{"id":"c439ccc9-f4fd-473b-9684-ea9c5b22d981","arxiv_id":"2506.17854","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper proves a wall-crossing formula expressing quadratic Gromov-Witten invariants of a general fiber of a d-surgery del Pezzo degeneration in terms of those of the untwisted fiber, with correction terms in the Grothendieck-Witt group.","lead":"This paper proves a surgery formula that relates quadratic (bilinear-form valued) counts of rational curves on del Pezzo surfaces obtained from the same nodal degeneration. The formula makes arithmetic curve counting over arbitrary fields computable, extending real Welschinger and complex Gromov-Witten techniques to a richer, number-theoretic setting.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Positive-characteristic relative enumerativity is the least secure condition: Theorem 5.32 and Corollary 5.33 are conditional on it, and Theorem 6.13 omits it entirely.","rationale":"The reader's weakest_assumption (Hypothesis 5.11(2) in positive characteristic) is exactly the condition on which the conditional theorem depends. In characteristic 0 the paper gives a complete argument via Proposition 5.10, so the central advertised formula over k of characteristic 0 is not threatened. In positive characteristic the theorem is honestly stated as conditional, but the scope is narrower than the applications suggest: Theorem 6.13 omits the hypothesis and the characteristic restriction, and the proof of Theorem 5.32 contains a missing reference ('Remark 4.14') at the point where a diagonal representative of φ_d^{-1}(D) is needed. These are fixable but real defects, so I keep the reader's CONDITIONAL verdict rather than upgrading to ACCEPT. The recommended concrete test directly checks whether the imported tropical correspondence settles Hypothesis 5.11(2) for the toric examples; if it does, the positive-characteristic applications stand, and if not, Theorem 6.1 and 6.2 need a characteristic-0 restriction or an extra hypothesis. No ad hominem assessment is involved; the concern is about missing support in the argument.","tokens_in":46205,"tokens_out":13590,"duration_ms":140750,"concrete_test":"Apply [Tyo12, Theorem 6.2] to eS = F2 over F_p for every class D-iE appearing in φ_1^{-1}(aH) in Theorem 6.1, e.g. for a=2 and p=5, and verify from the theorem's hypotheses that each such class yields a finite, reduced evaluation map with curves transverse to E and only ordinary double points outside E. If the cited theorem does not cover all listed classes, or if any class on the Newton polygon of Figure 6.1 fails the transversality/ordinary-node conditions, then the positive-characteristic part of the applications is unsupported. Separately, insert the missing Remark 4.14 or replace the appeal with a direct argument that φ_d^{-1}(D) has a diagonal representative, and state the d-square case of Theorem 5.32 explicitly.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing condition is Hypothesis 5.11(2), relative enumerativity of the uninodal component eS. In characteristic 0, Proposition 5.10 supplies it universally, so the advertised Theorem 1.1 is safe. In positive characteristic, however, the paper imports this hypothesis from tropical correspondence theorems [Mik05], [Tyo12] and verifies it only for the toric surfaces of Sections 6.1 and 6.2. The main theorem 5.32 and Corollary 5.33 are therefore conditional in char p on an assumption not established for an arbitrary uninodal del Pezzo surface. The gap is visible in Section 6.3: Theorem 6.13 is stated as an unconditional result for cubic surfaces but neither assumes Hypothesis 5.11 nor restricts to characteristic 0, and relative enumerativity for the blow-up of P^2 at six points on a conic is not proved in positive characteristic. A second, smaller defect is the broken cross-reference in the proof of Theorem 5.32: 'By Remark 4.14' is used to choose a diagonal representative (D',(i,i)) of φ_d^{-1}(D), but no such remark appears in Section 4; as written this leaves the representative choice unjustified and the case d∈(k*)^2 is not separated. Neither defect contradicts the char-0 core, but both should be fixed before the positive-characteristic claims are accepted.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a degeneration (wall-crossing) formula for quadratic genus-0 Gromov–Witten invariants of del Pezzo surfaces. For a 1-nodal Lefschetz fibration X→Spec k[[t]] of del Pezzo surfaces and d∈k*, the d-surgery X(d) has general fiber Σ(d); Theorem 5.32 (with characteristic-0 special case Theorem 1.1) states that, under Hypothesis 5.11, N_{Σ(d),D,σ} = N_{Σ(1),D,σ} + (⟨2⟩−⟨2d⟩)∑_{j≥1}(−1)^j N_{Σ(1),D−jγ,σ}. The proof constructs explicit moduli spaces M_0(X(d),D)_σ(ep), proves they are finite étale over F[[t]], identifies their special-fiber curves (Lemma 5.17), and reduces the count to the enriched binomial identity of Corollary 4.8. Applications include rational quadrics, blow-ups of P^2 at two points, some cubic surfaces, and a Dehn-twist invariance result.","tokens_in":46398,"tokens_out":7483,"duration_ms":68862,"significance":"If the result holds, this is a substantial advance in quadratic enumerative geometry: it enriches the Abramovich–Bertram degeneration method with GW(k)-valued binomial coefficients, gives an arithmetic degeneration formula over perfect fields, and produces new computations (Tables 6.1 and 6.2) as well as a quadratic analogue of Dehn-twist invariance. The wall-crossing term ⟨2⟩−⟨2d⟩ is not a fitted constant: it is the trace form of the quadratic algebra defined by d, and the enriched binomial identity Corollary 4.8 is parameter-free. The characteristic-0 core is carefully argued and appears sound, with detailed proofs of finiteness and étaleness of the relevant moduli spaces.","major_comments":[{"comment":"The positive-characteristic status of the main theorem is not as unconditional as the exposition suggests. Hypothesis 5.11(2) requires the uninodal component eS to be relatively enumerative. In characteristic 0, Proposition 5.10 proves this universally, so Theorem 1.1 is safe. In positive characteristic, however, the paper imports relative enumerativity from tropical correspondence theorems [Mik05], [Tyo12] and verifies it only for the toric surfaces treated in Sections 6.1 and 6.2. Consequently Theorem 5.32 and Corollary 5.33 are conditional in characteristic p on an assumption that is not established for an arbitrary uninodal del Pezzo surface. The gap is visible in Section 6.3: Theorem 6.13 is stated as an unconditional result for cubic surfaces, with neither Hypothesis 5.11 nor a characteristic-0 restriction, while relative enumerativity for the blow-up of P^2 at six points on a conic is not proved in positive characteristic. This should be repaired either by proving the missing relative enumerativity or by making the positive-characteristic statements explicitly conditional and marking Theorem 6.13 accordingly.","section":"§5.1, Hypothesis 5.11(2); §6.3, Theorem 6.13"},{"comment":"The proof uses the sentence 'By Remark 4.14, this element is of the form (D',(i,i))', but no Remark 4.14 appears in Section 4. For d∉(k*)^2 the diagonal form follows from (3.1), but for d∈(k*)^2 the Picard group Pic Q(d) is Z×Z and the chosen representative of φ_d^{-1}(D) need not be diagonal. The subsequent construction D0=D'+iE and the reduction to (D0,(0,0)) are used essentially in the computation of the wall-crossing sum, so as written the proof has a genuine gap for d∈(k*)^2. This case should be separated; since the wall-crossing term vanishes when d is a square, a separate argument (e.g. Σ(d)≅Σ(1)) should make the formula immediate, but that argument is not currently written.","section":"§5.6, proof of Theorem 5.32"}],"minor_comments":[{"comment":"In the second displayed formula of Theorem 5.30, the twisted binomial coefficient is written inom{\\sigma'[\\sqrt d]/k}{j} although σ' is an ℓ-algebra; the proof uses inom{\\sigma'[\\sqrt d]/k(f'_1)}{j} with k(f'_1)=ℓ. The displayed formula should read /ℓ.","section":"§5.6, Theorem 5.30"},{"comment":"The characteristic assumption is printed as 'either 0 or at least 2 + 1', which is nonsensical, and the proof refers to 'characteristic at least d+1' with d the parameter of the surgery. The intended bound should be stated consistently.","section":"§6.2, Theorem 6.6"},{"comment":"The degree of σ is given as '3d−3' with d not defined in the statement; presumably '3a−3' (or the correct integer for the class ae0) is intended.","section":"§6.2, Corollary 6.9"},{"comment":"In the bulleted description of f0, the condition 'f^{-1}(E)=C_1∩C_2' should read 'f_0^{-1}(E)=C_1∩C_2'.","section":"§5.3, Lemma 5.17"},{"comment":"In the final formula of Corollary 5.33, the term on the right is printed as N_{σ(1),D,σ((t))}; this should be N_{Σ(1),D,σ((t))}.","section":"§5.6, Corollary 5.33"},{"comment":"The proof cites 'Lemma 3.15' for the description of φ_d^{-1}(D); the correct reference is Proposition 3.15.","section":"§5.6, proof of Theorem 5.30"}],"recommendation":"major_revision","confidential_remarks":"I agree with the conditional assessment in the reader's report. The characteristic-0 core of the paper is convincing and the moduli-space construction is careful, but the positive-characteristic claims, especially Theorem 6.13, need to be made conditional or proved. The missing 'Remark 4.14' cross-reference and the unseparated d∈(k*)^2 case are small but genuine proof gaps that should be fixed before publication. The paper is within the scope of the journal and deserves a major-revision decision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Hi — quick take on arXiv:2506.17854. The paper delivers what it advertises: a Grothendieck–Witt-valued Abramovich–Bertram wall-crossing formula for genus-0 quadratic Gromov–Witten invariants over perfect fields of characteristic ≠ 2. The complex and real cases are genuine specializations, and the genuinely new content — the enriched twisted binomial coefficients and the surgery argument that cancels counts on the uninodal component — is substantive.\n\nI read the proof of Theorem 5.32 carefully. The characteristic-0 core holds up: the moduli spaces are constructed, shown to be finite étale over the base, and the double-point discriminant computation checks out. The enriched binomial identity (Corollary 4.8) is proved through Serre's cohomological invariants and verified over finite fields in [CW24]; that is real, checkable evidence. As a route to computing quadratic invariants of rational del Pezzo surfaces of degree at least 7, the approach is clearly effective, and the Dehn twist invariance (Theorem 6.14) is a nice dividend.\n\nThe soft spot is exactly where the stress-test note points. Hypothesis 5.11(2), relative enumerativity of the uninodal component, is automatic in characteristic 0 by Proposition 5.10, but in positive characteristic it is imported from tropical correspondence theorems and checked only for the toric surfaces of Sections 6.1 and 6.2. Theorem 6.13 on cubic surfaces states an unconditional formula without assuming or proving this hypothesis. As written, that is a gap in the positive-characteristic claim. It may well be fixable — arguably the same toric techniques could be adapted — but it has to be acknowledged, not silently bypassed.\n\nTwo smaller defects: the proof of Theorem 5.32 cites “Remark 4.14” to choose a diagonal representative (D', (i,i)), but no such remark exists in Section 4; the d \\in (k^*)^2 case is not separated, and the representative choice is left unjustified. And Corollary 5.33 has a subscript typo (“N_{\\sigma(1)}” instead of “N_{\\Sigma(1)}”). Both are small fixes, but the missing reference is not merely cosmetic because it hides a running assumption in the main proof.\n\nBottom line: this is a genuine advance in quadratic enumerative geometry, and the characteristic-0 theorem deserves a serious referee. I would send it to peer review and ask the authors to either prove the positive-characteristic relative enumerativity or explicitly conditionalize Theorem 6.13 and the related statements, and to repair the broken cross-reference.","headline":"A substantive quadratic Gromov–Witten wall-crossing formula whose characteristic-0 core is sound; the positive-characteristic statements need an explicit hypothesis or proof before acceptance.","tokens_in":47029,"tokens_out":3742,"would_cite":true,"duration_ms":36069,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14N35","14F42","53D45","19G38"],"pacs":[],"model":"deepseek-v4-flash","headline":"Quadratic Gromov–Witten invariants of a del Pezzo surface obtained by a d-surgery differ from those of the untwisted surgery by an explicit wall-crossing sum, with coefficients in the Grothendieck–Witt group.","keywords":["A1-homotopy theory","Quadratic Gromov-Witten invariants","del Pezzo surfaces","Lefschetz fibration","Grothendieck-Witt group","enriched binomial coefficients","vanishing cycle","wall-crossing formula"],"falsifier":"Find a 1-nodal Lefschetz fibration of del Pezzo surfaces in positive characteristic whose special component is not toric, choose a class $D-iE$ appearing in $\\varphi_d^{-1}(D)$, and check whether a generic point configuration admits finitely many unramified rational curves in that class meeting $E$ transversely with ordinary double points away from $E$; if any of these conditions fails, Lemmas 5.17 and 5.19 and the wall-crossing identity lose their foundation, so the equality of Theorem 5.32 should fail for that fibration.","tokens_in":45918,"feed_emoji":"🧮","tokens_out":10816,"duration_ms":97344,"temperature":0.7,"pith_summary":"The paper proves a wall-crossing formula for quadratic genus-0 Gromov–Witten invariants of del Pezzo surfaces over perfect fields of characteristic not 2. It shows that when a del Pezzo surface is produced from a 1-nodal Lefschetz fibration by a d-surgery—an algebraic analogue of cutting and gluing along a vanishing cycle—the enriched count of rational curves of a given class equals the count on the untwisted surface plus a correction sum, with coefficients in the Grothendieck–Witt group. The correction is governed by new twisted binomial coefficients in $GW(k)$, and cancelling the contribution of the singular component reduces the computation to the surface of largest Picard group. This matters because the invariants give arithmetically meaningful counts over arbitrary fields, not only $\\mathbb{C}$ or $\\mathbb{R}$, and the formula makes them computable for rational quadrics, two-point blow-ups, and cubic surfaces.","feed_headline":"Quadratic curve counts on del Pezzo surfaces obey a wall-crossing law","feed_subtitle":"A surgery identity expresses enriched counts over any field and reduces quadrics, blow-ups, and cubics to known toric cases.","key_machinery":"The load-bearing device is the collection of twisted enriched binomial coefficients $\\binom{Y[Z]/X}{j}$ in the Grothendieck–Witt sheaf: the trace form of the finite \\'etale $X$-scheme associated to the set of $j$-element subsets of a degree-$2j$ \\'etale cover $Y$, with the Galois action twisted by the quadratic cover $Z\\to X$. The paper proves the binomial identity, Corollary 4.8, used to convert the count over curves on $\\widetilde{S}$ into the wall-crossing sum, and combines it with the surgery construction $X(d)$ whose special fiber is $\\widetilde{S}\\cup_E Q(d)$. The moduli spaces of stable maps to $X(d)$ are shown to be finite \\'etale over $F[[t]]$ when the enumerative hypotheses hold, so weights specialize without change.","core_discovery":"At the center of the paper is Theorem 5.32: for a 1-nodal Lefschetz fibration of del Pezzo surfaces over a perfect field $k$, with vanishing cycle $\\gamma$, and for $d\\in k^*$, every class $D\\in \\operatorname{Pic}\\Sigma(d)$ satisfying Hypothesis 5.11 satisfies $$N_{\\Sigma(d),D,\\$\\sigma$((t))}=N_{\\Sigma(1),D,\\$\\sigma$((t))}+(\\langle 2\\rangle-\\langle 2d\\rangle)\\sum_{j\\ge 1}(-1)^jN_{\\Sigma(1),D-j\\gamma,\\$\\sigma$((t))}$$ in $GW(k)$, after the injective base change to $GW(k((t)))$. Here $\\sigma$ is any finite \\'etale $k$-algebra of degree $-K_{\\Sigma(d)}\\cdot D-1$, $\\Sigma(d)$ is the general fiber of the $d$-surgery, and $\\Sigma(1)$ is the untwisted surgery. The proof counts rational curves on the special fiber $\\widetilde{S}\\cup_E Q(d)$, identifies the reducible specializations through an intersection profile, and uses an enriched Pascal identity to eliminate all terms involving $\\widetilde{S}$.","pith_inferences":["One consequence the authors leave implicit is that the whole $d$-dependence of these invariants is carried by the single Grothendieck–Witt element $\\langle 2\\rangle-\\langle 2d\\rangle$; if that element vanishes, every $d$-surgery has the same enriched counts.","A testable extension would be to run the same moduli-space argument for a degeneration with several nodes; the same enriched binomial cancellation should then produce a formula with multinomial coefficients and a sum over several vanishing cycles.","Over finite fields, the enriched Pascal triangle computed in [CW24] gives concrete numbers for the twisted binomial coefficients; combining them with the formula would produce finite-field predictions for $N_{Q(d),aH,\\sigma}$ that can be checked by independent tropical or deformation counts.","If relative enumerativity could be established for all uninodal del Pezzo surfaces in positive characteristic, not only the toric ones, Corollary 5.33 would give a uniform reduction of quadratic invariants of all rational del Pezzo surfaces to toric computations, realizing the paper's stated motto as a theorem."],"forward_implications":["For rational quadrics $Q(d)\\subset\\mathbb{P}^3_k$, Theorem 6.1 reduces $N_{Q(d),aH,\\sigma}$ to counts on $Q(1)=\\mathbb{P}^1\\times\\mathbb{P}^1$ with classes $(a\\pm j)e_1,(a\\mp j)e_2$, and the tables in Section 6.1 give explicit $GW(k)$ values for $a=1,2,3,4$.","For the blow-up of $\\mathbb{P}^2_k$ at two points, Theorem 6.6 reduces counts of class $ae_0-bf_1$ to counts on the split blow-up, and Corollary 6.8 expresses them through Gromov–Witten and real enumerative invariants.","For cubic surfaces obtained by smoothing the union of a conic and a cubic, Theorem 6.13 gives a wall-crossing identity for classes orthogonal to the vanishing cycle $\\gamma=2e_0-e_1-\\cdots-e_6$, whenever the twisted surface is $\\mathbb{A}^1$-connected.","Theorem 6.14 yields the arithmetic analogue of Dehn-twist invariance: $N_{\\Sigma(1),D,\\sigma}=N_{\\Sigma(1),D+(D\\cdot\\gamma)\\gamma,\\sigma}$, so reflecting a class across the vanishing cycle does not change the enriched count.","Combined with toric computations [JPP23, JPMPR25], the main theorem computes quadratic invariants of rational del Pezzo surfaces of degree at least 7, including point constraints defined over quadratic extensions of $k$."],"supporting_citations":[{"why":"Defines the quadratic genus-0 Gromov–Witten invariants and gives their enumerative meaning as enriched sums over rational curves.","marker":"[KLSW23a]"},{"why":"Provides the moduli of stable maps, the relative orientation, and the double-point locus used to compare weights under specialization.","marker":"[KLSW23b]"},{"why":"Introduces the original Abramovich–Bertram degeneration strategy for $F_2$ and $\\mathbb{P}^1\\times\\mathbb{P}^1$ that this paper adapts to quadratic invariants.","marker":"[AB01]"},{"why":"Gives the real Welschinger wall-crossing formula whose quadratic-form version Theorem 5.32 generalizes.","marker":"[Bru20]"},{"why":"Supplies the trace-form operations and the multiquadratic-reduction principle used to prove the enriched binomial identities.","marker":"[GMS03]"},{"why":"Provides the tropical correspondence theorem used to verify relative enumerativity in positive characteristic for toric special components.","marker":"[Mik05]"},{"why":"Extends the tropical correspondence to the toric stacks needed for the same positive-characteristic enumerativity checks.","marker":"[Tyo12]"},{"why":"Computes quadratic Gromov–Witten invariants of toric del Pezzo surfaces for split and multiquadratic $\\sigma$, giving the baseline for the applications.","marker":"[JPP23]"},{"why":"Extends the toric computations to multiquadratic extensions, used in the applications of Section 6.","marker":"[JPMPR25]"},{"why":"Constructs the proper moduli stacks of stable maps used throughout the deformation-invariance argument.","marker":"[AO01]"}],"fun_headline_variants":["Surgery formula for quadratic curve counts on del Pezzo surfaces","Quadratic Gromov-Witten invariants obey a surgery identity over any field","Enriched curve counts on del Pezzo surfaces: a quadratic Abramovich-Bertram law","From nodal smoothing to toric: quadratic GW invariants for all fields","Algebraic surgery cuts quadratic curve counts on del Pezzo surfaces"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The formula collapses if the singular component $\\widetilde{S}$ fails to be relatively enumerative, meaning that for some curve class $D-iE$ a generic point configuration admits only finitely many rational curves, each unramified, transverse to $E$, and with only ordinary double points away from $E$; this is automatic in characteristic 0 but imported from tropical correspondence in positive characteristic and checked only for the toric surfaces treated in Sections 6.1 and 6.2.","fun_headline_variants_meta":{"raw":{"variants":["Surgery formula for quadratic curve counts on del Pezzo surfaces","Quadratic Gromov-Witten invariants obey a surgery identity over any field","Enriched curve counts on del Pezzo surfaces: a quadratic Abramovich-Bertram law","From nodal smoothing to toric: quadratic GW invariants for all fields","Algebraic surgery cuts quadratic curve counts on del Pezzo surfaces"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001134,"raw_usage":{"total_tokens":4724,"prompt_tokens":974,"completion_tokens":3750,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":590,"completion_tokens_details":{"reasoning_tokens":3666}},"tokens_in":590,"tokens_out":3750,"duration_ms":26749,"temperature":1.0,"reasoning_tokens":3666,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T19:00:26.242590+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Find a 1-nodal Lefschetz fibration of del Pezzo surfaces in positive characteristic whose special component is not toric, choose a class $D-iE$ appearing in $\\varphi_d^{-1}(D)$, and check whether a generic point configuration admits finitely many unramified rational curves in that class meeting $E$ transversely with ordinary double points away from $E$; if any of these conditions fails, Lemmas 5.17 and 5.19 and the wall-crossing identity lose their foundation, so the equality of Theorem 5.32 should fail for that fibration.","supporting_citations":[],"review_version":1}