.2] covers distributional solutions, non-homogeneous kernels of class L_n^s(λ,Λ), or this growth bound. Since the compactness argument by itself cannot add kernel or solution regularity, the contradiction step is unsecured as written. Please either quote [14, Theorem 6.2] in full and verify these points, or replace it by a self-contained half-space Liouville argument.
[Theorem 4.3, Eqs. (4.18)-(4.21)] The compactness chain also imports several black-box conclusions without stating them: the existence of scales r_k and the uniform growth bound (4.19) are taken from [4, Lemma 4.4.11], and the existence of the limiting operator L∞ is taken from [4, Proposition 2.2.36]. These results are load-bearing because (4.19) is needed both for the local uniform convergence and for the growth hypothesis of the Liouville theorem. The manuscript should state exactly what these external results deliver and confirm that the normalized sequence v_k satisfies the hypotheses of the stability result after the truncation/normalization performed at the beginning of the proof.

minor comments (4)

[Section 4.1, Eqs. (4.6)-(4.7)] The implication from (4.7) to (4.6) is not literal: the dyadic bound with constant one gives |ũ(x)| ≤ ρ^{-σ}|x|^σ when ρ^{k+1} ≤ |x| ≤ ρ^k. The missing factor ρ^{-σ} is harmless for the final theorem because constants may be renamed, but the displayed estimate in (4.6) should either carry the constant or be rephrased as a growth estimate with a universal constant.
[Lemma 3.2 and Lemma 3.4] The sentence 'up to decreasing κ we may assume R ≫ D without loss of generality' is not justified from assumption (P_{R,κ}), which is only assumed for 0 < r < R; decreasing κ does not extend the range of r. The argument is repairable because the only application in Section 4 uses complements of balls, for which the property holds with arbitrarily large R, but the lemmas as stated need a precise hypothesis such as R large relative to diam(Ω^c) or an explicit statement of the stronger property used in the application.
[Theorem 4.3, statement] The theorem states the conclusion for every γ > 0 and writes C^{s-γ}(B_{1/2}); this only makes sense for γ < s. The statement should be restricted to γ ∈ (0,s), which is also what Theorem 1.2 needs.
[General presentation] There are several small typos and imprecisions: 're lies' in the abstract, 'tha t that' in the proof of Theorem 4.3, 'the same result h olds' in Section 2, and the phrase 'uniformly bounded in C^σ(Ω_k ∩ B_{1/(2r_k)})' in Theorem 4.3 is ambiguous and should read Ω_k/r_k ∩ B_{1/(2r_k)} or be phrased locally on compact subsets of the limiting half-space. The integral limits in the second case of Lemma 3.2 involving '100D' should also be double-checked against the size of R.

Circularity Check

0 steps flagged · score 1.0 of 10

No substantive circularity: the C^{s-ε} result is derived via an independent barrier induction and compactness argument; the only self-citation is contextual and not load-bearing.

full rationale

The derivation chain for Theorem 1.2 proceeds through Theorem 4.1 (barrier construction plus induction, using Lemma 3.4 and the weak comparison principle) and Theorem 4.3 (compactness blow-up). The main external inputs are [14, Theorem 6.2] (half-space Liouville theorem), [4, Proposition 2.2.36] (stability of distributional solutions), and [4, Lemma 4.4.11] (blow-up selection). These are independent prior works, not results of the present paper, and none of them is a restatement of the target boundary-regularity estimate. The half-space Liouville theorem is used only to rule out nontrivial blow-up profiles, not to assume the desired C^{s-γ} bound. The one self-citation, [13], appears only as background: the introduction says Theorem 1.2 'completes the picture initiated in [13]', and the outline says the induction 'follow[s] the ideas of [12,13]'. No technical step in Sections 3 or 4 invokes [13] as a premise. The barrier estimates are proved directly from the kernel class (1.2), the regularized distance properties (2.1)-(2.2), and the measure estimate Lemma 3.1; they do not presuppose the Hölder exponent being proved. The compactness step does rely on black-box hypotheses from [14, Theorem 6.2], and the paper does not verify those hypotheses in detail against the blow-up objects; however, a missing verification is a correctness or completeness risk, not circularity, because the external theorem is not derived from the paper's conclusion and no equation reduces to its own input. There is no fitted parameter relabeled as a prediction and no load-bearing self-citation chain.

Assumptions & free parameters 0 free parameters · 7 assumptions · 0 invented entities

The theorem is proved from a handful of external results, all standard in the theory of integro-differential equations but not proved in the note: the weak maximum principle, interior C^{2s} regularity, L∞ bounds, regularized distance estimates, stability of distributional solutions, a blow-up selection lemma, and a half-space Liouville theorem. These are assumptions the reader is asked to take from [4] and [14]. No free parameters are fitted and no new entities are postulated.

assumptions (7)

standard math Weak maximum principle for weak and continuous distributional solutions (Theorem 2.4 and [4, Lemma 2.3.5])
Used in Theorem 4.1 to compare u with barriers via the comparison principle.
standard math Interior C^{2s} (or C^{1-ε} if s=1/2) regularity for distributional solutions (Theorem 2.5)
Used to combine boundary decay with interior estimates and to justify continuity inside Ω.
standard math Existence and L∞ bound for weak solutions to the Dirichlet problem ([4, Theorem 2.2.24, Lemma 2.3.9])
Used in the proof of Theorem 1.2 to pass from local to global estimates and to obtain (1.5).
standard math Regularized distance function δ with comparable bounds and derivative estimates (2.1)-(2.2), [4, Lemma B.0.1]
The barriers in Lemma 3.4 are built from powers of this distance.
standard math Stability of distributional solutions under kernel convergence ([4, Proposition 2.2.36])
Used in Theorem 4.3 to pass to the limit and obtain the half-space equation.
standard math Half-space Liouville theorem for operators in the class L_n^s(λ,Λ) ([14, Theorem 6.2])
Used in Theorem 4.3 to force the blow-up limit to be zero, giving the contradiction.
standard math Blow-up selection lemma ([4, Lemma 4.4.11])
Used in Theorem 4.3 to extract scales r_k with ‖u_k‖_{L∞(B_{r_k})} ≥ C r_k^{s-γ}.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Boundary regularity for nonlocal elliptic equations over Reifenberg flat domains." pith.science (2026). https://pith.science/paper/PZ7Y2P3M

@misc{pith2026250204107,
  author       = {Pith},
  title        = {Pith review of: Boundary regularity for nonlocal elliptic equations over Reifenberg flat domains},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PZ7Y2P3M}},
  note         = {Machine review of arXiv:2502.04107}
}

abstract

We prove sharp boundary regularity of solutions to nonlocal elliptic equations arising from operators comparable to the fractional Laplacian over Reifenberg flat sets and with null exterior condition. More precisely, if the operator has order $2s$ then the solution is $C^{s-\varepsilon}$ regular for all $\varepsilon>0$ provided the flatness parameter is small enough. The proof relies on an induction argument and its main ingredients are the construction of a suitable barrier and the comparison principle.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

14 extracted references · 11 canonical work pages

[14]

Optimal regularity for nonlocal elliptic equations and fre e boundary problems

Ros-Oton, X., and Weidner, M. Optimal regularity for nonlocal elliptic equations and fre e boundary problems. arXiv preprint arXiv:2403.07793 (2024). CMAP, ´Ecole Polytechnique, Institut Polytechnique de P aris, 911 20 P alaiseau, France. Email address : adriano.prade@polytechnique.edu

arXiv 2024
[13]

Boundary H\"older regularity for the fractional Laplacian over Reifenberg flat domains via ABP maximum principle

Prade, A. Boundary H¨ older regularity for the fractional Laplacian o ver Reifenberg ﬂat domains via ABP max- imum principle. arXiv preprint arXiv:2501.14639 (2025)

work page Pith review arXiv 2025
[4]

Integro-diﬀerential elliptic equations

Fern´andez-Real, X., and Ros-Oton, X. Integro-diﬀerential elliptic equations. Progress in Mathematics, Birkh¨ auser Cham(2024)

work page 2024
[12]

Boundary H\"older Regularity for Elliptic Equations on Reifenberg Flat Domains

Lian, Y., and Zhang, K. Boundary H¨ older regularity for elliptic equations on Reif enberg ﬂat domains. arXiv preprint arXiv:1812.11354 (2018)

work page Pith review arXiv 2018
[1]

Nonlocal diﬀusion and applications , vol

Bucur, C., and V aldinoci, E. Nonlocal diﬀusion and applications , vol. 20. Springer, 2016

work page 2016
[2]

Regularity results for nonlocal equations by approximatio n

Caffarelli, L., and Sil vestre, L. Regularity results for nonlocal equations by approximatio n. Archive for rational mechanics and analysis 200 (2011), 59–88

work page 2011
[3]

Higher order elliptic and parabolic systems with variably p artially BMO coeﬃcients in regular and irregular domains

Dong, H., and Kim, D. Higher order elliptic and parabolic systems with variably p artially BMO coeﬃcients in regular and irregular domains. Journal of Functional Analysis 261 , 11 (2011), 3279–3327

work page 2011
[5]

Fractional thoughts

Garofalo, N. Fractional thoughts. arXiv preprint arXiv:1712.03347 (2017). A. PRADE 15

arXiv 2017

Show all 14 references

[6]

Rigidity of the ball for an isoperimetric problem with stron g capacitary repulsion

Goldman, M., Novaga, M., and Ruffini, B. Rigidity of the ball for an isoperimetric problem with stron g capacitary repulsion. Journal of the European Mathematical Society (2024)

2024
[7]

Boundary regularity and Hopf lemma for nondegenerate stabl e operators

Grube, F. Boundary regularity and Hopf lemma for nondegenerate stabl e operators. arXiv preprint arXiv:2410.00829 (2024)

2024 arXiv
[8]

The obstacle problem for nonlinear integro-diﬀerential op era- tors

Korvenp¨a¨a, J., Kuusi, T., and P alatucci, G. The obstacle problem for nonlinear integro-diﬀerential op era- tors. Calculus of Variations and Partial Diﬀerential Equations 55 , 3 (2016), 63

2016
[9]

Lemenant, A., Milakis, E., and Spinolo, L. V. On the extension property of Reifenberg-ﬂat domains. Annales Academiæ Scientiarum Fennicæ Math. 39 , 1 (2014), 51–71

2014
[10]

Boundary regularity for the polyharmonic Dirichlet proble m

Lemenant, A., and Mougenot, R. Boundary regularity for the polyharmonic Dirichlet proble m. arXiv preprint arXiv:2502.02964 (2025)

2025 arXiv
[11]

Boundary regularity for the Poisson equation in Reifenberg -ﬂat domains

Lemenant, A., and Sire, Y. Boundary regularity for the Poisson equation in Reifenberg -ﬂat domains. In Geometric Partial Diﬀerential Equations proceedings (2013), Springer, pp. 189–209

2013

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.

[1] [14]

Optimal regularity for nonlocal elliptic equations and fre e boundary problems

Ros-Oton, X., and Weidner, M. Optimal regularity for nonlocal elliptic equations and fre e boundary problems. arXiv preprint arXiv:2403.07793 (2024). CMAP, ´Ecole Polytechnique, Institut Polytechnique de P aris, 911 20 P alaiseau, France. Email address : adriano.prade@polytechnique.edu

arXiv 2024

[2] [13]

Boundary H\"older regularity for the fractional Laplacian over Reifenberg flat domains via ABP maximum principle

Prade, A. Boundary H¨ older regularity for the fractional Laplacian o ver Reifenberg ﬂat domains via ABP max- imum principle. arXiv preprint arXiv:2501.14639 (2025)

work page Pith review arXiv 2025

[3] [4]

Integro-diﬀerential elliptic equations

Fern´andez-Real, X., and Ros-Oton, X. Integro-diﬀerential elliptic equations. Progress in Mathematics, Birkh¨ auser Cham(2024)

work page 2024

[4] [12]

Boundary H\"older Regularity for Elliptic Equations on Reifenberg Flat Domains

Lian, Y., and Zhang, K. Boundary H¨ older regularity for elliptic equations on Reif enberg ﬂat domains. arXiv preprint arXiv:1812.11354 (2018)

work page Pith review arXiv 2018

[5] [1]

Nonlocal diﬀusion and applications , vol

Bucur, C., and V aldinoci, E. Nonlocal diﬀusion and applications , vol. 20. Springer, 2016

work page 2016

[6] [2]

Regularity results for nonlocal equations by approximatio n

Caffarelli, L., and Sil vestre, L. Regularity results for nonlocal equations by approximatio n. Archive for rational mechanics and analysis 200 (2011), 59–88

work page 2011

[7] [3]

Higher order elliptic and parabolic systems with variably p artially BMO coeﬃcients in regular and irregular domains

Dong, H., and Kim, D. Higher order elliptic and parabolic systems with variably p artially BMO coeﬃcients in regular and irregular domains. Journal of Functional Analysis 261 , 11 (2011), 3279–3327

work page 2011

[8] [5]

Fractional thoughts

Garofalo, N. Fractional thoughts. arXiv preprint arXiv:1712.03347 (2017). A. PRADE 15

arXiv 2017