{"id":"9d03f9e3-70bf-4f8f-9d2b-cfe15432a96f","arxiv_id":"2411.17016","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For uniformly degenerate elliptic equations whose positive characteristic exponent varies along the boundary, solutions decompose into a regular part plus explicit terms ρ^{m+j}(logρ)^j, with sharp Hölder regularity.","lead":"The paper proves sharp boundary regularity for a class of elliptic equations that degenerate at the boundary, showing that solutions split into a smooth piece plus explicit singular terms built from powers of the distance to the boundary and powers of the log of that distance, even when the key power varies from point to point. Specialists in geometric PDEs may use the decomposition as a tool for nonlinear problems such as the Loewner-Nirenberg and Monge-Ampere equations.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The α<γ branch and the smooth theorem rest on Theorem 6.2, whose proof is omitted; without it, Theorem 3.2 in the case α<γ and Theorem 3.3 are unsupported.","rationale":"The reader identified Theorem 4.1, imported from preprint [31], as the weakest assumption. That is a legitimate concern: Theorem 4.1 is the base-regularity engine, and the paper gives no proof. However, Theorem 6.2 is an internally stated theorem with an omitted proof, and it is essential for the α<γ half of Theorem 3.2 and for the claimed proof of Theorem 3.3 via the unproved extension of Theorem 5.4. The proof of Theorem 5.2 is detailed and suggests the method is plausible, but the α<γ regime has a different expansion pattern and different coefficient regularities, so the omitted induction is genuinely load-bearing. The concrete test is to write out the missing induction and confirm the indexing and regularity close; this would settle whether the central claim holds in full generality. Since the gap is real but potentially fillable, the reader's CONDITIONAL verdict is appropriate and should be retained.","tokens_in":53810,"tokens_out":5731,"duration_ms":53685,"concrete_test":"Extract the proof of Theorem 6.2 by following the induction pattern of Theorem 5.2 in the α<γ regime: write F = Σ_{i=0}^{k} a_i t^i + Σ_{i=[m]+1}^{k} Σ_{j=0}^{i-[m]} a_{i,j} t^{i+γ}(logt)^j + S_k and check that substituting into (4.21) yields exactly (6.17), with the coefficient c_{k,0},...,c_{k,k-[m]} appearing only in the second expansion and with (6.18)–(6.22) satisfied. In particular verify that c_{i,j} has C^{ℓ-i-1,ε} regularity, one derivative weaker than in Theorem 5.2, and that the remainder estimates (6.19)–(6.22) close. If any step fails, the α<γ branch of Theorem 3.2 and the claimed extension of Theorem 5.4 to [m]≥1 collapse.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 3.2's α<γ case is proved via Theorem 6.3, whose proof invokes Theorem 6.2 in full (expansion (6.17), estimates (6.18)–(6.22)). Theorem 6.2 is stated with 'The proof ... is similar to that of Theorem 5.2 and hence omitted.' This is not a routine difference: in the α<γ expansion the singular term t^{[m]+γ} is followed by t^{[m]+1}, t^{[m]+1+γ}logt, ..., t^k, whereas for γ<α the remainder R_k controls t^{k+γ}. The induction in Theorem 5.2 repeatedly differentiates t^γ via D_{x'} t^γ = t^γ logt D_{x'}γ, creating new log powers; the same mechanism must work in Theorem 6.2 but with different index bookkeeping and with coefficient regularity c_{i,j}∈C^{ℓ-i-1,ε} (one derivative lower in x'). The omitted proof is exactly what must show this closes. Moreover, after Theorem 6.3 the authors state 'By using Theorem 6.2, we can prove Theorem 5.4 without the additional assumption [m]≥1' and then 'Theorem 3.3 follows' — again without proof. Since Theorem 3.3 is a central claim, this is a genuine gap, not a stylistic omission. Theorem 4.1 imported from [31] is also unproved here, but at least it is identified as an external result; Theorem 6.2's omission is internal and load-bearing.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies boundary regularity for linear uniformly degenerate elliptic operators L = ρ²a^{ij}∂_{ij} + ρb^i∂_i + c, where the characteristic polynomial P(μ) = μ(μ−1)a^{ij}ν_iν_j + μb^iν_i + c may have a positive root m = [m] + γ that varies along the boundary. The main results are Theorem 3.2, a finite-differentiability decomposition u = v + Σ_{j=0}^{k*−[m]} w_j ρ^γ (log ρ)^j with v ∈ C^{k,α} and w_j ∈ C^{k*,ε}, and Theorem 3.3, a smooth analogue with an absolutely and uniformly convergent infinite series. The proof flattens the boundary, rewrites the equation as a singular ODE in the normal variable t, and then derives expansions in powers t^i and t^{i+γ}(log t)^j with remainder estimates. Section 5 carries out the case γ < α in detail; Section 6 treats α < γ, but its central expansion theorem, Theorem 6.2, is stated without proof.","tokens_in":54154,"tokens_out":5836,"duration_ms":49027,"significance":"If the main theorems hold, the paper provides a systematic polyhomogeneous boundary expansion for uniformly degenerate elliptic equations with a varying indicial exponent, going beyond the constant-exponent results of the authors' earlier work and the classical microlocal expansions. The explicit identification of the logarithmic factors produced by the variation of γ, and the uniform convergence of the infinite expansion without analyticity, are valuable and novel contributions. The γ < α half is supported by a detailed induction with explicit remainder estimates in Lemmas 5.1 and Theorem 5.2, and the paper is well informed by the geometry of the problem. However, the α < γ branch and the smooth theorem are not fully proved in the manuscript, so the contribution is currently conditional on filling that gap.","major_comments":[{"comment":"Theorem 6.2 is the exact analogue for the case α < γ of the expansion theorem Theorem 5.2, and it is stated with 'The proof ... is similar to that of Theorem 5.2 and hence omitted.' This omission is load-bearing: Theorem 6.3's proof invokes Theorem 6.2 in full (its expansions (6.17) and estimates (6.18)–(6.22)), and Theorem 3.2 in the case α < γ follows only through Theorem 6.3. The differences from Theorem 5.2 are not routine: in the α < γ expansion the integer term t^{[m]+1} precedes the singular terms t^{[m]+1+γ}(log t)^j, and the coefficient regularity c_{i,j} ∈ C^{ℓ−i−1,ε} is one derivative lower, while the remainder R'_{k−1} has integer order. The induction in Theorem 5.2 uses repeatedly D_{x'} t^γ = t^γ log t D_{x'} γ to generate new log-powers; the same mechanism has to be verified with different index bookkeeping and with the lower coefficient regularity in Theorem 6.2. The manuscript should contain this induction.","section":"6, Theorem 6.2"},{"comment":"The paragraph following Theorem 6.3 asserts: 'By using Theorem 6.2, we can prove Theorem 5.4 without the additional assumption [m] ≥ 1. Then, Theorem 3.3 follows by renaming the functions w_j, for j ≥ 0.' This is the only justification given for the smooth Theorem 3.3 in the general case, and it is asserted without proof. Since Theorem 3.3 is one of the two main results and includes the uniform absolute convergence of the infinite series and the differentiated identity (3.3), this is a genuine gap in the manuscript, not a matter of presentation. A proof, or at least a precise reduction to the (provided) proof of Theorem 6.2, is needed.","section":"6, after Theorem 6.3"},{"comment":"Theorem 4.1 supplies the base regularity estimates for u, t∂_t u, and t^2 ∂_t^2 u in the flat model; every subsequent expansion in Sections 5 and 6 starts from these estimates. The theorem is imported from the authors' own preprint [31] and is stated in the present paper only with the reference 'Refer to [31] for details.' If [31] is not yet published, the manuscript is not self-contained at a load-bearing point. Either a proof should be included in an appendix, or the dependence on [31] should be made explicit in the statements of Theorems 3.2 and 3.3.","section":"4, Theorem 4.1"}],"minor_comments":[{"comment":"In the proof, the first sentence 'By a^{ij}, b^i, c, f ∈ C^β(¯G_1)' silently uses that the assumed C^α regularity implies C^β for β ≤ α; this is true but should be stated once, as the phrase may be read as a stronger assumption than the theorem actually makes.","section":"4, Theorem 4.3"},{"comment":"The index convention for c_{i,j}, with j counting powers of log t, is explained in the paragraph after the statement of Theorem 5.2; however, the same convention is used in the proof for a_{i,j} and b_{p,j}, where the index ranges are set implicitly by statements such as 'we set c_{i,j}=0 for ...' — a one-sentence reminder before (5.20) would avoid confusion.","section":"5, Theorem 5.2"},{"comment":"Several results in the appendix are described with 'The proof is similar to that of Lemma 7.2/7.8 and is omitted.' The analogy is plausible, but the statements are long and vary in the Hölder exponents used in x' and t, so a short derivation, especially of the key integral conditions, would make the appendix easier to verify.","section":"7, Lemmas 7.4, 7.5, 7.6, 7.9"}],"recommendation":"major_revision","confidential_remarks":"The paper has a detailed and convincing proof for one half of the main theorem (the case γ < α), but the other half (α < γ) and the smooth theorem rest on an omitted proof of Theorem 6.2 and on an unproved assertion at the end of Section 6. This is the main obstacle to acceptance. The reliance on the authors' own preprint [31] for Theorem 4.1 is acceptable if the preprint is available, but it should be stated more prominently. With the missing proofs supplied, the paper would be a strong contribution to the boundary regularity of degenerate elliptic equations."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Let me give you the short version before the details. The paper's central claim is right in spirit: when the positive characteristic exponent m(x') varies, solutions pick up logarithmic factors (logρ)^j alongside ρ^{m+j}, and the authors give explicit finite and infinite decompositions. That is genuinely new and useful, and the proof for the case γ<α (where γ is the fractional part of m) is thorough. The problem is the other half, α<γ, which rests on Theorem 6.2, and Theorem 6.2 is stated with the proof omitted. That omission is load-bearing, not cosmetic.\n\nWhat the paper does well: Example 3.1 shows exactly why log factors appear—differentiating ρ^γ in a tangential direction produces ρ^γ logρ, and the induction in Section 5 tracks these terms carefully. The remainder estimates are explicit, and the infinite series in Theorem 3.3 is shown to converge absolutely and uniformly without analyticity. The constant-exponent results in Section 2 are special cases, which is the right framing. The paper is honest about its reliance on the authors' own preprint [31] for base regularity (Theorem 4.1); that is identified as an external input, though it carries a lot of weight.\n\nThe soft spot: Theorem 6.2 is the analogue of Theorem 5.2 for α<γ, and it is not a routine rerun. The pattern of exponents changes—the singular term t^{[m]+γ} is followed by t^{[m]+1}, t^{[m]+1+γ}logt, etc., and the coefficient regularity in (6.18) is one derivative lower in x'. The remainder estimates (6.19)–(6.22) are also different from (5.14)–(5.16). The authors say \"the proof is similar\" and mention two differences, but they don't provide the bookkeeping. Then Theorem 6.3, the decomposition for α<γ, is proved assuming Theorem 6.2, and Theorem 5.4 without the assumption [m]≥1—which is needed for Theorem 3.3—is asserted to follow from Theorem 6.2 with no argument. So the smooth result and the α<γ finite result both hang on that missing proof.\n\nAnyone working on boundary regularity for degenerate elliptic operators, or on polyhomogeneous expansions, will want to know these decompositions. The paper deserves a serious referee, because the method and the γ<α half are solid and the result is significant. But a referee should require the proof of Theorem 6.2, or at least a detailed sketch that addresses the changes in exponent bookkeeping and coefficient regularity, before the paper is accepted. If that proof works, this is a strong contribution. Right now it is conditional.","headline":"Variable-exponent degenerate elliptic regularity with explicit log terms: the γ<α half is solid, but the α<γ half and the smooth theorem rest on an omitted proof; referee it and demand the proof.","tokens_in":54678,"tokens_out":3474,"would_cite":false,"duration_ms":29933,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35J70","35B65","35J25","35C20"],"pacs":[],"model":"deepseek-v4-flash","headline":"With a varying characteristic exponent, solutions split into a regular part plus explicit singular factors ρ^{m+j}(log ρ)^j — the only obstructions to higher regularity.","keywords":["uniformly degenerate elliptic equations","characteristic exponents","indicial roots","boundary regularity","polyhomogeneous expansions","varying characteristic exponents","Hölder spaces","logarithmic singularities"],"falsifier":"Take the flat model $t^2u_{tt}+p(x')tu_t+q(x')u=f$ with $m(x')=1+\\gamma(x')$, $\\gamma$ smooth with $0<\\gamma<1$, and $f$ smooth, and compute the expansion coefficients recursively as in Theorem 5.2. The theorem predicts that after subtracting $v+\\sum_{j=0}^{k-1}w_j t^\\gamma(\\log t)^j$, the remainder is $C^{k,\\alpha}$ with the stated decay; a remainder that fails that decay, or a solution with $\\partial_t^{[m]}w_0=0$ that is still not $C^{k,\\alpha}$, would falsify the decomposition. This can be checked explicitly in one tangential variable because the coefficient recursion and the integral formula are fully explicit.","tokens_in":99,"feed_emoji":"📐","tokens_out":15001,"duration_ms":188392,"temperature":0.7,"pith_summary":"Uniformly degenerate elliptic operators are second-order operators of the form $Lu = \\rho^2 a^{ij}\\partial_{ij}u + \\rho b^i\\partial_i u + cu$, where $\\rho$ is the distance to the boundary, so the equation is elliptic inside the domain but degenerates at the boundary. Earlier results for constant characteristic exponents say that solutions are a regular function plus a singular factor $\\rho^\\gamma$ (or $\\rho^m\\log\\rho$ when $m$ is an integer). This paper removes the constancy assumption and lets the positive characteristic exponent $m=[m]+\\gamma$ vary with the tangential position, with $0<\\gamma<1$. It proves that solutions decompose into a regular part plus a finite sum (or, for smooth data, an absolutely and uniformly convergent infinite sum) of terms $\\rho^{m+j}(\\log\\rho)^j$, with explicit Hölder estimates and with the condition that vanishing of the leading singular coefficient restores full regularity. That matters because many geometric and stochastic boundary-value problems produce exactly this kind of boundary degeneracy, and knowing the precise singular structure is what makes sharp regularity theorems possible.","feed_headline":"Varying singular exponents add log terms to solution series","feed_subtitle":"Near the boundary, solutions split into a smooth part plus explicit ρ^γ(log ρ)^j singular terms — and nothing worse.","key_machinery":"The engine is the reduction, on a flattened boundary portion, of $Lu=f$ to an ordinary differential equation in the normal variable $t$: $t^2u_{tt}+p(x')tu_t+q(x')u=F$, where $p=1-(m+\\underline{m})$ and $q=m\\,\\underline{m}$, with $\\underline{m}<0<m$ the two characteristic exponents. Lemma 4.4 solves this ODE explicitly by variation of parameters, expressing $u$ as a combination of $t^m$ and $t^{\\underline{m}}$ with integrals of $F$. The proof expands $F$ into powers $t^i$ and $t^{i+\\gamma}(\\log t)^j$, feeds the expansion into the integral formula, and controls the Hölder regularity of the remainders through the singular-integral estimates of Lemmas 7.1–7.9. Because $\\gamma$ varies, tangential derivatives of $t^\\gamma$ bring in factors $t^\\gamma\\log t$, which is exactly how the logarithmic terms are born. The estimates are formulated in mixed Hölder spaces $C^{\\beta,\\alpha}_{x',t}$ in which tangential and normal directions may have different Hölder exponents, tracking the anisotropic regularity loss.","core_discovery":"The paper's central claim is that a varying characteristic exponent creates no singularities beyond the expected powers $\\rho^{m+j}$ multiplied by integer powers of $\\log\\rho$. Theorem 3.2 states that if $m=[m]+\\gamma$ with $0<\\gamma<1$ on a boundary neighborhood, then any continuous solution of $Lu=f$ with $u=f/c$ on the boundary can be written as $u=v+\\sum_{j=0}^{k^*-[m]} w_j \\rho^\\gamma(\\log\\rho)^j$, where $v\\in C^{k,\\alpha}$, each $w_j\\in C^{k^*,\\varepsilon}$, and explicit estimates hold; if the $[m]$-th normal derivative of $w_0$ vanishes on the boundary, then $u\\in C^{k,\\alpha}$. Theorem 3.3 is the smooth analogue: with smooth coefficients and data, $u=v+\\sum_{j=0}^{\\infty} w_j \\rho^{m+j}(\\log\\rho)^j$, the series converging absolutely and uniformly on compact subregions, with $v$ and all $w_j$ smooth, and the same vanishing condition forcing $u$ to be $C^\\infty$. The logarithmic factors are the genuinely new feature: because $\\gamma$ depends on the tangential variables, differentiating $\\rho^\\gamma$ tangentially produces $\\rho^\\gamma\\log\\rho$ factors, so even a non-integer exponent that varies generates a logarithmic ladder.","pith_inferences":["This suggests a general rule for geometric boundary problems whose indicial root varies along the boundary: polyhomogeneous expansions will generically contain integer powers of $\\log\\rho$ even when all characteristic exponents are non-integers, so constant-exponent models understate the singularity structure.","One testable extension is to compute the recursion displayed in Theorem 5.2 for a concrete operator in one tangential variable and verify numerically that the remainder decays at the stated rate; this would provide an independent check of the decomposition.","The use of mixed Hölder spaces $C^{\\beta,\\alpha}_{x',t}$ here hints that these spaces may be the natural scale for variable-exponent degeneracies in other contexts, such as stochastic control or conformal geometry, where the same kind of anisotropic regularity loss is expected.","A further question the paper leaves open is whether the infinite polyhomogeneous sum persists with less than $C^\\infty$ boundary regularity; the current smooth result requires $C^\\infty$ data, so the optimal finite-regularity threshold for the infinite series is not addressed."],"forward_implications":["At any Hölder level $k+\\alpha$ with $k+\\alpha>m$, a solution differs from a $C^{k,\\alpha}$ function only by a finite, explicitly identified singular sum $\\rho^\\gamma(\\log\\rho)^j$; no further unknown boundary singularities can occur.","If the leading singular coefficient satisfies $\\partial_\\nu^{[m]}w_0=0$ on the boundary, the singular terms drop out and the solution is genuinely $C^{k,\\alpha}$ (and $C^\\infty$ in the smooth case), so the regularity of the data is then fully transmitted to the solution.","The infinite expansion of Theorem 3.3 may be differentiated term by term to any order, with the differentiated remainder series converging absolutely and uniformly, making the series a usable computational tool for boundary asymptotics.","The method is developed for linear equations and the authors state it can be adapted to nonlinear equations, so the same decomposition should govern the singular boundary behavior in the nonlinear problems listed in the introduction."],"supporting_citations":[{"why":"Supplies Theorem 4.1, the base flat-boundary regularity estimates for $u$, $t\\partial_t u$, and $t^2\\partial_t^2 u$ from which every later expansion starts.","marker":"[31]"},{"why":"Provides the extension lemma (Lemma 4.5) used to construct the regular part $v$ and the singular coefficients $w^{(j)}$ with prescribed normal derivatives.","marker":"[20]"},{"why":"Supplies the interpolation estimate used in Theorem 4.3 to raise tangential Hölder regularity and prove the mixed-Hölder improvement behind the decomposition.","marker":"[8]"}],"fun_headline_variants":["Varying singular exponents add log terms to elliptic solutions","Log terms emerge when singular exponents vary","Variable exponents turn solution series into log ladders","Non-integer exponent variation forces log factors in expansions","Degenerate elliptic solutions gain log series from variable exponents"],"cache_read_input_tokens":56704,"weakest_assumption_plain":"The argument starts from a base regularity statement for the flat boundary model — that $u$, $t\\partial_t u$, and $t^2\\partial_t^2 u$ are Hölder up to the boundary under the stated hypotheses — which the paper imports from the authors' own preprint and does not prove here; if that statement were false or required stronger assumptions, the decompositions built on it would collapse.","fun_headline_variants_meta":{"raw":{"variants":["Varying singular exponents add log terms to elliptic solutions","Log terms emerge when singular exponents vary","Variable exponents turn solution series into log ladders","Non-integer exponent variation forces log factors in expansions","Degenerate elliptic solutions gain log series from variable exponents"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000429,"raw_usage":{"total_tokens":2331,"prompt_tokens":847,"completion_tokens":1484,"prompt_tokens_details":{"cached_tokens":768},"prompt_cache_hit_tokens":768,"prompt_cache_miss_tokens":79,"completion_tokens_details":{"reasoning_tokens":1412}},"tokens_in":79,"tokens_out":1484,"duration_ms":41953,"temperature":1.0,"reasoning_tokens":1412,"cache_read_input_tokens":768,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T12:36:50.122276+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the flat model $t^2u_{tt}+p(x')tu_t+q(x')u=f$ with $m(x')=1+\\gamma(x')$, $\\gamma$ smooth with $0<\\gamma<1$, and $f$ smooth, and compute the expansion coefficients recursively as in Theorem 5.2. The theorem predicts that after subtracting $v+\\sum_{j=0}^{k-1}w_j t^\\gamma(\\log t)^j$, the remainder is $C^{k,\\alpha}$ with the stated decay; a remainder that fails that decay, or a solution with $\\partial_t^{[m]}w_0=0$ that is still not $C^{k,\\alpha}$, would falsify the decomposition. This can be checked explicitly in one tangential variable because the coefficient recursion and the integral formula are fully explicit.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies Theorem 4.1, the base flat-boundary regularity estimates for $u$, $t\\partial_t u$, and $t^2\\partial_t^2 u$ from which every later expansion starts."},{"cited_title":"Gilbarg, L","cited_arxiv_id":null,"evidence_quote":"Provides the extension lemma (Lemma 4.5) used to construct the regular part $v$ and the singular coefficients $w^{(j)}$ with prescribed normal derivatives."},{"cited_title":"Caffarelli, X","cited_arxiv_id":null,"evidence_quote":"Supplies the interpolation estimate used in Theorem 4.3 to raise tangential Hölder regularity and prove the mixed-Hölder improvement behind the decomposition."}],"review_version":1}