{"id":"62e61307-9dc6-49a0-a385-ace73e5c5489","arxiv_id":"2502.08318","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Any solution with finite total e^u mass of -Δu=K(x)e^u in a punctured disk satisfies u(x)=α ln|x|+O(1) near the singularity, with α>-2, and the same log asymptotics hold for polyharmonic Q-curvature equations in all dimensions.","lead":"The paper proves that solutions of the conformal Gaussian curvature equation and its higher-order Q-curvature analogues grow like a logarithmic term near an isolated singularity, with a Hölder continuous remainder. It replaces complex-analysis arguments with a unified PDE method and treats variable, bounded, nonnegative curvature functions, which was not covered before.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 3.2 asserts that v defined in (67) lies in L^{n/2}(R^n) because |v(x)|≤C ln|x|; this implication is false in R^n for n≥3, so the proof of Theorem 1.3 as written has a gap.","rationale":"I read the paper in good faith. The proofs of Theorems 1.1 and 1.2 are plausible and most technical steps check out; the flagged Brezis–Merle invocation in the proof of Theorem 1.1 is a genuine gap in exposition but is readily patched by the same splitting argument used later in Lemma 3.1, because f=Ke^u∈L^1. The truly load-bearing problem is internal to the odd-dimensional case. Lemma 3.2 explicitly asserts that logarithmic growth implies L^{n/2}(R^n) integrability, which is false for n≥3. This is not a matter of citing an external theorem; it is a concrete mathematical error inside the proof. It affects Theorem 1.3, a stated central contribution of the paper, and therefore the manuscript as written does not fully establish its claimed odd-dimensional result. The flaw is repairable—standard truncation of the potential gives the needed decay—so the appropriate verdict remains CONDITIONAL rather than REJECT. My disagreement with the Reader is about which assumption is weakest: the unbounded-potential Brezis–Merle step is secondary, while the false L^{n/2} assertion in §3.3 is the main obstacle to the proof as written. Since my concern does not change the overall verdict, I recommend UNCHANGED.","tokens_in":15345,"tokens_out":12343,"duration_ms":130331,"concrete_test":"Take n=3 and f=χ_{B_{1/2}}. Define v by (67). For large |x|, v(x)∼(c_3∫f)ln|x|, so ∫_{|x|>R}|v(x)|^{3/2}dx ∼ C∫_R^∞ r^2(ln r)^{3/2}dr = ∞. Computing this tail integral, or merely noting the divergence, directly refutes the claim in Lemma 3.2 that |v(x)|≤C ln|x| implies v∈L^{n/2}(R^n). If the authors intend a different normalization, the test is to verify explicitly that the modified potential v~ satisfies (−Δ)^{n/2}v~=f in B1\\{0} and lies in L^{n/2}(R^n).","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing gap is in §3.3, Lemma 3.2. After defining v(x)=c_n∫_{|y|<1} ln(5/|x-y|) f(y)dy, the proof claims: \"|v(x)|≤C ln|x| for large |x|. This implies that v∈L^{n/2}(R^n).\" For n≥3, this implication is false: if ∫f>0, then |v(x)|~C ln|x| as |x|→∞, and ∫_{|x|>R} |v|^{n/2}dx ≥ c∫_R^∞ r^{n-1}(ln r)^{n/2}dr = ∞. In the setting of Theorem 1.3, f=Ke^u≥0 and, unless K≡0, ∫f>0. The proof uses v∈L^{n/2}(R^n) to conclude that w=u−v is in L^{n/2}(R^n) and that (−Δ)^{n/2}w is a well-defined tempered distribution supported at {0}; without the false integrability assertion, the representation (66) and hence Theorem 1.3 are not established by the written argument. This is repairable, for example by replacing v with a truncated potential such as v~(x)=∫[ln(5/|x-y|)−ln(5/|x|)]f(y)dy, which has |v~(x)|≤C(1+|x|)^{-1} and still solves (−Δ)^{n/2}v~=f in B1\\{0}; however, the authors must state this and re-derive the subsequent regularity estimates for the modified potential. The Brezis–Merle unbounded-potential issue identified by the Reader is secondary: in the two-dimensional case one can obtain e^v∈L^p for every p>1 by splitting f=Ke^u into a small-mass L^1 part and an L^∞ part and applying Jensen's inequality, so that point is a missing detail rather than a false assertion.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies isolated singularities of the conformal Gaussian curvature equation -Δu=K(x)e^u in a punctured disk and of the higher-order analogue (-Δ)^{n/2}u=K(x)e^u in dimensions n≥3. Under the assumptions that K∈L^∞ is nonnegative and that ∫e^u is finite (plus, in higher dimensions, a mild integrability condition on |u| near the puncture), the authors claim that every such solution has the asymptotic form u(x)=α ln|x|+ϕ(x) with a Hölder continuous remainder ϕ, where α>-2 for n=2 and α>-n for n≥3. The proof is based on a representation theorem for polyharmonic Poisson equations with sign-changing solutions, followed by sector arguments that exclude derivative singularities, and potential estimates that yield Hölder regularity of the remainder. The paper also states an extension to odd dimensions using a distributional formulation of the nonlocal operator (-Δ)^{n/2}.","tokens_in":15692,"tokens_out":12137,"duration_ms":118571,"significance":"If the proof is completed, the results are significant: they provide a PDE proof of the known constant-K logarithmic asymptotics for the Gaussian curvature equation, extend the conclusion to variable L^∞ nonnegative K with no flatness conditions, and exhibit stability of the asymptotics under C^0 perturbations of K, in contrast to the scalar curvature equation. The approach is unified across the second-order, higher-order even-dimensional, and nonlocal odd-dimensional settings, and the representation theorem in Section 2 is of independent interest. The paper is largely self-contained and the main architecture of the proof is convincing; however, one of the key integrability claims in the odd-dimensional case is false as stated, and one application of a cited estimate needs explicit verification.","major_comments":[{"comment":"The proof of Lemma 3.2 asserts that for v defined in (67), the bound |v(x)|≤C ln|x| for large |x| implies v∈L^{n/2}(R^n). This implication is false for n≥3: if ∫f>0, then |v(x)| is comparable to ln|x| on a set of positive measure, and ∫_{|x|>R}(ln|x|)^{n/2}dx diverges. In the application to Theorem 1.3, f=Ke^u≥0 and ∫f>0 unless K≡0, so this is a genuine obstruction. The subsequent step defining w=u−v∈L^{n/2}(R^n) and treating (−Δ)^{n/2}w as a tempered distribution supported at {0} is therefore not justified as written. The gap appears repairable, for example by replacing v with a modified potential v~(x)=∫[ln(5/|x−y|)−ln(5/|x|)]f(y)dy, which has |v~(x)|≤C(1+|x|)^{-1} and satisfies (−Δ)^{n/2}v~=f in B1\\{0}; however, the representation (66) and the subsequent estimates in the proof of Theorem 1.3 must be re-derived for the modified potential.","section":"§3.3, Lemma 3.2"},{"comment":"The proof applies Brézis–Merle [3, Theorem 1, Corollary 1 and Remark 2] to the equation −Δv=V e^v with V=5^{−α}K e^h |x|^α. This potential is unbounded near the origin because α>−2, so the hypotheses of the cited theorem are not automatically satisfied if that theorem is stated for bounded potentials. The authors should either state the precise variant of [3] they are invoking and verify that V∈L^p for some p>1 and ∫V e^v<∞ are sufficient, or provide a self-contained argument as is done in Lemma 3.1 and the odd-dimensional case. This point is load-bearing because the conclusion e^v∈L^p underlies the W^{2,p} estimate and the Hölder regularity of ϕ in Theorem 1.1.","section":"§3.1, proof of Theorem 1.1, equations (42)-(43)"}],"minor_comments":[{"comment":"The text 'n /greaterorequalslant2m' should read 'n≥2m'.","section":"Theorem 1.4 statement"},{"comment":"There is a typo 'h ∈∈ L^{n/2}(R^n)'; it should be 'h∈L^{n/2}(R^n)'.","section":"Lemma 3.2"},{"comment":"The inequality ∫_{B_r}(5/|x−y|)^{k c_m}dx ≤ ∫_{B_r}(5/|x|)^{k c_m}dx is not literally true for every y∈B_r because the ball is not translation-invariant; the right-hand side should be an integral over a slightly enlarged ball with a harmless constant. The intended estimate is correct, but the line as written is imprecise.","section":"Lemma 3.1, inequality (60)"},{"comment":"The proof of conclusion (i) is omitted with the remark that it is similar to that of Theorem 1.2. Since the odd-dimensional setting is nonlocal and the fundamental solution is logarithmic, the analogous sector argument should be sketched or at least the necessary modifications explained.","section":"Theorem 1.3, proof of conclusion (i)"},{"comment":"There is a typographical error 'con formal' in the abstract; it should be 'conformal'.","section":"Abstract"}],"recommendation":"major_revision","confidential_remarks":"The paper fits the scope of the journal and the main results are likely correct, but the proof of Theorem 1.3 contains a false integrability assertion in Lemma 3.2 that must be repaired before the nonlocal case is established. The 2D and even-dimensional results appear sound modulo the Brézis–Merle verification requested in the major comments. I recommend major revision rather than rejection because the gap is fixable without changing the claimed theorems."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Let me give you the bottom line first: this is a worthwhile paper with one gap that needs fixing. The genuinely new parts are the logarithmic asymptotics for the Gaussian curvature equation with variable, nonnegative L∞ coefficient in the punctured disk, and the extension to even-dimensional polyharmonic Q-curvature equations. Those results read as solid, and the proof machinery—the representation theorem with source terms, the sector arguments that kill derivative singularities, the exponential-integrability lemma in the even case—is in place. The odd-dimensional case is where the written argument fails.\n\nIn Lemma 3.2, v(x)=c_n∫_{|y|<1} ln(5/|x-y|) f(y)dy. The proof asserts |v(x)|≤C ln|x| for large |x|, and therefore v∈L^{n/2}(R^n). That implication is false for n≥3. Here f=Ke^u is nonnegative and not identically zero, so ∫f>0 and |v(x)| behaves like C ln|x| at infinity, which is not in L^{n/2}. This membership is exactly what allows w=u−v to lie in L^{n/2} and gives a tempered distribution (−Δ)^{n/2}w supported at {0}. Without it, representation (66) and Theorem 1.3 are not established by the written proof. The fix is standard—replace the kernel by φ(x−y)−φ(x) or otherwise truncate the logarithmic term—and I suspect the theorem is true, but the authors need to state the modification and redo the estimates that follow.\n\nThe reader's Brezis–Merle concern is secondary. In the n=2 proof, the potential 5^{−α}K e^h |x|^α is unbounded, and the paper cites Brezis–Merle without stating a version covering that setting. That is a missing detail rather than a false assertion, and it can be handled by splitting f into a small-mass part and an L∞ part.\n\nCitation practice is honest, no fitted parameters, and the paper is largely self-contained. I would send it to a serious referee. The even-dimensional results deserve to be in the literature, and the odd-dimensional gap is repairable. For a reading group, bring it only if someone is willing to work out the truncated potential.","headline":"Variable-K Gaussian and even-dimensional Q-curvature results are solid, but the odd-dimensional proof rests on a false L^{n/2} integrability claim for the logarithmic potential, so Theorem 1.3 needs repair.","tokens_in":16283,"tokens_out":3231,"would_cite":true,"duration_ms":31756,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35J91","35B40","35A21"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that any finite-volume solution of $-\\Delta u=K(x)e^u$ near an isolated singularity must take the form $u(x)=\\alpha\\ln|x|+\\phi(x)$ with $\\alpha>-2$ and $\\phi$ H\\\"older continuous, for any nonnegative bounded curvature…","keywords":["isolated singularities","conformal Gaussian curvature equation","Q-curvature equation","Liouville-type equation","polyharmonic equation","asymptotic expansion","Brezis-Merle estimates"],"falsifier":"Compute $\\int_{B_{1/2}}e^{k v(x)}dx$ for the logarithmic potential $v(x)=\\int\\ln(5/|x-y|)K(y)e^{u(y)}dy$ with an admissible unbounded coefficient $|x|^{\\alpha}e^h$, $\\alpha>-2$; if for some nonnegative bounded $K$ and finite-volume $u$ this integral diverges for every $k>0$, the regularity step in Theorem 1.1 fails. Equivalently, a solution satisfying the hypotheses whose angular first-derivative coefficient $a_{(1,0)}$ in (38) is nonzero would refute the theorem, because the paper's sector integrals show such a term forces $\\int e^u$ to diverge.","tokens_in":15069,"feed_emoji":"📐","tokens_out":7065,"duration_ms":63670,"temperature":0.7,"pith_summary":"This paper establishes the asymptotic profile of solutions to the conformal Gaussian curvature equation $-\\Delta u=K(x)e^u$ in a punctured disk near the puncture. The result: if $K$ is bounded and nonnegative and the solution has finite total volume $\\int e^u<\\infty$, then $u(x)=\\alpha\\ln|x|+\\phi(x)$ near the origin, with $\\alpha>-2$ and $\\phi$ H\\\"older continuous. The proof is purely PDE-based, in contrast to the earlier complex-analysis proof for $K\\equiv 1$, and it handles variable $K(x)$. The same argument is extended to the conformal $Q$-curvature equation $(-\\Delta)^{n/2}u=K(x)e^u$ in every dimension $n\\ge 3$, where finite volume plus a mild integrability condition on $|u|$ yields the same logarithmic structure with $\\alpha>-n$.","feed_headline":"Every finite-volume blow-up is a log plus a H\\\"older term","feed_subtitle":"New PDE proof handles variable curvature and Q-curvature equations in all dimensions.","key_machinery":"The load-bearing tool is Theorem 1.4, a representation formula for sign-changing solutions of the polyharmonic Poisson equation $(-\\Delta)^m u=f$: any solution decomposes as a Newtonian or logarithmic potential $N(x)$, a smooth homogeneous solution $h$, and a finite sum of derivatives $D^\\beta\\varphi$ of the fundamental solution with $|\\beta|\\le 2m-1$. Finite volume forces coefficients with $|\\beta|\\ge 1$ to vanish and bounds the coefficient of the logarithmic term, yielding $u=\\alpha\\ln|x|$ plus a regular remainder. The remainder's H\\\"older regularity comes from Br\\'ezis-Merle exponential integrability: the weighted nonlinearity $K e^h |x|^\\alpha e^v$ lies in $L^{p_0}$, so classical $W^{2m,p}$ estimates apply.","core_discovery":"The central claim is Theorem 1.1: for $K\\in L^\\infty(B_1)$, $K\\ge 0$, and $u\\in C^2(B_1\\setminus\\{0\\})$ solving $-\\Delta u=K e^u$ with $\\int e^u<\\infty$, there is $\\alpha>-2$ and $\\phi\\in C^\\gamma_{\\rm loc}(B_1)$, $0<\\gamma<1$, such that $u=\\alpha\\ln|x|+\\phi$ near $0$. The higher-dimensional analogue, Theorems 1.2 and 1.3, asserts the same logarithmic form $u=\\alpha\\ln|x|+\\phi$ with $\\alpha>-2m$ for even $n=2m\\ge 4$ and $\\alpha>-n$ for odd $n\\ge 3$, under $\\int e^u<\\infty$ and an additional $|u|$-integrability condition. The authors prove that all derivative-of-fundamental-solution singularities are killed by the finite-volume condition, leaving only the logarithmic term; the remainder is regularized via exponential integrability estimates.","pith_inferences":[],"forward_implications":["For the exterior-domain Gaussian curvature problem, the Kelvin transform turns Theorem 1.1 into the statement that solutions have $v(x)=\\beta\\ln|x|+O(1)$ near infinity with $\\beta<-2$.","The logarithmic asymptotics are stable under $C^0$ perturbations of $K$, unlike the scalar-curvature equation in higher dimensions, where $C^0$ and $C^1$ perturbations can destroy Fowler-type estimates.","Unbounded kernels of the form $|x|^{-\\gamma}K(x)$ are covered by the same theorem after the shift $w=u-\\gamma\\ln|x|$, giving $(\\alpha+\\gamma)\\ln|x|+\\phi$ with $\\alpha>-2$.","For the $Q$-curvature equation, the same form holds in even and odd dimensions, including the nonlocal odd-dimensional case, under finite volume and $\\int_{B_r}|u|\\,dx=o(r^{n-2})$.","The representation theorem also applies when $f$ is not integrable, satisfying only $\\int |x|^s|f|dx<\\infty$, so the method is ready for finite-volume-type assumptions that fail.","The proof never uses differentiability or flatness of $K$, so the value of $\\alpha$ should vary continuously under $L^\\infty$ perturbations of $K$ while the logarithmic form persists; the paper does not discuss this continuity.","The method suggests that the extra $|u|$-integrability condition in Theorems 1.2 and 1.3 may be relaxable: the representation formula already controls the singular terms, so a weaker integrability hypothesis might suffice.","A natural test of sharpness is the radial family $u_\\alpha=\\alpha\\ln r$ with $K=0$; finite volume holds exactly for $\\alpha>-2$ (or $\\alpha>-n$), matching the theorem's range, so the range is likely optimal."],"supporting_citations":[{"why":"It supplies the Br\\'ezis-Merle exponential integrability estimate used to prove $e^v\\in L^p$ and hence the H\\\"older regularity of the remainder.","marker":"[3]"},{"why":"It proved the $K\\equiv 1$ case by complex analysis, providing the baseline result this paper reproves and generalizes by PDE methods.","marker":"[14]"},{"why":"It proved the Bochner-type representation for polyharmonic functions with zero right-hand side, which Theorem 1.4 extends to nonzero $f$ and sign-changing $u$.","marker":"[17]"},{"why":"It classified solutions of the constant-curvature equation on $\\mathbb{R}^2$, the background classification result that motivates the singular-profile question.","marker":"[9]"},{"why":"It supplies the classical $W^{2,p}$ estimates and potential estimates used to bootstrap the regular part of the solution.","marker":"[20]"},{"why":"It gives the classical isolated-singularity results for linear elliptic equations that pattern the representation formula.","marker":"[2]"}],"fun_headline_variants":["Finite volume forces log + Hölder blow-up","Variable curvature in all dimensions: log + Hölder","PDE method resolves singularities for variable K","Logarithmic blow-up for curvature equations","Q-curvature singularities tamed by PDE proof"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof assumes that the Br\\'ezis-Merle exponential-integrability estimate remains valid when the coefficient $V(x)=5^{-\\alpha}K(x)e^{h(x)}|x|^{\\alpha}$ is unbounded near the origin; the paper invokes this as a direct consequence of [3] without spelling out a proof for unbounded $V$.","fun_headline_variants_meta":{"raw":{"variants":["Finite volume forces log + Hölder blow-up","Variable curvature in all dimensions: log + Hölder","PDE method resolves singularities for variable K","Logarithmic blow-up for curvature equations","Q-curvature singularities tamed by PDE proof"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001233,"raw_usage":{"total_tokens":5080,"prompt_tokens":978,"completion_tokens":4102,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":594,"completion_tokens_details":{"reasoning_tokens":4028}},"tokens_in":594,"tokens_out":4102,"duration_ms":30097,"temperature":1.0,"reasoning_tokens":4028,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-08T05:39:37.038911+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute $\\int_{B_{1/2}}e^{k v(x)}dx$ for the logarithmic potential $v(x)=\\int\\ln(5/|x-y|)K(y)e^{u(y)}dy$ with an admissible unbounded coefficient $|x|^{\\alpha}e^h$, $\\alpha>-2$; if for some nonnegative bounded $K$ and finite-volume $u$ this integral diverges for every $k>0$, the regularity step in Theorem 1.1 fails. Equivalently, a solution satisfying the hypotheses whose angular first-derivative coefficient $a_{(1,0)}$ in (38) is nonzero would refute the theorem, because the paper's sector integrals show such a term forces $\\int e^u$ to diverge.","supporting_citations":[{"cited_title":"Br´ ezis, F","cited_arxiv_id":null,"evidence_quote":"It supplies the Br\\'ezis-Merle exponential integrability estimate used to prove $e^v\\in L^p$ and hence the H\\\"older regularity of the remainder."},{"cited_title":"Chou, Y .-H","cited_arxiv_id":null,"evidence_quote":"It proved the $K\\equiv 1$ case by complex analysis, providing the baseline result this paper reproves and generalizes by PDE methods."},{"cited_title":"Futamura, K","cited_arxiv_id":null,"evidence_quote":"It proved the Bochner-type representation for polyharmonic functions with zero right-hand side, which Theorem 1.4 extends to nonzero $f$ and sign-changing $u$."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It classified solutions of the constant-curvature equation on $\\mathbb{R}^2$, the background classification result that motivates the singular-profile question."},{"cited_title":"Gilbarg, N","cited_arxiv_id":null,"evidence_quote":"It supplies the classical $W^{2,p}$ estimates and potential estimates used to bootstrap the regular part of the solution."},{"cited_title":"Br´ ezis, P .-L","cited_arxiv_id":null,"evidence_quote":"It gives the classical isolated-singularity results for linear elliptic equations that pattern the representation formula."}],"review_version":1}