{"id":"7cc63e8e-08b7-47ff-b088-23605dcc651e","arxiv_id":"2412.00041","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The BCOV invariant of Camere-Garbagnati-Mongardi Calabi-Yau fourfolds is proportional to the author's equivariant torsion invariant and is expressed by Borcherds products in the Enriques and rational-curve-fixed-locus cases.","lead":"This paper proves that the BCOV invariant of certain Calabi-Yau fourfolds equals, up to a constant, an equivariant analytic torsion invariant defined earlier by the author for the underlying hyperkähler manifold with involution. In special cases it uses this to express the BCOV invariant as the Petersson norm of Borcherds products.","discovery_kind":"unification","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified: the comparison theorem is explicitly conditional and the proof's boundary argument is internally consistent.","rationale":"The reader identified the irreducibility of the discriminant divisor as the weakest assumption, and I agree that it is the crucial hypothesis on which the proof of Theorem 2.16 depends. However, I do not regard this as an objection: the theorem is explicitly conditional on it, and all applications in the paper verify it. I also checked the core curvature computation in §2.3, the comparison of Chern class identities in Lemmas 2.6-2.12, and the Euler-characteristic calculation in Proposition 2.14; no algebraic or sign error surfaced. The only genuine risk is the dependence on the companion preprints [19] and [20], which supply the construction of τ and the boundary lemmas used to derive (2.48)-(2.50). That risk is concrete but testable, and it does not by itself invalidate the paper's internal logic. Hence the reader's ACCEPT verdict should stand.","tokens_in":29061,"tokens_out":23884,"duration_ms":230136,"concrete_test":"Obtain [20, Lemma 4.2] and verify that it proves, exactly as used in (2.50), the following statement: a pluriharmonic function on Ω_{M0^\\perp} \\setminus D_{M0^\\perp} with local expansion a log|s|^2 + O(log log |s|^{-1}) near each point of D_{M0^\\perp} satisfies the global current equation dd^c log u = a δ_{\\bar D}. In particular, check that the O(log log) remainder contributes zero to the distributional dd^c and that the residue calculation in Theorem 2.16 is therefore justified.","verdict_should_be":"UNCHANGED","load_bearing_attack":"I find no load-bearing flaw in the central claim. Theorem 2.16 is stated under the explicit hypothesis that rk(M0) ≤ 17 and that the discriminant divisor \\bar D_{M0^\\perp} is irreducible (equivalently, the roots of M0^\\perp form one O(M0^\\perp)-orbit). This hypothesis enters at a specific and well-localized point: it upgrades the local pole asymptotics (2.49) to the global current equation dd^c log u = a δ_{\\bar D} in (2.50), and it makes the residue theorem step in the proof of Theorem 2.16 valid, since all intersection points of a chosen curve with the divisor then carry the same residue a. The O(log log |s|^{-1}) remainder in (2.49) is not a problem: after subtracting a log|s|^2, the remaining term is pluriharmonic on the punctured domain and has sub-logarithmic growth, hence extends across the divisor; the delta-current equation therefore follows from the stated asymptotics together with the cited [20, Lemma 4.2]. The theorem does not overclaim: the irreducibility assumption is stated clearly, and the applications in Section 3 satisfy it, via [24, (1.10)] for the Enriques case and via [34, 11.3] and [13, FIGURE 1] for the Λ_k(2)^\\perp cases. The main residual risk is reliance on the author's companion preprints [19] and [20], but those citations are specific, the supported steps are identified, and I found no circularity or hidden assumption in the present text.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper is the third in a series on equivariant analytic torsion for irreducible holomorphic symplectic fourfolds of K3[2]-type with antisymplectic involution. The main result, Theorem 2.16, shows that for a primitive hyperbolic 2-elementary sublattice M0 of the K3 lattice with rk(M0) ≤ 17 and irreducible discriminant divisor \\bar D_{M0^\\perp}, the BCOV invariant of the Camere–Garbagnati–Mongardi Calabi–Yau fourfold Z_Y agrees with the author's earlier invariant τ_{\\tilde M0,K}(X_Y,ι_Y) up to a constant depending only on M0. The proof compares the curvature equations satisfied by the two invariants (Theorem 2.4 and Eq. (2.1)), then analyses the boundary behaviour on the Baily–Borel compactification via residues. Applications in Section 3 express the BCOV invariant as the Petersson norm of the Borcherds Φ-function for the universal cover of the Hilbert scheme of two points on an Enriques surface (Theorem 3.2) and of the Borcherds products Φ_k for 2-elementary K3 surfaces of type Λ_k(2)^\\perp (Theorem 3.4).","tokens_in":29310,"tokens_out":5588,"duration_ms":60898,"significance":"If the result stands, this is a substantial contribution: it gives the first comparison between the BCOV invariant of Calabi–Yau fourfolds with moduli dimension greater than one and a torsion-type invariant, and it yields explicit Petersson-norm formulas for those BCOV invariants. The paper has no fitted parameters: the curvature comparison in Theorem 2.4 is a direct identity of characteristic forms, and the constant in Theorem 2.16 is obtained from rigidity of the pluriharmonic extension rather than from normalization. The main limitations are explicitly stated in the manuscript: the comparison is conditional on the irreducibility of \\bar D_{M0^\\perp}, and several load-bearing ingredients are quoted from the author's companion preprints [19] and [20] (Theorem 1.5, Theorem 1.7, Eq. (2.48), and the current equation (2.50)). These references are specific and the supported steps are identified, and I do not see circularity; independent verification is, however, currently limited by the availability of those preprints.","major_comments":[],"minor_comments":[{"comment":"There are several typographical errors, including 'in vaiants' in the abstract and 'donoted' in the paragraph preceding Lemma 2.8 in Section 2.3; these should be corrected.","section":"Abstract and Section 1"},{"comment":"The statement uses the notation r(M0) whereas the hypotheses elsewhere use rk(M0); the notation should be unified throughout.","section":"Section 2.4, Theorem 2.16"},{"comment":"The constant a0 is first introduced in the local degeneration statement (2.48) and then converted to a for the curve coordinate. The phrase 'Replacing t with tε(t)^{1/ν}' is acceptable, but it would help to spell out explicitly that a0 is rational and independent of the chosen degeneration, since that fact is used for the global residue argument.","section":"Section 2.4, Eq. (2.48)"},{"comment":"The introduction says the applications hold 'if every small deformation of Z remains a Calabi–Yau fourfold of Camere–Garbagnati–Mongardi.' The proofs of Theorems 3.2 and 3.4 verify this via [7, Theorem 5.1 and A], but the statement of these theorems does not mention this condition; the formulation should be made consistent so that the conditional nature is transparent.","section":"Section 3.1 and 3.2"},{"comment":"The sentence 'At least to the author, it is unclear what is the mirror of the Calabi–Yau 4-folds in Examples 3.1 and 3.3' is a useful honest caveat, but it is placed in the final discussion rather than in the introduction; consider moving or echoing it where the mirror-symmetry context is introduced.","section":"Section 3.2"}],"recommendation":"accept","confidential_remarks":"For the editor: the manuscript is technically detailed and the sampled algebraic identities are consistent. The main risk is the dependence on the author's own preprints [19] and [20] for the curvature equation of τ, the degeneration asymptotics, and the extension lemma used in (2.50). I did not find an internal inconsistency, but the acceptance of this paper is contingent on those preprints being made available in verifiable form. The reviewer's stress-test concern about reducibility of the discriminant divisor is addressed by the explicit hypothesis in Theorem 2.16 and by the verification of that hypothesis in the two applications."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Hi,\n\nThis paper deserves a real referee. The main theorem (2.16) is a genuine new result: under explicit lattice-theoretic hypotheses, the BCOV invariant of the Camere–Garbagnati–Mongardi Calabi-Yau fourfold agrees, up to a constant, with the author's equivariant analytic torsion invariant for the associated K3[2]-type manifold with involution. The applications—expressing the BCOV invariant as Petersson norms of Borcherds products for Enriques surfaces and for the Λ_k(2)^⊥ cases—are new and give the kind of explicit formula that is useful for genus-one mirror symmetry.\n\nThe proof is careful. The curvature equation for τ (Theorem 2.4) is checked through a sequence of Chern-class identities; I sampled Lemmas 2.5, 2.7, 2.9, 2.11, and Proposition 2.14, and the algebra is consistent. The overall strategy follows Yoshikawa's method: compare curvature equations, then analyze boundary behavior on the moduli space. The use of the irreducibility of the discriminant divisor to conclude that all boundary residues are equal and therefore vanish is legitimate; without that hypothesis the argument would not close, and the paper says so explicitly.\n\nThe main soft spot is the heavy reliance on the author's two earlier preprints, [19] and [20]. Theorems 1.5 and 1.7, which are the backbone of the comparison, are quoted from those papers. That is not a flaw in itself, but it means the correctness of this paper is contingent on results that have not yet appeared in final form. A referee should check those carefully. The second limitation is the irreducibility assumption on the discriminant divisor; it is satisfied in the applications but not derived. The paper is transparent about this, so it is a scope restriction rather than an error.\n\nOne more thing: the paper claims to be the first to give explicit BCOV formulas for Calabi-Yau fourfolds with moduli dimension greater than one. That claim is plausible, and the examples indeed have moduli dimensions from 4 to 10, so it holds in this setting.\n\nOverall, this is a solid, honest paper. The dependence on the preprints is a reason to be cautious, but not a reason to desk-reject. Send it to a referee who knows Yoshikawa's work and the BCOV literature.","headline":"A technically careful paper that proves a genuine comparison between two analytic torsion invariants and yields explicit Borcherds product formulas for the BCOV invariant of certain Calabi-Yau fourfolds.","tokens_in":29875,"tokens_out":5147,"would_cite":true,"duration_ms":45528,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14J32","14J28","58J52","14D07","11F55"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that the BCOV invariant of a Calabi-Yau fourfold obtained from a K3[2]-type manifold with antisymplectic involution equals the author's equivariant analytic torsion up to a constant, and derives Borcherds-product…","keywords":["equivariant analytic torsion","BCOV invariant","Calabi-Yau fourfold","K3[2]-type manifold","antisymplectic involution","Borcherds product","period domain","2-elementary K3 surface"],"falsifier":"Compute the quotient $u=\\tau_{\\mathrm{BCOV}}/\\tau_{\\widetilde M_0,K}$ along a one-parameter degeneration of 2-elementary K3 surfaces satisfying the hypotheses of Theorem 2.16; the theorem predicts that $\\log u$ has zero residue at the discriminant divisor, so $u$ tends to a nonzero constant. Observing a nonzero residue, or a ratio that varies between two points of the moduli space, would refute the comparison.","tokens_in":28808,"feed_emoji":"🧮","tokens_out":8874,"duration_ms":81056,"temperature":0.7,"pith_summary":"The paper tries to prove that two invariants that look unrelated actually coincide on a moduli space: the BCOV invariant of a Calabi-Yau fourfold obtained by resolving the quotient of a hyperkähler fourfold by an antisymplectic involution, and an equivariant analytic torsion invariant of the underlying K3[2]-type manifold with involution. The claimed statement is that, under a rank and irreducibility condition, one is a constant multiple of the other for every 2-elementary K3 surface of a fixed type. The proof compares curvature equations and then uses the boundary of the period domain to force the log-ratio to be constant. In the concrete examples covered, this turns the BCOV invariant into the Petersson norm of a reflective modular form $\\Phi$ or $\\Phi_k$. A sympathetic reader would care because it connects analytic torsion, mirror-symmetric BCOV invariants, and modular forms on period domains.","feed_headline":"Calabi-Yau invariant equals equivariant torsion up to a constant","feed_subtitle":"The BCOV invariant of these fourfolds turns into a Borcherds product on the moduli space.","key_machinery":"The mechanism is a curvature identity plus a boundary residue computation. For a family of K3[2]-type manifolds with antisymplectic involution, the invariant $\\tau_{M,K}$ satisfies $$dd^c\\log\\tau_{M,K}=\\sum_{k=0}^{8}(-1)^k\\omega_{H^k}-\\frac{\\chi}{12}\\omega_{WP},$$ and the BCOV invariant of the associated fourfold satisfies the same equation. The proof compares Chern forms of direct image bundles through the blowup isomorphism $H^q(X,\\Omega_X^p)^+\\oplus H^{q-1}(X^\\iota,\\Omega_{X^\\iota}^{p-1})\\cong H^q(Z,\\Omega_Z^p)$, together with hyperkähler identities for the $(-1)$-eigenspaces on $H^1(X,\\Omega_X^1)$. On the moduli space this makes $\\log u$ pluriharmonic away from the discriminant divisor; the boundary law $dd^c\\log u=a\\,\\delta_{\\bar D}$ and the residue theorem force $a=0$, so $\\log u$ extends over the compactification and is constant.","core_discovery":"The central claim is Theorem 2.16: if $M_0$ is a primitive hyperbolic 2-elementary sublattice of the K3 lattice with $\\operatorname{rk}(M_0)\\le 17$ and the roots of $M_0^\\perp$ form a single $O(M_0^\\perp)$-orbit, then there is a positive constant $C_{M_0}$, depending only on $M_0$, such that for every 2-elementary K3 surface $(Y,\\sigma)$ of type $M_0$, $$\\tau_{\\mathrm{BCOV}}(Z_Y)=C_{M_0}\\,\\tau_{\\widetilde M_0,K}(X_Y,\\iota_Y).$$ Here $Z_Y$ is the Calabi-Yau fourfold obtained as the crepant resolution of the quotient of the Hilbert scheme $X_Y=Y^{[2]}$ by the involution induced by $\\sigma$. The paper derives this by showing both invariants satisfy the same curvature equation, so their log-ratio is pluriharmonic off the discriminant divisor, and then uses the residue theorem to show the log-ratio extends to a constant on the Baily-Borel compactification.","pith_inferences":["If the discriminant divisor has several irreducible components, the same residue argument would yield one residue per component, so the natural expectation is a piecewise-constant ratio rather than a global constant; this could be checked by computing the ratio near each boundary component.","The equality of curvature equations suggests the two invariants live in the same determinant-line theory, so known birational invariance of the BCOV invariant would imply the paper's Conjecture 2.17 for equivariant torsion whenever a resolution exists.","For any mirror of these fourfolds, the Borcherds-product formulas predict that the genus-one Gromov-Witten series is a power of $\\Phi$ or $\\Phi_k$, giving a concrete way to search for a mirror."],"forward_implications":["For any Enriques surface $S$, the BCOV invariant of the universal cover of the Hilbert scheme of two points on $S$ equals a constant multiple of the Petersson norm of the automorphic form $\\Phi$.","For each $k=1,\\dots,6$, the BCOV invariant of the fourfold built from a 2-elementary K3 surface of type $\\Lambda_k(2)^\\perp$ equals $C_k\\|\\Phi_k([Y])\\|^{k+1}$.","The comparison shows the ratio $\\tau_{\\mathrm{BCOV}}/\\tau_{\\widetilde M_0,K}$ is constant along the whole moduli space, so the BCOV invariant carries no information beyond the equivariant torsion for these families.","These are explicit BCOV formulas on moduli spaces of dimension greater than one, beyond the previously known hypersurface cases."],"supporting_citations":[{"why":"Defines the equivariant analytic torsion invariant $\\tau_{M,K}$ and proves its curvature equation, which Theorem 2.4 matches to the BCOV equation.","marker":"[19]"},{"why":"Establishes the singularity expansion of $\\log\\tau$ at the discriminant divisor and the comparison with the K3-surface invariant $\\tau_{M_0}$ used in the applications.","marker":"[20]"},{"why":"Constructs the BCOV invariant for Calabi-Yau manifolds and provides the curvature equation (2.1) and the normalization factors.","marker":"[9]"},{"why":"Supplies the higher-dimensional BCOV/mirror-symmetry framework and the degenerating behavior used in the boundary analysis.","marker":"[10]"},{"why":"Constructs the Calabi-Yau fourfolds as crepant resolutions of quotients by antisymplectic involutions and identifies the Hodge structure of $Z_Y$.","marker":"[7]"},{"why":"Introduces the invariant $\\tau_{M_0}$ of 2-elementary K3 surfaces and proves its relation to the automorphic form $\\Phi$, a key input for the Enriques application.","marker":"[32]"},{"why":"Provides the reflective modular form $\\Phi_k$ and its Petersson-norm relation used for the $k=1,\\dots,6$ families.","marker":"[33]"}],"fun_headline_variants":["BCOV invariant equals equivariant torsion","BCOV equals equivariant torsion up to a constant","Equivariant torsion matches BCOV on CY 4-folds","BCOV and equivariant torsion agree"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof needs the discriminant divisor $\\bar D_{M_0^\\perp}$ to be irreducible, so that every intersection of a curve with the divisor gives the same residue; if the divisor has several components, the residue theorem no longer forces the log-ratio to be constant, and the comparison is not derived.","fun_headline_variants_meta":{"raw":{"variants":["BCOV invariant equals equivariant torsion","BCOV equals equivariant torsion up to a constant","Equivariant torsion matches BCOV on CY 4-folds","BCOV and equivariant torsion agree"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001104,"raw_usage":{"total_tokens":4598,"prompt_tokens":936,"completion_tokens":3662,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":552,"completion_tokens_details":{"reasoning_tokens":3598}},"tokens_in":552,"tokens_out":3662,"duration_ms":30785,"temperature":1.0,"reasoning_tokens":3598,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T14:45:17.065367+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the quotient $u=\\tau_{\\mathrm{BCOV}}/\\tau_{\\widetilde M_0,K}$ along a one-parameter degeneration of 2-elementary K3 surfaces satisfying the hypotheses of Theorem 2.16; the theorem predicts that $\\log u$ has zero residue at the discriminant divisor, so $u$ tends to a nonzero constant. Observing a nonzero residue, or a ratio that varies between two points of the moduli space, would refute the comparison.","supporting_citations":[{"cited_title":"Analytic torsion for irreducible holomorphic symplectic fourfolds with involution, I: Construction of an invariant","cited_arxiv_id":"2406.18023","evidence_quote":"Defines the equivariant analytic torsion invariant $\\tau_{M,K}$ and proves its curvature equation, which Theorem 2.4 matches to the BCOV equation."},{"cited_title":"Analytic torsion for irreducible holomorphic symplectic fourfolds with involution, II: the singularity of the invariant (with an Appendix by Ken-Ichi Yoshikawa)","cited_arxiv_id":"2411.13911","evidence_quote":"Establishes the singularity expansion of $\\log\\tau$ at the discriminant divisor and the comparison with the K3-surface invariant $\\tau_{M_0}$ used in the applications."},{"cited_title":"Eriksson, G","cited_arxiv_id":null,"evidence_quote":"Constructs the BCOV invariant for Calabi-Yau manifolds and provides the curvature equation (2.1) and the normalization factors."},{"cited_title":"Pi 10 (2022), Paper No","cited_arxiv_id":null,"evidence_quote":"Supplies the higher-dimensional BCOV/mirror-symmetry framework and the degenerating behavior used in the boundary analysis."},{"cited_title":"Camere, A","cited_arxiv_id":null,"evidence_quote":"Constructs the Calabi-Yau fourfolds as crepant resolutions of quotients by antisymplectic involutions and identifies the Hodge structure of $Z_Y$."},{"cited_title":"Yoshikawa, K3 surfaces with involution, equivariant analytic torsion, a nd automorphic forms on the moduli space , Invent","cited_arxiv_id":null,"evidence_quote":"Introduces the invariant $\\tau_{M_0}$ of 2-elementary K3 surfaces and proves its relation to the automorphic form $\\Phi$, a key input for the Enriques application."},{"cited_title":"328, 351–389","cited_arxiv_id":null,"evidence_quote":"Provides the reflective modular form $\\Phi_k$ and its Petersson-norm relation used for the $k=1,\\dots,6$ families."}],"review_version":1}