{"id":"f868ff1e-a058-4bcf-897e-147f051d4d56","arxiv_id":"2412.06318","paper_version":3,"verdict":"REJECT","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper claims that for small values of the fractional perimeter parameter s, the only stable s-minimal cones in R^2 are half-planes, but a key integral estimate in the proof is false.","lead":"For very small values of the fractional perimeter parameter, the paper claims the only stable cone-shaped surfaces in the plane are half-planes. The proof relies on a sharp Hardy inequality, but a central integral estimate contains a scaling error.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 3.1 uses a false exponent: the ray-to-ray integral is ~(1-cos θ)^{-(1+s)/2}, not ~(1-cos θ)^{-(1+s)}; this invalidates the proof of Theorem 1.3.","rationale":"The reader's verdict identifies precisely the same load-bearing flaw: Lemma 3.1 asserts a lower bound with exponent -(1+s) in (1-cos θ), but the true asymptotic of the integral is -(1+s)/2. I re-derived the integral: for small θ the dominant contribution comes from t≈1, where the denominator behaves like (t-1)^2 + 2(1-cos θ); scaling t-1 by sqrt(1-cos θ) yields (1-cos θ)^{-(1+s)/2}. This directly disproves Lemma 3.1. The main theorem's two key consequences depend on the false exponent: the Claim that some j satisfies the 1/100 bound would need re-evaluation, and the final step forcing θ_j^{j+2}=π uses 1/(1-cos θ)^s ≤ 1/100, which only arises from the erroneous exponent. With the corrected exponent, nearby rays produce small terms and the contradiction disappears. Hence the central classification theorem is unproven. No other part of the paper independently establishes the main claim; the Appendix's proof that the cross is unstable for small s is a useful but much weaker result. I therefore agree with the REJECT verdict: the main result rests on a demonstrably false estimate, and the proof as written is invalid. A numerical check of the integral would settle the matter conclusively, but the analytical scaling argument is already decisive.","tokens_in":16819,"tokens_out":4555,"duration_ms":45470,"concrete_test":"Compute numerically I(θ) = ∫_0^∞ dt/(1+t^2-2t cos θ)^{(2+s)/2} for s=0.1 and θ=10^{-2}, 10^{-3}, 10^{-4}, e.g. by adaptive quadrature. Compare log I(θ) with log(1-cos θ) for these values: the slope should approach -(1+s)/2 ≈ -0.55, while Lemma 3.1 predicts slope -(1+s) = -1.1. If the slope is -0.55, Lemma 3.1 is false and the proof of Theorem 1.3 collapses.","verdict_should_be":"REJECT","load_bearing_attack":"The central proof of Theorem 1.3 rests on Lemma 3.1. For x on ray Σ_j, the exact integral over Σ_i is I(θ) = |x|^{-(1+s)} ∫_0^∞ dt/(1+t^2-2t cos θ)^{(2+s)/2}. The paper claims I(θ) ≥ c/(1-cos θ)^{1+s}. This is false for small θ. Writing a = 1-cos θ and u = t-1, the integrand near u=0 is (u^2 + 2a + O(a|u|))^{-(2+s)/2}. Setting u = √a v gives the leading contribution a^{-(1+s)/2} ∫ dv/(v^2+2)^{(2+s)/2} = O((1-cos θ)^{-(1+s)/2}). Thus the true asymptotic is (1-cos θ)^{-(1+s)/2}, not (1-cos θ)^{-(1+s)}; the paper's bound is larger by a factor (1-cos θ)^{-(1+s)/2}, which diverges as θ→0. The proof of Lemma 3.1 makes the invalid substitution that replaces (t^2+3a) by (t^2+3) after pulling out a^{-(1+s)}, ignoring the t^2-vs-a balance. With the correct exponent, the terms entering (8) become (1-(-1)^{i+j}cos θ)/(1-cos θ)^{(1+s)/2}; for nearly parallel rays this is ~(1-cos θ)^{(1-s)/2} → 0, so the subsequent Claim (9) and the final contradiction using 1/(1-cos θ_{j}^{j+2})^s ≤ 1/100 no longer follow. The argument no longer forces N=1, and cones with many nearly parallel rays are not excluded. The Appendix's cross-instability proof is independent and appears sound, but it is far weaker than the stated classification.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims that for sufficiently small s in (0,1), the only s-minimal cones in R^2 that are stable in R^2 \\ {0} are half-planes (Theorem 1.3), and it derives from this a classification of finite-Morse-index cones (Corollary 1.4). The strategy is to use a BV-estimate for small s to show the cone has finitely many rays, then to use the second variation formula with a radial test function that nearly saturates Hardy's inequality on (0,∞). A key quantitative lemma (Lemma 3.1) is used to force a contradiction unless the cone has exactly one sector of angle π. The appendix gives a separate proof that the cross {xy>0} is unstable for small s.","tokens_in":17171,"tokens_out":7395,"duration_ms":73909,"significance":"If Theorem 1.3 were correct, it would be a striking purely nonlocal phenomenon: it would contradict the classical-perimeter behavior and the expected behavior for s close to 1, and it would imply regularity of the fractional min-max surfaces discussed in Section 1.1. The paper also contains useful auxiliary ingredients, notably Lemma 2.6 on the boundary-integral representation of the nonlocal mean curvature for non-smooth sets, a BV-estimate with explicit dependence on s as s→0, and an independent instability proof for the cross in the Appendix. These components appear to be of independent interest. However, the central classification rests on Lemma 3.1, and that lemma is false; the main theorem is therefore not established.","major_comments":[{"comment":"The claimed lower bound is false. Let a = 1 - cos(θ) and write the integral in the lemma as |x|^{-(1+s)} I(a), with I(a) = ∫_0^∞ dt ((t-1)^2 + 2a t)^{-(2+s)/2}. For small a the dominant contribution comes from t near 1; setting u = t - 1 gives I(a) ~ a^{-(1+s)/2} ∫_R (u^2 + 2)^{-(2+s)/2} du. The correct growth is therefore a^{-(1+s)/2}, not a^{-(1+s)}. The step in the proof that replaces (t^2 + 3a)^{-(2+s)/2} by a^{-(1+s)} (t^2 + 3)^{-(2+s)/2} is invalid, because the integral over the t-variable has a concentration window of width √a, which changes the exponent by (1+s)/2 instead of 1+s. Consequently the factor (1 - cos(θ))^{-(1+s)} in equation (8) is not available.","section":"Section 3, Lemma 3.1"},{"comment":"Even if Lemma 3.1 were replaced by the correct asymptotic, the final contradiction forcing N = 1 does not follow. With the correct exponent, the term for i = j + 2 in the displayed sum becomes (1 - cos(θ_j^{j+2}))/(1 - cos(θ_j^{j+2}))^{(1+s)/2} = (1 - cos(θ_j^{j+2}))^{(1-s)/2}, which tends to 0 for nearly parallel rays. Thus the inequality ≤ 1/100 is no longer impossible, and the argument does not exclude cones with N ≥ 2 rays. Since Lemma 3.1 is the only estimate producing the critical divergence in (8), the proof of Theorem 1.3 collapses. Corollary 1.4 and the regularity discussion in Section 1.1 inherit this gap.","section":"Section 3, proof of Theorem 1.3 after equation (8)"},{"comment":"There is a further algebraic issue in the derivation of the contradiction from (7). The inequalities s^2 ≥ c/(100N) and N ≤ C/s imply s ≥ c/(50C), not that s is small; the contradiction holds only for s below that fixed constant, which can be absorbed in s_0. This point is fixable and is not the main obstruction, but it should be corrected if the argument is revised.","section":"Section 3, Claim (9)"}],"minor_comments":[{"comment":"The denominator in the display after the Hardy-saturation step is written as |x-y|^{1+s}, whereas the stability inequality and the preceding displays use |x-y|^{2+s}. This appears to be a typo, but it creates a dimensional mismatch and should be corrected.","section":"Section 3, equation (8)"},{"comment":"The change of variables in the displayed chain of inequalities has reversed integration limits (∫_{1/2}^{-1/2} and ∫_{1/10}^{-1/10}); these should read ∫_{-1/2}^{1/2} and ∫_{-1/10}^{1/10}. Correcting the limits does not repair the exponent error noted above.","section":"Section 3, proof of Lemma 3.1"},{"comment":"There is a typo in 'conclusialon'; it should be 'conclusion'.","section":"Section 1, paragraph before Corollary 1.4"},{"comment":"The definition of θ_j^i as 'the counterclockwise angle from Σ_i and Σ_j' is ambiguous; it should specify the angle from Σ_i to Σ_j in the chosen orientation, and the modulo-2π convention should be stated explicitly.","section":"Section 3, notation"}],"recommendation":"reject","confidential_remarks":"The paper is within the journal's scope and contains some potentially useful technical tools, but the central lemma is mathematically false and the main theorem is unsupported. The independent appendix result on the cross is sound but far weaker than the classification claimed in Theorem 1.3. A rejection is appropriate because the proof of the main theorem cannot be repaired by local editing within the present framework; it requires a different estimate or a different argument."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: interesting question, fresh mechanism, but Lemma 3.1 is wrong, and it carries the whole proof of Theorem 1.3. I side with the stress-test note.\n\nWhat's new: Theorem 1.3, if true, would be the first classification of stable s-minimal cones in the punctured plane for small s, contrasting with s near 1 and the classical perimeter. The Hardy-saturation idea for s→0 is genuinely new, and the BV-estimate for small s (Theorem 2.8) looks like a useful standalone result. The appendix proof that the cross is unstable for small s is independent and appears sound.\n\nThe soft spot is not minor. In Lemma 3.1, the claimed bound is off by a factor (1-cosθ)^{-(1+s)/2}. The integral over the ray is asymptotic to c/(|x|^{1+s}(1-cosθ)^{(1+s)/2}); the substitution in the proof pulls out the wrong power and then renormalizes the t² term incorrectly. With the correct exponent, equation (8) no longer forces the blow-up used in the Claim, and the argument does not rule out many nearly parallel rays. The contradiction that yields N=1 evaporates. Everything after Lemma 3.1 depends on that false lower bound, so Theorem 1.3 is not proven. The finite-index extension in Section 4 inherits the failure.\n\nThat said, the paper is honest and competently written; the flaw is a technical error rather than fraud or confusion. The appendix and the BV estimate deserve to survive independently.\n\nBottom line: this deserves a serious referee, but not publication as is. I would send it to review with the expectation that the author must either fix Lemma 3.1 or develop a different mechanism. If the correct exponent still allows a similar argument, the result might be salvageable; as written, it does not.","headline":"The intended classification may be true, but the proof stands on a false exponent in Lemma 3.1; the paper needs a major repair before it can be trusted.","tokens_in":17726,"tokens_out":2467,"would_cite":false,"duration_ms":23485,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35R11","49Q20"],"pacs":[],"model":"deepseek-v4-flash","headline":"For the fractional perimeter with $s$ very close to 0, any stable cone in the plane is a half-plane.","keywords":["fractional perimeter","s-minimal cones","stability","Hardy inequality","nonlocal minimal surfaces","Morse index","classification of cones","cones in the plane"],"falsifier":"Evaluate the integral $\\int_0^\\infty dt/(1+t^2-2t\\cos\\theta)^{(2+s)/2}$ numerically for a small $s$ and a sequence of small angles $\\theta$, comparing its growth as $\\theta\\to0$ with the two candidate powers of $1-\\cos\\theta$; the result settles whether Lemma 3.1, and with it the proof of Theorem 1.3, is valid. A second check is to apply the radial Hardy-saturating test function to a two-ray cone with a very small angle and see whether the stability inequality fails.","tokens_in":16552,"feed_emoji":"📐","tokens_out":10166,"duration_ms":89973,"temperature":0.7,"pith_summary":"This paper establishes that, for the fractional (nonlocal) perimeter with exponent $s$ small enough, the only cones in $\\mathbb{R}^2$ that are stationary and stable away from the origin are half-planes. The result is nonlocal in character: for the classical perimeter (formally $s=1$) and for $s$ close to 1, the cross-shaped cone $\\{xy>0\\}$ is stable in $\\mathbb{R}^2\\setminus\\{0\\}$. The paper only assumes stability away from the cone's vertex, so the theorem does not use any stability information at the singular point itself. It also derives the same classification for $s$-minimal cones with finite Morse index.","feed_headline":"Stable fractional cones in the plane are flat for tiny s","feed_subtitle":"Stability away from the vertex leaves only half-planes, unlike the classical cross.","key_machinery":"The engine is Hardy's inequality for the $H^\\sigma(\\mathbb{R})$ seminorm with $\\sigma=(1+s)/2$, whose optimal constant is of order $s^2$ as $s\\to0$, together with radial functions in $C_c^2(\\mathbb{R}\\setminus\\{0\\})$ that nearly saturate it. Plugging such a radial function into the second variation inequality for the cone separates the problem into a Hardy part on each ray and an angular sum; the coefficient $1/s^2$ from saturation competes with the number of rays, forcing the cone to have exactly one sector. The auxiliary machinery consists of the BV-estimate for stable $s$-minimal surfaces with constant $C/s$, the representation of nonlocal mean curvature as a boundary integral, and the finite-index-to-stability lemma that yields the corollary.","core_discovery":"The central claim is Theorem 1.3: there exists $s_0\\in(0,1/2)$ such that for every $s\\in(0,s_0)$, every $s$-minimal cone $E\\subset\\mathbb{R}^2$ that is stable in $\\mathbb{R}^2\\setminus\\{0\\}$ in the sense of inner variations is a half-plane. The proof shows that a stable cone can have at most a controlled number of rays, then uses radially symmetric test functions that nearly saturate Hardy's inequality on the half-line to force $N=1$, and finally uses the first variation formula at a smooth boundary point to force the single angle to be $\\pi$. Corollary 1.4 extends the conclusion to cones with finite Morse index.","pith_inferences":["A natural stress test is to replace Lemma 3.1's angular lower bound with the true small-angle scaling and see whether the Hardy-saturation argument still forces $N=1$; if it does, the classification survives with a different quantitative range for $s_0$.","The same radial Hardy-saturation mechanism could be tried in higher dimensions, where it might separate radial from angular effects for other conical singularities and yield new flatness results for stable $s$-minimal cones.","The instability of the cross for small $s$ suggests a phase transition as $s$ varies, so locating the threshold would connect this small-$s$ classification to the known stability of crosses near $s=1$."],"forward_implications":["For small $s$, the fractional widths on a Riemannian surface are attained by smooth $s$-minimal surfaces: every blow-up cone is flat and the improvement-of-flatness theorem gives regularity.","The cross $\\{xy>0\\}$ is unstable in $\\mathbb{R}^2\\setminus\\{0\\}$ for small $s$, despite being stable for the classical perimeter and expected to be stable for $s$ close to 1.","The number of rays of a stable cone is bounded by a constant times $1/s$.","The classification extends from stability to finite Morse index.","There is a sharp contrast with higher dimensions: in $\\mathbb{R}^7$, smooth stable non-flat $s$-minimal cones exist for small $s$."],"supporting_citations":[{"why":"Proves the sharp Hardy inequality and the existence of radial almost-saturating functions, the main test-function tool.","marker":"[FS08]"},{"why":"Gives the BV-estimate and the stability-by-rearrangements classification whose strategy is adapted with explicit $s\\to0$ constants.","marker":"[CSV19]"},{"why":"States the same Hardy constant and develops the stability framework for $s$-minimal cones in $\\mathbb{R}^3$ for $s$ close to 1, a comparison target.","marker":"[CCS20]"},{"why":"Supplies the criterion that finite Morse index implies stability for regular $s$-minimal cones, used for Corollary 1.4.","marker":"[CFSS23]"},{"why":"Provides the second variation formula and the finite-index lemma needed for the Morse-index extension.","marker":"[FS24]"},{"why":"Supplies the first and second variation formulas for the fractional perimeter that justify the stability inequality.","marker":"[FFM+15]"},{"why":"Introduces the $s$-perimeter and the nonlocal minimal surface equation, and provides the improvement-of-flatness used for the regularity consequence.","marker":"[CRS10]"},{"why":"Proves that minimizing cones in $\\mathbb{R}^2$ are half-planes for every $s$, the minimizer counterpart that stability-only results extend.","marker":"[SV13a]"}],"fun_headline_variants":["Stable fractional cones in R^2 are half-planes for small s","For tiny s, stable s-minimal cones in the plane must be flat","Small s forces stable fractional cones to be flat in R^2","Only half-planes: stable s-cones for s near 0"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof depends on Lemma 3.1's lower bound for the interaction between two rays, which decays like $(1-\\cos\\theta)^{-(1+s)}$ in the angle $\\theta$ between the rays; if the correct small-angle decay is the weaker $(1-\\cos\\theta)^{-(1+s)/2}$, the argument forcing a stable cone to have only two rays and then angle $\\pi$ collapses.","fun_headline_variants_meta":{"raw":{"variants":["Stable fractional cones in R^2 are half-planes for small s","For tiny s, stable s-minimal cones in the plane must be flat","Small s forces stable fractional cones to be flat in R^2","Only half-planes: stable s-cones for s near 0"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000259,"raw_usage":{"total_tokens":1502,"prompt_tokens":775,"completion_tokens":727,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":391,"completion_tokens_details":{"reasoning_tokens":647}},"tokens_in":391,"tokens_out":727,"duration_ms":6849,"temperature":1.0,"reasoning_tokens":647,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T19:50:50.667949+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Evaluate the integral $\\int_0^\\infty dt/(1+t^2-2t\\cos\\theta)^{(2+s)/2}$ numerically for a small $s$ and a sequence of small angles $\\theta$, comparing its growth as $\\theta\\to0$ with the two candidate powers of $1-\\cos\\theta$; the result settles whether Lemma 3.1, and with it the proof of Theorem 1.3, is valid. A second check is to apply the radial Hardy-saturating test function to a two-ray cone with a very small angle and see whether the stability inequality fails.","supporting_citations":[],"review_version":1}