REVIEW 3 major objections 4 minor 3 references
Examples of optimal H\"older regularity in semilinear equations involving the fractional Laplacian
T0 review · 3 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash
Pith's one-line read Explicit examples show that for $(-\Delta)^s u=f(u)$ with $f\in C^\beta$, the Hölder exponent $2s/(1-\beta)$ is optimal when $2s\le 1-\beta$, and $2s+\beta$ is optimal otherwise.
desk verdict Sharpness of the Hölder exponent 2s/(1-β) is proved with explicit nonperiodic examples, but the periodic version rests on an unproved interpolation. 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 engine is a corner-shaped model function. In the nonperiodic case one takes $u(x)=\operatorname{sign}(x)\,|x|^r$ on $[-1,1]$ with $r=2s/(1-\beta)$, extended to $\pm1$ outside. Lemma 2.3 computes $(-\Delta)^s u$ for such functions as $c_1x^{r-2s}+c_2x^r+c_sG(x)-c_sH(x)$, proves $G\in C^1$, and shows that $x\mapsto H(x^{1/r})$ is Lipschitz; this is exactly what is needed to define $f(t)$ from $t=x^r$ and verify $f\in C^\beta$ with $\beta=(r-2s)/r$. For the periodic version the same lemma is applied to an 8-periodic interpolant satisfying six structural properties (symmetry, monotonicity, flatness, and quadratic behaviour near the top). In the complementary regime $2s>1-\beta$ the model is $u(x)=\operatorname{sign}(x)|x|^{2s+\beta}+x$, whose inverse is Lipschitz, which makes the composition $f\circ u$ amenable to the same argument.
What would settle it
Check whether the six properties (a)-(f) in Section 3 are consistent: if one can prove that no smooth 8-periodic function can simultaneously satisfy all of them, the periodic construction fails; conversely, producing any such interpolant and verifying the Lipschitz estimate (3.1) numerically would test the proof directly.
Extended reading notes
Core claim
The central result, Theorem 1.2, is a pair of optimality examples. For every $0<s<1$, $0<\beta<1$, and every period $2L$, there is a $2L$-periodic function $u$ and a function $f\in C^\beta(\mathbb{R})$ such that $(-\Delta)^s u=f(u)$ holds on all of $\mathbb{R}$. In the regime $2s\le 1-\beta$, $u\in C^{2s/(1-\beta)}(\mathbb{R})$ yet $u\notin C^{2s/(1-\beta)+\epsilon}([-\rho,\rho])$ for any $\epsilon,\rho>0$; in the complementary regime $2s>1-\beta$, the same non-improvement holds with the classical exponent $2s+\beta$. Together with the upper bounds of Theorem 1.1, this shows the new exponent $2s/(1-\beta)$ is sharp and that the dichotomy between the two regimes is not a proof artefact. The boundary case $2s=1-\beta$ is included and gives optimal Lipschitz ($C^1$) regularity.
Load-bearing premise
The load-bearing premise is the unproved existence, in the periodic construction, of a smooth 8-periodic interpolant that is $|x|^r$ on $(-1,1)$, odd about 0, even about 2, increasing on (0,2), flatter on (1,2) than on (0,1), and quadratic near $x=2$; if no such function exists, the periodic optimality example in Theorem 1.2(i) is not established, although the nonperiodic example is explicit and does not rely on it.
Editorial extensions
If this is right
- The regularity estimate $C^{2s/(1-\beta)-\epsilon}$ from Theorem 1.1(i) is optimal: no general theorem can replace the exponent by anything larger when $2s\le 1-\beta$.
- In the complementary case $2s>1-\beta$, the classical exponent $2s+\beta$ is also optimal, so the full dichotomy in Theorem 1.1 is sharp.
- At the boundary $2s=1-\beta$, the sharp regularity is $C^1$ (Lipschitz), the same value the linear theory would predict.
- Because the examples are periodic, they show optimality for the whole-line, periodic, and Dirichlet-type settings covered by the earlier theorem.
Reading between the lines
- The same corner construction likely extends to radially symmetric solutions in higher dimensions, where the fractional Laplacian of a radial profile has a similar one-dimensional kernel near the origin; the exponent $2s/(1-\beta)$ would then control the radial Hölder regularity at the origin.
- A natural stress test is $\beta=0$, i.e. continuous nonlinearities: the formula predicts $C^{2s}$ as the limiting regularity, but the present proof requires $\beta>0$ for the composition law, so whether the phenomenon survives for merely continuous $f$ is left open.
- The explicit nonperiodic example could serve as a numerical benchmark: discretizing $(-\Delta)^s$ on the corner function with fixed $s,\beta$ should reproduce the predicted lack of $C^{r+\epsilon}$ regularity at zero, giving a cheap computational check of the theory.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper studies Hölder regularity of solutions to the one-dimensional semilinear fractional Laplacian equation (-Δ)^s u = f(u) with f ∈ C^β. It constructs examples showing that, for 2s ≤ 1-β, the regularity exponent 2s/(1-β) from the upper bound in Theorem 1.1 cannot be improved, and for 2s > 1-β that the exponent 2s+β is sharp. In the nonperiodic case the construction is explicit: for 2s < 1-β the authors define u(x) = sign(x)|x|^r, truncated outside [-1,1], with r = 2s/(1-β), and they compute (-Δ)^s u in closed form in Lemma 2.3, then define f by f(u(x)) = (-Δ)^s u(x) and prove f ∈ C^β. The case 2s = 1-β is handled separately in Lemma 2.5. For the periodic case, the paper announces an 8-periodic analogue in Proposition 3.1 by postulating a smooth interpolant satisfying six properties (a)-(f), and Lemma 3.2 is used to establish the needed Lipschitz property of H(t^{1/r}). For 2s > 1-β, the nonperiodic example is u(x) = sign(x)|x|^{2s+β} + x (truncated), and the periodic case is only outlined. The main theorem, Theorem 1.2, states both cases in the periodic setting.
Significance. If fully established, Theorem 1.2 would be a valuable contribution: it proves that the unusual exponent 2s/(1-β) discovered in Theorem 1.1 is not an artifact of the iteration argument but a genuine regularity threshold for semilinear nonlocal equations. The nonperiodic construction is explicit and transparent, and Lemma 2.3 gives a precise formula for the fractional Laplacian of the power-like function, which is a solid basis for the f ∈ C^β conclusion. The paper also includes a simple local example in Appendix A that motivates the nonlocal construction. The main weakness is that the periodic optimality claims depend on an unproved existence assertion for an interpolant satisfying properties (a)-(f) in Section 3, and the proof of Lemma 3.2 is sketchy at a load-bearing point. The nonperiodic results, however, are not affected by this gap.
major comments (3)
- [Section 3, Proposition 3.1(a)-(f)] The proof of Proposition 3.1 postulates the existence of an 8-periodic function u satisfying all six properties (a)-(f), including the flatness condition (e) and the quadratic behavior near x=2 in (f), but no construction or existence proof is given. Lemma 3.2 uses property (e) essentially in the A_k^1 estimate leading to (3.3), and the proof of Proposition 3.1 uses property (f) to prove that f is Lipschitz near u(2). Since every step of the periodic optimality argument depends on this interpolant, Theorem 1.2(i) is conditional on an unproved existence assertion that could hide an incompatibility among the six properties. Please supply an explicit interpolant or a rigorous existence proof.
- [Lemma 3.2, A_k^1 estimate and Eq. (3.3)] The proof that |u(x^{1/r}+y) - u(z^{1/r}+y)| ≤ |x-z| for y ∈ A_k^1 is the cornerstone of the Lipschitz estimate for H_+, but the argument is only sketched. In the first case the sentence 'using (3.4) ... one easily obtains (3.3)' omits the verification when the points x^{1/r}+y and z^{1/r}+y lie in different monotonicity regions, and in the second case the reflection argument is described geometrically without a complete algebraic check. Since Lemma 3.2 is needed for the periodic construction, please provide a fully detailed proof of (3.3).
- [Section 4, outline of Theorem 1.2(ii)] The periodic u for the case 2s > 1-β is said to coincide with (4.1) on [-1,1] and to be smooth on R \ 4Z, but (4.1) has a corner at x = ±1 when 2s+β > 1: the derivative from the left at x=1 is 2s+β+1, while the derivative from the right is 0. This contradicts smoothness at x=1, which is an interior point of (0,4). The construction therefore needs modification near x=±1, and the required interpolation properties (with a proof of existence) should be stated explicitly, as in the previous comment.
minor comments (4)
- [Throughout] There are several typographical issues in the TeX source, such as 'H¨older' and 'flatter', which should be corrected in the final version.
- [Lemma 2.5] The proof defines f only on [0,1/2] and then concludes the equation for all x ∈ [-1/2,1/2] using that u is even; this is correct because u([-1/2,0]) ⊂ [0,1/2], but the extension of f to the full real line by zero for negative arguments should be stated explicitly.
- [Proof of Proposition 3.1] The sentence 'using the smoothness of g and a Taylor expansion we get g'(x) = O(|x-2|)' should specify the neighborhood and the constant, since property (f) is used only locally near x=2.
- [Appendix A] The claim that the function t ↦ 1 - t^β - ((y^{1+β}+1) - t(x^{1+β}+1))^β is negative 'due to convexity' would benefit from a one-line justification, as it is not immediate.
Circularity Check
No circularity: the sharpness examples are self-contained constructions with f defined from u, and the only self-citation supplies context, not input.
full rationale
The paper's derivation chain is not circular. In the nonperiodic construction (Section 2), the function u is chosen first as an explicit power function (2.1), and f is then defined by the identity f(u(x)) = (-Delta)^s u(x). The proof that such an f lies in C^beta is a genuine regularity estimate: Lemma 2.3 computes (-Delta)^s u explicitly, and Lemma 2.4 shows that the composed function f(t) has the claimed Holder exponent because the relation beta = (r - 2s)/r is exactly the algebra of the computed leading term. This is an identity satisfied by the constructed example, not a parameter fitted to a target regularity claim, and it does not presuppose the conclusion. The periodic construction in Section 3 follows the same scheme: it postulates an interpolant u with properties (a)-(f), verifies the needed Lipschitz estimate for H in Lemma 3.2, and again defines f by f(u(x)) = (-Delta)^s u(x). The existence of the interpolant is asserted rather than explicitly constructed, which is a completeness gap in the manuscript but not a circularity, since the asserted properties do not include the conclusion that f is in C^beta or that u has the sharp Holder exponent. The only self-citation is to the authors' prior paper [2] for the upper-bound theorem 1.1, which is used as context for what is being shown sharp; the examples are built independently of that theorem and do not invoke it in constructing u or f. No fitted input is renamed as a prediction, no uniqueness theorem is imported from the authors, and no known result is merely relabeled. Thus the appropriate circularity score is 0.
Assumptions & free parameters
assumptions (2)
- ad hoc to paper Existence of an 8-periodic function u, smooth on R \ 4Z, satisfying properties (a)-(f) in Section 3.
- standard math Standard fractional Laplacian calculus: the P.V. definition (1.1), the scaling identity (-Δ)^s(u(λ·))(x)=λ^{2s}(-Δ)^s u(λx), and the Taylor expansion (2.5) for the kernel.
Cite this review
Pith. "Pith review of Examples of optimal H\"older regularity in semilinear equations involving the fractional Laplacian." pith.science (2026). https://pith.science/paper/3ZKXN722
@misc{pith2026241202762,
author = {Pith},
title = {Pith review of: Examples of optimal H\"older regularity in semilinear equations involving the fractional Laplacian},
year = {2026},
howpublished = {\url{https://pith.science/paper/3ZKXN722}},
note = {Machine review of arXiv:2412.02762}
}
abstract
We discuss the H\"older regularity of solutions to the semilinear equation involving the fractional Laplacian $(-\Delta)^s u=f(u)$ in one dimension. We put in evidence a new regularity phenomenon which is a combined effect of the nonlocality and the semilinearity of the equation, since it does not happen neither for local semilinear equations, nor for nonlocal linear equations. Namely, for nonlinearities $f$ in $C^\beta$ and when $2s+\beta <1$, the solution is not always $C^{2s+\beta-\epsilon}$ for all $\epsilon >0$. Instead, in general the solution $u$ is at most $C^{2s/(1-\beta)}.$
Reference graph
Works this paper leans on
-
[2]
X. Cabr´ e, G. Csat´ o, A. Mas,Periodic solutions to integro-differential equations: var iational formula- tion, symmetry, and regularity , arXiv:2404.06462
-
[1]
Bass, Regularity results for stable-like operators , J
R. Bass, Regularity results for stable-like operators , J. Funct. Anal. 257 (2009), no. 8, 2693–2722
work page 2009
-
[3]
H. Shahgholian, Regularity issues for semilinear PDE-s (a narrative approa ch), Algebra i Analiz 27 (2015), no. 3, 311–325. 18 G. CSAT ´O AND A. MAS G. Csat ´o 1,2 1 Departament de Matem`atiques i Inform `atica, Universitat de Barcelona, Gran Via 585, 08007 Barcelona, Spain 2 Centre de Recerca Matem `atica, Edifici C, Campus Bellaterra, 08193 Bellaterra, ...
work page 2015
Reviewed August 11, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.