{"id":"cd5e98b5-1d11-43bd-aa11-1f0be0edbf57","arxiv_id":"2608.11960","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"For every toric mirror curve with a generic one-node degeneration and every genus at least 2, the conifold free energy equals B_{2g}/(2g(2g-2)) t^{2-2g} plus a holomorphic remainder.","lead":"This paper proves a general form of the conifold gap, a universal singularity of the free energy near a node, for all toric mirror curves. The result gives a rigorous boundary condition for the holomorphic anomaly equations across the entire toric Calabi-Yau class.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Rauch variational sign check in Theorem 4.2 is invalid: the P1 example uses the wrong Bergman kernel, so the sign of Eq. (4.3) is unverified; a sign flip would break the Euler-defect cancellation.","rationale":"The reader identified the uniform scaled estimates of Theorem 3.4 as the weakest assumption. That is a reasonable concern, but I did not find a concrete flaw there: the paper supplies a detailed Neumann-series proof with a moving-family jump operator, and the radius/epsilon bookkeeping appears internally consistent. By contrast, the P1 sign check for the Rauch variational formula is demonstrably incorrect: it uses a Bergman kernel incompatible with the chosen projection, and the computation reduces to 0 = 0 for the correct kernel. Since the sign of Eq. (4.3) controls the exact cancellation of t I_g^neck in the Euler defect, this is more load-bearing than an exponent typo. The exponent inconsistency in (5.7) is real but cosmetic: the displayed algebra uses 4^{2g-2}, which corresponds to T^{2-2g}, so correcting the quoted exponent preserves the final constant. The sign issue, however, is unresolved in the text; an independent check of (4.3) is needed. I therefore retain the reader's CONDITIONAL verdict rather than moving to ACCEPT or REJECT.","tokens_in":45851,"tokens_out":55639,"duration_ms":561837,"concrete_test":"Independently verify Eq. (4.3) on a one-parameter hyperelliptic family, e.g., y^2 = x^2 - q or y^2 = x(x-1)(x-q). Compute the A-normalized B numerically from period matrices, form Ω_t = (∂_t y)|_x dx, and compare the fixed-x transported derivative ∇_t^x B(p,r) with the residue sum in (4.3) at generic p,r and several small q. Also rerun the paper's P1 check using B = dx1dx2/(x1-x2)^2; it should yield 0 = 0 and hence no sign information, confirming that the written check is vacuous. A nontrivial sign or coefficient mismatch against (4.3) would invalidate the Euler-defect cancellation and the gap theorem's proof.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The special-geometry identity ∂_t F_g^c = I_g^neck + R_g in Theorem 4.2 rests on the Rauch variational formula (4.3). The sign of that formula is load-bearing: Theorem 4.6 bounds the Euler defect E_g = H_g - t R_g only after the t I_g^neck terms cancel exactly. If the sign in (4.3) were opposite, E_g would contain +t I_g^neck, which carries the polar order t^{2-2g} and would destroy the boundedness used by Lemma 5.3. The paper's only explicit sign check is the P1 computation in Theorem 4.2, Step 2, with x(z) = z^2/2 + λ(t), y(z) = z, and B(z1,z2) = dz1dz2/(z1-z2)^2. But for the projection x, the normalized bidifferential on P1 is B = dx1dx2/(x1-x2)^2 = 4z1z2 dz1dz2/(z1^2-z2^2)^2, not dz1dz2/(z1-z2)^2. With the correct B, the family consists of mutually biholomorphic P1s and both sides of (4.3) vanish identically, so the computation provides no sign information. Thus the sign of (4.3) is asserted only via the quoted convention from [EO07], and the paper's own verification of that sign is invalid as written.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves a conifold gap theorem for Eynard–Orantin topological recursion on toric mirror curves. In the setting of a local analytic family of reduced compactified toric mirror curves acquiring a generic one-node degeneration, a good integer framing, and a conifold-adapted A-normalization, the paper defines an exact vanishing period t and shows that for every fixed g≥2 the conifold-polarized free energy has the form F_g^c(s,t) = B_{2g}/(2g(2g-2)) t^{2-2g} + H_g^gap(s,t) with H_g^gap jointly holomorphic in (s,t), no logarithmic term, and no intermediate negative powers. The proof combines a projection-compatible node normal form, uniform analytic estimates for the degenerating bidifferential and the stable correlators, an exact-period Gaussian leading limit, special geometry for the deformation with a bounded residual, and a Laurent-coefficient argument that uses the boundedness of an Euler defect; the universal coefficient is calibrated against the Gaussian one-matrix model. The separation of the argument into normal form, estimates, special geometry, and calibration is clear, and the paper explicitly covers both separating and nonseparating nodes.","tokens_in":46138,"tokens_out":34555,"duration_ms":337394,"significance":"If the result is correct, it is a substantial advance: it establishes the conifold gap for the whole class of toric mirror curves within the recursion, without closed-form BPS formulas, and it covers general transverse deformations with spectator moduli. The underlying mechanism is attractive: the Euler defect is bounded by an exact cancellation of the neck terms between the residue identity and the special-geometry identity, and the Gaussian comparison supplies the universal constant. The paper is careful with markings, framings, and orientation conventions. I do not share the concern that the P1 sign check in Theorem 4.2 Step 2 is invalid; on P1 the fundamental bidifferential in the global coordinate z is dz1dz2/(z1-z2)^2, and the residue computation checks the sign in (4.3). The main issue I found is an internal inconsistency in the calibration algebra (Eq. (5.7) versus Step 5 of Theorem 5.5), which is load-bearing for the universal constant. The proof does not ship machine-checked code, but the analytic estimates are presented in enough detail that the central line is checkable.","major_comments":[{"comment":"The calibration algebra is internally inconsistent. Eq. (5.7) states F^P_g = B_{2g}/(2g(2g-2)) T^{2g-2}; specializing to T=1/4 gives the factor 4^{-(2g-2)}, whereas the display in Step 5 of Theorem 5.5 uses the factor 4^{2g-2}. The two factors are reciprocals, so the displayed derivation of (5.4) does not follow from the stated input. If (5.7) is correct, the final coefficient would be different; if the final formula (5.4) is correct, then (5.7) should presumably read T^{2-2g}, which is also dimensionally natural for this family. Please correct the exponent and re-verify the coefficient. This is load-bearing because Corollary 5.2 is the calibration of the universal polar term in Theorem 1.1.","section":"Section 5.2, Eq. (5.7) and Step 5 of Theorem 5.5"},{"comment":"The uniform Neumann bound for the seam jump operator is the technical foundation for the O(1) bounds of Theorem 3.7 and the Gaussian limit of Theorem 3.10, but the proof invokes the fixed-component Cauchy-kernel jump construction of [GKN19, HN20] 'adapted to the present moving family' without proving the family-wise uniformities. In particular, the construction of the kernels K_{μ;q,s}(x,w) needs a global choice of base sections o_μ and of integration paths that vary holomorphically with (q,s), and the bound on the regular-part coefficients k_{m,n;e,e'}(q,s) near q=0 needs the uniform invertibility of the A-normalization on the capped family; the text asserts these bounds by Cauchy estimates. Please either provide the family version as a lemma with a proof or state precisely which theorem in [Yam80, HN20] supplies it. As written, this is a gap in the most delicate estimate of the paper.","section":"Section 3.1, Theorem 3.4 Step 1"}],"minor_comments":[{"comment":"The sentence 'The finite-set assertions concern all but finitely many integer f at the fixed base points 0' is unclear; presumably it means 'at the fixed base point (0,s_0)'.","section":"Proposition 2.1"},{"comment":"The notation warning for [ACPPRS12] objects (letters p, q, T and the hat convention) appears rather late in the proof; moving it before Step 1 would improve readability.","section":"Section 5.2"},{"comment":"The reference list contains a typo: 'Bohan F ang' should be 'Bohan Fang'.","section":"References"},{"comment":"The statement that a full loop of ε has monodromy T_γ^2 would benefit from a one-line justification using the plumbing relation z_+z_-=ε^2 and the orientation convention of Proposition 2.4.","section":"Theorem 4.3, Step 1"}],"recommendation":"major_revision","confidential_remarks":"The mathematical core is strong and the main theorem is plausible, but the calibration inconsistency and the need for a fuller family-wise proof of the sewing estimates justify major revision. The journal should also check that the AI-assistance statement and the references [LGS26, JGJ26] comply with its policies."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know this is the first proof of the conifold gap for arbitrary toric mirror curves, not another special-curve computation. Previous results covered local P^2, nine hypergeometric curves, and Painlevé I; this paper handles every fixed Newton polygon with a generic one-node degeneration, in both separating and nonseparating topologies, with general transverse deformations and spectator moduli. That is a real step forward.\n\nThe architecture is sound. The proof has distinct modules: a projection-compatible normal form using the vanishing period itself as the transverse coordinate, uniform scaled estimates for the bidifferential in the moving family, special geometry applied to the deformation form, a bounded Euler defect argument, and a Gaussian calibration. The Euler-defect logic is clean: exact cancellation of the neck term, boundedness of the remainder, and the Gaussian limit force the Laurent expansion to have exactly one polar term. The descent from the epsilon-cover to the t-disc via deck symmetry is carefully done.\n\nThe soft spots are proportionate to how complex the proof is. Section 3 is the part to scrutinize: the uniform Neumann-series bound on the seam jump operator assumes the Grushevsky–Krichever–Norton/Hu–Norton fixed-component construction extends to families with transverse variation. If that fails, the whole chain collapses. I did not find a concrete error there, but independent verification is warranted.\n\nThere is one real typo: in Theorem 5.5, the quoted ACPPRS12 free energy is written as F^P_g = B_{2g}/(2g(2g-2)) T^{2g-2}, which is wrong. The subsequent substitution T=1/4 uses the correct value 4^{2g-2}, so the final constant is right, but the displayed formula should be T^{2-2g}. The reader who spotted this is correct.\n\nThe stress-test note about the P1 sign check in Theorem 4.2 does not land. The B_t in that example is the intrinsic Bergman kernel of P1 in the z-coordinate, namely dz1 dz2/(z1-z2)^2, not the pullback dx1 dx2/(x1-x2)^2. The latter has spurious poles at the deck-involute points and is not the Bergman kernel. With the correct B, the residue computation and the fixed-x derivative both give \\dot{\\lambda} dz1 dz2/(z1^2 z2^2), and the sign checks out. So the Rauch formula is not unverified on this ground.\n\nThis paper deserves a serious referee. The right reader is someone working on topological recursion or conifold boundary conditions, and the referee should spend most of their time on Section 3. Send it to peer review; the typos are fixable and the central argument is substantial.","headline":"A genuinely general conifold gap theorem for toric mirror curves, proved by a dense but coherent analytic tour de force; the one real defect I found is a typo in the calibration constants, not a hole in the proof.","tokens_in":46680,"tokens_out":8259,"would_cite":true,"duration_ms":75961,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14J33","14H81","14N35","32G20"],"pacs":[],"model":"deepseek-v4-flash","headline":"The conifold gap theorem holds for topological recursion of every toric mirror curve: for each genus at least two, the free energy has one universal polar term and no other negative powers or logarithms.","keywords":["conifold gap","topological recursion","toric mirror curve","one-node degeneration","free energy","vanishing period","Euler defect","Gaussian matrix model"],"falsifier":"Compute, for one explicit toric one-node family with spectator moduli and genus $g=2$, the Laurent coefficients of $F_2^c(s,t)$ in $t$ by the recursion of the paper: if any negative coefficient other than the one at exponent $-2$ is nonzero, or if $t^2 F_2^c(s,t)$ does not converge to the Gaussian constant as $t\\to 0$, then Theorem 1.1 is false.","tokens_in":45653,"feed_emoji":"📐","tokens_out":7826,"duration_ms":74867,"temperature":0.7,"pith_summary":"The paper proves a conifold gap theorem for topological recursion of toric mirror curves. For every local analytic family whose fiber acquires a single ordinary node, and for every fixed genus $g \\ge 2$, the conifold-polarized free energy has exactly one universal polar term, and the remainder is jointly holomorphic in the transverse and spectator parameters. Concretely, after normalizing the vanishing period $t$, the Laurent expansion of $F_g^c(s,t)$ in $t$ contains no logarithmic term and no negative power other than $t^{2-2g}$. If the theorem is right, the gap supplies $2g-2$ boundary conditions at every conifold point of the moduli space for the whole toric class.","feed_headline":"Toric mirror curves obey a universal conifold gap at every genus","feed_subtitle":"Each genus≥2 free energy has one polar term, no logarithms, and a holomorphic remainder across the node.","key_machinery":"The load-bearing object is the one-neck normal form $v^2 = u^2 - \\lambda$ together with the exact vanishing period $t = c_{\\mathrm{per}} \\int_\\gamma \\omega_{0,1}$ used as the transverse coordinate. The proof controls the $A$-normalized bidifferential of the moving family by a capped outer family and a two-seam construction, inverting a seam jump operator by a Neumann series; then special geometry expresses $\\partial_t F_g^c$ as a neck integral plus a bounded residual. The bounded Euler defect $E_g = (2-2g)F_g^c - t\\partial_t F_g^c$ plus the Gaussian leading limit forces the Laurent gap in one assembly step.","core_discovery":"The central claim, Theorem 1.1, is that for every fixed integer $g \\ge 2$, after shrinking to a product $U_s^p \\times D_t$, there is a unique jointly holomorphic function $H_g^{gap}$ such that for $t \\neq 0$, $$F_g^c(s,t) = \\frac{B_{2g}}{2g(2g-2)}\\, $t^{{2-2g}}$ + $H_g^{{gap}}$(s,t).$$ In particular the Laurent expansion has no logarithm and no negative power other than $t^{2-2g}$. The result covers separating and nonseparating nodes, allows general transverse deformations beyond the pure filling-fraction case, and is uniform on compact sets of spectator moduli.","pith_inferences":["A natural extension: the same Laurent assembly would give one polar term per vanishing period for families acquiring several nodes simultaneously, provided the uniform bidifferential estimates extend to multi-neck degenerations.","The bounded Euler defect may be the more general phenomenon: wherever special geometry plus a Gaussian leading limit hold, boundedness of the defect is equivalent to the absence of logarithms and intermediate powers in the free energy.","The proof's explicit contours suggest the remainder $H_g^{gap}$ carries effective bounds, not just existence, so the gap could be turned into numerical predictions for higher-genus amplitudes near the conifold."],"forward_implications":["Every genus-at-least-two free energy of a toric mirror curve near a one-node degeneration is determined up to holomorphic data by its polar term; no logarithms or intermediate negative powers appear.","The polar coefficient is universal: it depends only on the period normalization and the recursion conventions, not on the Newton polygon, the spectator moduli, or whether the node is separating or nonseparating.","The gap gives $2g-2$ boundary conditions at each conifold point for holomorphic anomaly equations in the whole toric class.","Both node topologies and general transverse deformations are covered, so the result is not confined to the pure filling-fraction case.","After specializing the period normalization, the constant equals $B_{2g}/(2g(2g-2))$, matching the standard conifold-gap coefficient."],"supporting_citations":[{"why":"supplies the ordinary topological recursion, the normalized bidifferential, and the residue conventions the theorem runs on.","marker":"[EO07]"},{"why":"provides the sewing formulas for normalized bidifferentials on degenerating families used in the moving-family expansion.","marker":"[Yam80]"},{"why":"introduces the fixed-component Cauchy-kernel jump construction behind the seam operator.","marker":"[GKN19]"},{"why":"gives the holomorphic A-normalized form of the Cauchy-kernel jump construction used in the uniform bidifferential estimates.","marker":"[HN20]"},{"why":"carries the Gaussian one-matrix-model free-energy evaluation that calibrates the universal polar constant.","marker":"[ACPPRS12]"},{"why":"supplies the Bernoulli/Euler-characteristic value identifying the constant with $B_{2g}/(2g(2g-2))$.","marker":"[HZ86]"},{"why":"supplies the criterion that an empty vital set identifies logarithmic with ordinary recursion, which the framing hypothesis uses.","marker":"[HMO26]"}],"fun_headline_variants":["Conifold gap: one pole, no logs, every genus≥2","Universal pole, zero logs: genus≥2 toric mirror free energy","One polar term, holomorphic remainder: conifold gap at genus≥2","Toric mirror curves: conifold gap is universal with no logs","Conifold gap: universal pole, holomorphic rest for genus≥2"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument rests on one premise: the analytic control of the degenerating two-point kernel remains uniform as the family moves in the transverse and spectator directions. If that uniform control fails at any stage of the degeneration, the bounded correlators, the Gaussian limit, and the bounded Euler defect—and with them the gap—do not follow.","fun_headline_variants_meta":{"raw":{"variants":["Conifold gap: one pole, no logs, every genus≥2","Universal pole, zero logs: genus≥2 toric mirror free energy","One polar term, holomorphic remainder: conifold gap at genus≥2","Toric mirror curves: conifold gap is universal with no logs","Conifold gap: universal pole, holomorphic rest for genus≥2"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001142,"raw_usage":{"total_tokens":4647,"prompt_tokens":760,"completion_tokens":3887,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":376,"completion_tokens_details":{"reasoning_tokens":3787}},"tokens_in":376,"tokens_out":3887,"duration_ms":27695,"temperature":1.0,"reasoning_tokens":3787,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T00:23:07.677417+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute, for one explicit toric one-node family with spectator moduli and genus $g=2$, the Laurent coefficients of $F_2^c(s,t)$ in $t$ by the recursion of the paper: if any negative coefficient other than the one at exponent $-2$ is nonzero, or if $t^2 F_2^c(s,t)$ does not converge to the Gaussian constant as $t\\to 0$, then Theorem 1.1 is false.","supporting_citations":[],"review_version":1}