{"id":"55f93fbd-3fc9-4267-8cb0-4a464edf3c85","arxiv_id":"2411.12931","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The Picard group of the Baily-Borel compactification of the moduli space of quasi-polarized K3 surfaces is Z, spanned by the extended Hodge line bundle.","lead":"This paper proves that the Baily-Borel compactification of the moduli space of quasi-polarized K3 surfaces has Picard group Z, the smallest possible, generated by the Hodge line bundle. It gives a general method for computing Picard groups of boundary compactifications of orthogonal Shimura varieties, with applications to moduli of Calabi-Yau pairs.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Basis problem (Thm. 4.5) is load-bearing and its odd-rank proof is only sketched; a gap or counterexample would break Thm. 6.4 and hence the main theorem.","rationale":"The reader identified the basis problem, Theorem 4.5, as the weakest assumption, and I concur: it is the unique step that upgrades the obstruction space to the full cusp space, and its odd-rank proof is delegated to an unpublished preprint and a 'very similar' sketch. I agree with the reader's assessment that this is not an observed mathematical error but a gap in verifiability, which justifies a conditional rather than an unconditional acceptance. I note that the omitted determinant-zero h=0 case of Theorem 4.4 is not load-bearing for the applications in Theorem 6.4, since all relevant weights satisfy h=2, but the odd-rank case of Theorem 4.5 is directly used for the negative definite lattices L' of rank n−2 with n>8, and these have odd signature. The concrete test above would settle the issue: it asks for the missing constructive step to be supplied, or a counterexample to be exhibited. If the construction works, the proof is complete; if not, the central argument has a genuine hole. This does not change the reader's conditional verdict, so I recommend UNCHANGED.","tokens_in":33921,"tokens_out":13994,"duration_ms":144253,"concrete_test":"Complete the proof of Theorem 4.5 for sign(M) odd by explicitly verifying the analogue of [32, Lem. 6.6] in odd rank: for a positive definite even lattice M of rank r>6 with M⊗Z_p splitting a hyperbolic plane for all p, show that there exists a sublattice L⊆M of index N (the level of M) with G_L≅G_M⊕G' and an isotropic subgroup H≤G_L with H⊥/H≅G_M, and then check the two bullet conditions of Proposition 3.7 for every p|N, including the p=2 case. If this verification succeeds, Theorem 4.5 follows and the concern is resolved; if it fails for some lattice appearing in §6 (e.g., a rank-8 lattice with G_M≅(Z/2)^6), Theorem 6.4 and consequently Theorem 1.1 would need to be re-examined.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim Pic(F̄_g)=Z depends on Theorem 6.4, whose key step (Proposition 5.13) requires equality of the obstruction space with the full cusp space. This equality is obtained from the basis problem, Theorem 4.5: for a positive definite even lattice M of rank r>6 with M⊗Z_p splitting a hyperbolic plane for all primes p, the cusp forms of weight k≥r/2 are spanned by theta series. This theorem is load-bearing: if it fails, the obstruction space may be strictly smaller than Cusp_k(ρ_M^*), and the Hodge-bundle criterion (Theorem 5.10) would not force every extendable Heegner class to be proportional to λ. The proof of the odd-signature case (sign(M) odd) is only sketched in §4.3, which says the argument is 'very similar' to [32, Thm. 6.7] and refers to the same author's preprint [32] for the even case. The adaptation hinges on (ii) of the proof: the construction in [32, Lem. 6.6] of an index-N sublattice L⊆M with G_L≅G_M⊕G' and an isotropic subgroup H with H⊥/H≅G_M, together with verification of the conditions (∗_p) of Proposition 3.7 for all p|N. This is not written down for odd rank and is exactly where a parity or local-splitting obstruction could invalidate the argument. The omitted determinant-zero h=0 case of Theorem 4.4 is secondary here—all lattices in Theorem 6.4 have h=2—but it leaves Theorem 4.5 as stated unproven. If Theorem 4.5 fails for an odd-rank lattice arising in the p-elementary families, then Cusp^obs is strictly smaller than Cusp, and the proof of Theorem 6.4 collapses.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the rational Picard group of Baily–Borel compactifications of orthogonal Shimura varieties and proves that, for K3-type lattices and for certain p-elementary lattices of large rank, the Heegner Picard group is one-dimensional, spanned by the extended Hodge line bundle. The main geometric consequence is Theorem 1.1: the Picard group of the Baily–Borel compactification of the moduli space of quasi-polarized K3 surfaces is isomorphic to Z. The proof introduces an arithmetic obstruction space for extending Heegner divisors to the boundary, relates it to spaces of theta series, and then uses a basis-problem statement for vector-valued modular forms to show that the obstruction space is maximal. Applications include Picard-group computations for moduli spaces of K3 surfaces with non-symplectic involutions and a torsion-freeness result for the Picard group of K3-type Shimura varieties.","tokens_in":34225,"tokens_out":8067,"duration_ms":80371,"significance":"If the main theorems are correct, Theorem 1.1 is a clean and surprising contrast to the curve case, and the general framework for computing Picard groups of Baily–Borel compactifications is valuable. The paper is well structured, the geometric–arithmetic dictionary is clearly explained, and the extension to p-elementary lattices and to boundary polarized Calabi–Yau pairs is a substantial contribution. The reliance on the same-author preprint [32] for the even-rank basis problem is stated transparently and is not circular. However, the odd-rank basis problem, which is load-bearing for the K3 application, is only sketched in the present manuscript, so the central claim is not yet fully verified in the text.","major_comments":[{"comment":"The odd-signature case of the basis problem is load-bearing for the main theorem: it is applied in Theorem 6.4 to the negative definite lattice L of rank n−2, which is odd when n is odd, and for the K3 lattice Λ_g one has n=17 and rank L=15. The proof of Theorem 4.5 in the odd-signature case consists of a sketch: the construction of the sublattice L with G_L ≅ G_M ⊕ G′ and the verification of conditions (∗_p) are delegated to [32, Lem. 6.6] and described as \"very similar.\" This is exactly the step where a parity or local-splitting obstruction could make the argument fail; if Cusp^θ is strictly smaller than Cusp, then the equality Cusp^obs = Cusp in Theorem 6.4 and hence Theorem 1.1 would not follow. Please provide the full odd-rank proof, or a precise reduction to a stated theorem in [32] that includes all parity details.","section":"§4.3, Theorem 4.5"},{"comment":"The proof of Theorem 4.4 explicitly omits the determinant-zero case when h=0, saying \"we omit the details.\" Since Theorem 4.5 is stated for all k ≥ r/2, including h=0, and its proof invokes Theorem 4.4 without restriction, the printed statement of Theorem 4.5 is not fully proven as stated. For the main geometric application one has h=3, so this omission does not by itself invalidate Theorem 1.1; nevertheless, the theorem as stated needs a completed proof or an added hypothesis excluding h=0.","section":"§4.2, Theorem 4.4"},{"comment":"In the 2-elementary case, the proof that the isotropic lifts of the relevant cusp spaces span Cusp_k(ρ_M^*) combines the decomposition from [33, Thm. 5.1] with the assertion that every non-characteristic isotropic subgroup H with H^⊥/H ≅ G_L is associated with an admissible isotropic plane. The latter is justified through Remark 6.5, which only sketches the order-4 conjugacy statement. Since this correspondence is what makes the p-elementary case, and hence Corollary 1.5, go through, the missing details should be supplied or replaced by a precise reference.","section":"§6.2, Theorem 6.4, Case 2"}],"minor_comments":[{"comment":"The displayed formula for the rank r_g is difficult to parse because of the large fraction and Jacobi-symbol notation; a reference to the original formula in [8] or [19] would help the reader.","section":"§1.1, Eq. (1.1)"},{"comment":"The reference list gives the second author of [9] only as \"Liu\"; the full name should be completed.","section":"References, [9]"},{"comment":"Affiliation 3 says \"Technical University of Darmstadt, Berlin\"; TU Darmstadt is located in Darmstadt, so this is likely a typo.","section":"Author affiliations"},{"comment":"The phrase \"which only split a hyperpbolic plane\" contains a typo: \"hyperpbolic\" should be \"hyperbolic.\"","section":"Remark 1.4"},{"comment":"The sentence \"the classification result in [1] shows that Λ_ρ is a 2-elementary lattice and splits a hyperbolic plane\" should give a precise location in [1] or [35], since this splitting is used to apply Corollary 6.7.","section":"Proof of Corollary 1.5"}],"recommendation":"major_revision","confidential_remarks":"The main theorem depends on the odd-rank basis problem, Theorem 4.5, whose proof is only sketched and refers to the same-author preprint [32]. This is not circular, but it does mean that a central input is not independently verified within the refereed manuscript. I would recommend that the editor require either a complete odd-rank proof in the paper or a public and citable version of [32] with the exact statement used here. The paper's scope and ambitions are appropriate for a top journal; the issue is one of verification, not of novelty."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Main thing to know: this paper gives a strong, credible argument that the Picard group of the Baily–Borel compactification of the quasi-polarized K3 moduli space is Z, and it develops a general framework for orthogonal Shimura varieties. If correct, this is the natural completion of earlier work by Bergeron–Li–Millson–Mœglin and Di Lorenzo–Fringuelli–Vistoli, and it makes a sharp contrast with the curve case.\n\nWhat is genuinely new and good: Theorem 1.1 is new, and the architecture is sensible. The paper constructs an obstruction space from admissible isotropic lifts of theta series, shows under arithmetic hypotheses that this obstruction space equals the full cusp space, and then uses the Hodge-bundle criterion to force every extendable Heegner class to be proportional to the Hodge bundle. The p-elementary cases and the corollary for Calabi–Yau pair moduli spaces broaden the reach. The torsion-freeness result extends [19], and the appendices supply useful details on the Siegel–Weil formula and Hecke kernels. There is no fitting, no parameter tuning, and no circularity; the self-citation to [32] is legitimate because that preprint supplies the even-rank basis problem.\n\nThe soft spot is exactly where the reader put it: Theorem 4.5, the basis problem in odd rank, is load-bearing, and its proof in §4.3 is a sketch. The text says the argument is “very similar” to [32, Thm. 6.7] and delegates the even case to an arXiv preprint by one of the authors. The odd-rank adaptation has to check the local splitting conditions and the isotropic-subgroup construction, and those checks are not written out. Theorem 4.4 also omits the determinant-zero h = 0 case, though that case is not needed for the main application. None of this is an observed mathematical error, and the sketch is plausible; a parity or local-splitting obstruction could hide there, but I do not see one. Still, as written, the central equality Cusp^obs = Cusp is not independently verifiable.\n\nThis paper is for moduli theorists and arithmetic geometers with some tolerance for vector-valued modular forms. It deserves a serious referee. I would send it to review, with the instruction that the referee check Theorem 4.5 in the odd-rank case, or require the authors to expand §4.3 and make the reduction to [32] precise. If that holds up, this is a clean and important theorem.","headline":"A likely correct and genuinely new computation of Pic(F̄_g) = Z whose main risk is the sketched odd-rank basis problem, which a referee should push to be written out.","tokens_in":34843,"tokens_out":2684,"would_cite":true,"duration_ms":32133,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14J28","14J15","11F27","11F37","14G35"],"pacs":[],"model":"deepseek-v4-flash","headline":"The Baily–Borel compactification of the moduli space of quasi-polarized K3 surfaces has Picard group isomorphic to ℤ, generated by a multiple of the extended Hodge line bundle.","keywords":["Picard group","Baily-Borel compactification","K3 surfaces","orthogonal Shimura varieties","Heegner divisors","theta series","vector-valued modular forms","basis problem"],"falsifier":"Compute, for a positive definite even lattice $L$ of odd rank $r > 6$ with $L \\otimes \\mathbb{Z}_p$ splitting a hyperbolic plane for every prime $p$, the orthogonal complement of the $\\theta$-series subspace inside $\\mathrm{Cusp}_{r/2}(\\rho_L)$: if it contains a nonzero cusp form, Theorem 4.5 is false, and the maximality of the obstruction space needed for the $p$-elementary cases collapses. A natural test case is a $2$-elementary lattice of rank 11 with discriminant group $(\\mathbb{Z}/2\\mathbb{Z})^n$.","tokens_in":33658,"feed_emoji":"🔢","tokens_out":15418,"duration_ms":121063,"temperature":0.7,"pith_summary":"This paper proves that the Baily–Borel compactification $\\overline{F}_g$ of the moduli space of quasi-polarized K3 surfaces of genus $g$ has the smallest possible Picard group: $\\mathrm{Pic}(\\overline{F}_g) \\cong \\mathbb{Z}$, generated by a multiple of the extended Hodge line bundle. The proof establishes a broader statement: for even lattices $M$ of signature $(2,n)$ with $n > 8$ that are of K3 type or $p$-elementary, the rational Picard group of the Baily–Borel compactification of the associated orthogonal Shimura variety is one-dimensional, spanned by the Hodge class. The central mechanism is an arithmetic obstruction space built from $\\theta$ series and their isotropic lifts, which controls exactly when Heegner divisors extend across the boundary. The paper shows this obstruction space is maximal by resolving the basis problem for vector-valued half-integral weight cusp forms. Because the global Torelli theorem identifies $F_g$ with such a Shimura variety, the rigidity theorem applies and answers a natural question about the divisor theory of the compactification.","feed_headline":"K3 moduli's Baily-Borel compactification has Picard group ℤ","feed_subtitle":"Smallest possible divisor group: every Cartier divisor is a multiple of the extended Hodge line bundle.","key_machinery":"The load-bearing object is the obstruction space $\\mathrm{Cusp}^{\\mathrm{obs}}_{(n+2)/2}(\\rho^*_M)$: the span, inside the cusp forms of weight $(n+2)/2$ for the dual Weil representation, of all isotropic lifts $\\uparrow_{H_J}(\\sigma^* \\Theta_{J^\\perp/J,F})$ attached to admissible isotropic planes $J$ in $M$ and degree-2 harmonic polynomials $F$. Heegner divisors fail to extend to the boundary exactly when they pair nontrivially with such lifts (Theorem 5.11), so if this space exhausts the full cusp space, every extendable Heegner class must annihilate all cusp forms and hence, by the Hodge bundle criterion (Theorem 5.10), be proportional to $\\lambda$. The maximality of the obstruction space is proved by the basis problem (Theorem 4.5): for a positive definite even lattice of rank $r > 6$ whose localization at every prime splits a hyperbolic plane, every cusp form of weight $k \\geq r/2$ is a $\\theta$ series. The proof of that basis result runs through a $\\theta$-lifting map $\\Psi$ built from the vector-valued Siegel–Eisenstein series, identified with a weighted sum of the Hecke operators of Section 3 (Theorem 4.4), whose non-vanishing is controlled by L-series.","core_discovery":"The central discovery is that the Baily–Borel compactification of the quasi-polarized K3 moduli space carries no new line bundles beyond the Hodge class: for every genus $g \\geq 2$, $\\mathrm{Pic}(\\overline{F}_g) \\cong \\mathbb{Z}$, spanned by some positive multiple of the extended Hodge line bundle $\\lambda$. The proof goes through a general result (Theorem 6.4 and Corollaries 6.7, 6.10): if $M$ is an even lattice of signature $(2,n)$ with $n > 8$ that either splits two hyperbolic planes and splits three locally at every prime (K3 type), or is $p$-elementary and splits one hyperbolic plane, then every Heegner divisor class that extends to the Baily–Borel boundary is rationally proportional to $\\lambda$. For the stable orthogonal group, the integral Picard group is then exactly $\\mathbb{Z}$, using a new torsion-freeness result for K3-type lattices of rank greater than 10. In the special case of the K3 lattice $\\Lambda_g = \\langle 2-2g \\rangle \\oplus E_8(-1)^{\\oplus 2} \\oplus U^{\\oplus 2}$, the global Torelli theorem identifies $F_g$ with the Shimura variety $\\mathrm{Sh}(\\Lambda_g)$, and Noether–Lefschetz divisors become Heegner divisors, so the general theorem yields the $\\mathbb{Z}$-Picard group. The paper also deduces that the rational Picard group of the normalization of the moduli space of boundary polarized Calabi–Yau pairs arising from K3 surfaces with non-symplectic involutions is spanned by the CM line bundle, outside one 11-dimensional family.","pith_inferences":["Beyond the paper: the same obstruction-space method should compute Picard groups for other Baily–Borel compactifications realized by orthogonal Shimura varieties, such as moduli of cubic fourfolds or of hyper-Kähler varieties of K3 type, whenever the lattice satisfies the local hyperbolic-splitting conditions; the paper proves the K3-type and $p$-elementary cases but does not claim a general class","Beyond the paper: Theorem 4.5 is a statement in pure modular forms—vector-valued cusp forms of weight at least half the rank are theta series for locally hyperbolic-splitting lattices; if it survives in odd rank, it gives an Eichler-type basis result independent of the geometry.","A computationally testable consequence: for a low-rank $p$-elementary lattice satisfying the hypotheses, the constant term of the restriction of any extendable Heegner divisor to a boundary component should vanish identically; a single counterexample there would bound the generality of Theorem 6.4."],"forward_implications":["$\\mathrm{Pic}(\\overline{F}_g) \\cong \\mathbb{Z}$ for every genus $g \\geq 2$: the Baily–Borel boundary adds no new Cartier divisor classes, and the Noether–Lefschetz divisors that generate $\\mathrm{Pic}(F_g)$ all collapse to multiples of $\\lambda$ at the boundary.","For any K3-type lattice of signature $(2,n)$ with $n > 8$, $\\mathrm{Pic}_\\mathbb{Q}(\\mathrm{Sh}_\\Gamma(M))^{\\mathrm{Heegner}} \\cong \\mathbb{Q}$ and, when $\\Gamma = \\widetilde{O}(M)$, $\\mathrm{Pic}(\\overline{\\mathrm{Sh}}(M)) \\cong \\mathbb{Z}$.","For $p$-elementary lattices with $\\Gamma = \\Gamma_0$, the same one-dimensional rational Picard group holds, covering moduli spaces of lattice-polarized K3 surfaces and Calabi–Yau pairs that do not split two hyperbolic planes.","The integral, torsion-free statement holds for K3-type lattices of rank greater than 10, so the one-dimensionality is an honest isomorphism, not merely a rational statement.","A by-product: the normalization of the boundary polarized Calabi–Yau moduli space of K3 surfaces with non-symplectic involution has rational Picard group spanned by the CM line bundle, outside the 11-dimensional exceptional family."],"supporting_citations":[{"why":"Shows that the rational Picard group of an orthogonal Shimura variety is spanned by natural divisors and, when M contains two hyperbolic planes, by Heegner divisors (Theorem 5.4).","marker":"[8]"},{"why":"Supplies the local Borcherds product criterion for a Heegner divisor combination to be trivial at generic boundary points (Theorem 5.11).","marker":"[14]"},{"why":"Proves the basis problem for even-rank lattices and gives the Hecke-operator identity and L-value non-vanishing that are adapted here to odd rank.","marker":"[32]"},{"why":"Provides the modularity of theta series and the Borcherds products underlying the geometric-arithmetic dictionary.","marker":"[10]"},{"why":"Identifies the Heegner subspace with the dual of almost cusp forms via coefficient functionals (Theorem 5.8).","marker":"[16]"},{"why":"Defines the vector-valued Hecke operators whose invertibility drives the proof of the basis problem.","marker":"[17]"},{"why":"Supplies the discriminant-form theory, the classification and local splitting criteria for p-elementary lattices, and the surjectivity of the natural map from O(M) to O(G_M).","marker":"[35]"},{"why":"The torsion-freeness argument for the Picard group of K3-type Shimura varieties extends this prior work.","marker":"[19]"},{"why":"Decomposes arbitrary cusp forms into isotropic lifts from subgroups, the input that spans the obstruction space in the elementary cases.","marker":"[33]"},{"why":"Computes the abelianization of the stable orthogonal group, pinning down the torsion of the Picard group of the quotient stack.","marker":"[22]"}],"fun_headline_variants":["K3 moduli Baily-Borel Picard group is Z","No extra line bundles on K3 compactification","K3 moduli boundary: all divisors are Hodge multiples","Picard group of K3 compactification is trivial beyond Hodge","K3 moduli compactification: Picard = Z from Hodge class"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the basis problem: every cusp form of weight at least half the lattice rank is a linear combination of theta series for lattices of rank greater than 6 that locally split a hyperbolic plane, and the paper proves this in full only when the rank is even, with the odd-rank case sketched and the determinant-zero case of weight equal to half the rank omitted.","fun_headline_variants_meta":{"raw":{"variants":["K3 moduli Baily-Borel Picard group is Z","No extra line bundles on K3 compactification","K3 moduli boundary: all divisors are Hodge multiples","Picard group of K3 compactification is trivial beyond Hodge","K3 moduli compactification: Picard = Z from Hodge class"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000852,"raw_usage":{"total_tokens":3800,"prompt_tokens":1138,"completion_tokens":2662,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":754,"completion_tokens_details":{"reasoning_tokens":2574}},"tokens_in":754,"tokens_out":2662,"duration_ms":19412,"temperature":1.0,"reasoning_tokens":2574,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T17:02:53.648929+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute, for a positive definite even lattice $L$ of odd rank $r > 6$ with $L \\otimes \\mathbb{Z}_p$ splitting a hyperbolic plane for every prime $p$, the orthogonal complement of the $\\theta$-series subspace inside $\\mathrm{Cusp}_{r/2}(\\rho_L)$: if it contains a nonzero cusp form, Theorem 4.5 is false, and the maximality of the obstruction space needed for the $p$-elementary cases collapses. A natural test case is a $2$-elementary lattice of rank 11 with discriminant group $(\\mathbb{Z}/2\\mathbb{Z})^n$.","supporting_citations":[{"cited_title":"Math.208(2017), no","cited_arxiv_id":null,"evidence_quote":"Shows that the rational Picard group of an orthogonal Shimura variety is spanned by natural divisors and, when M contains two hyperbolic planes, by Heegner divisors (Theorem 5.4)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the local Borcherds product criterion for a Heegner divisor combination to be trivial at generic boundary points (Theorem 5.11)."},{"cited_title":"Borcherds,Automorphic forms with singularities on Grassmannians, Invent","cited_arxiv_id":null,"evidence_quote":"Provides the modularity of theta series and the Borcherds products underlying the geometric-arithmetic dictionary."},{"cited_title":"Algebra397(2014), 315–","cited_arxiv_id":null,"evidence_quote":"Identifies the Heegner subspace with the dual of almost cusp forms via coefficient functionals (Theorem 5.8)."},{"cited_title":"Z.264(2010), no","cited_arxiv_id":null,"evidence_quote":"Defines the vector-valued Hecke operators whose invertibility drives the proof of the basis problem."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the discriminant-form theory, the classification and local splitting criteria for p-elementary lattices, and the surjectivity of the natural map from O(M) to O(G_M)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The torsion-freeness argument for the Picard group of K3-type Shimura varieties extends this prior work."},{"cited_title":"Number Theory263(2024), 206–233","cited_arxiv_id":null,"evidence_quote":"Decomposes arbitrary cusp forms into isotropic lifts from subgroups, the input that spans the obstruction space in the elementary cases."},{"cited_title":"Gritsenko, K","cited_arxiv_id":null,"evidence_quote":"Computes the abelianization of the stable orthogonal group, pinning down the torsion of the Picard group of the quotient stack."}],"review_version":1}