{"id":"222a42ee-5a37-454d-8d99-3d8e4bc4a48b","arxiv_id":"1908.08390","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Assuming the Bloch-Beilinson conjecture, the Chow group valued generating series for special cycles on orthogonal Shimura varieties over totally real fields is modular in all codimensions and for all d_+.","lead":"This paper proves that, assuming the Bloch-Beilinson conjecture, generating series of special algebraic cycles on orthogonal Shimura varieties over totally real fields are modular forms. It covers a case where cycles exist only in codimensions that are multiples of a fixed integer, which was previously open.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The proof of Prop. 7.1/8.2 is written only for integral weight; half-integral weight, needed for m odd, is explicitly left to the reader, leaving a gap in the 'all m' form of Theorem 1.1.","rationale":"The reader's weakest assumption was the Bloch-Beilinson conjecture. That is indeed the external hypothesis, but for a conditional theorem it is not a defect: the paper explicitly assumes it. The more vulnerable point inside the proof is the formal Fourier-series transfer that makes the embedding trick work. The paper itself flags the half-integral case as left to the reader, and Theorem 1.1 is not confined to m even. A reviewer cannot verify the theorem as stated without this case. I also noted that Prop. 8.1 and Lemma 8.3 are only sketched, but those at least have cited references (Knoeller, Freitag-Kiehl). The half-integral extension has no citation. Thus I would keep the CONDITIONAL verdict but require the half-integral details; if they are supplied, the remaining objections are expositional.","tokens_in":30011,"tokens_out":18798,"duration_ms":211756,"concrete_test":"Write out the metaplectic version of Section 8 for a fixed multiplier system: (1) define the appropriate formal Fourier series ring with the shifted exponent lattice and prove it is an integral domain, either directly via Knoeller or by embedding it into the integral-weight FFS through the square of a half-integral form; (2) verify Prop. 8.2 in this setting, with g_z a metaplectic theta series of weight 2*ell and c of half-integral weight k, ensuring the multiplier of f_z = g_z * c is compatible with the metaplectic modular form at every z. If the square-map reduction works, the gap is cosmetic; if not, Theorem 1.1 should be restricted to m even or the half-integral transfer proved separately.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 8 proves the modularity transfer by showing that the ring FFS^bullet_Lambda of Lambda-invariant formal Fourier series is an integral domain (Prop. 8.1) and that every c satisfying the two conditions of Prop. 8.2 is the expansion of a holomorphic modular form. The proof of Prop. 8.1 is itself only a sketch: Lemma 8.3 identifies FFS^bullet_Lambda with the completion R-hat of a local ring at the Baily-Borel boundary point and then invokes Knoeller's Satz 3.1.3. More importantly, the whole argument is formulated for the integral-weight graded ring M_*(Gamma). The sentence 'The case of half-integral weight can be formulated in exactly the same way... We leave this to the reader' is not a proof. Theorem 1.1 covers all m, including m odd, where the weight m/2+1 is half-integral, as in the m=1 quaternionic surface of Section 3. The formal series then live on the metaplectic double cover, have a multiplier system, and the relevant completed ring is not literally the ring M_*(Gamma) used to obtain injectivity of Q(phi). If the metaplectic analogue of Prop. 8.1/8.3 fails, or if Lemma 7.2 cannot be combined with multiplier compatibility at arbitrary z, the embedding trick does not transfer modularity from S-tilde to S in these cases. This is a proof gap independent of the Bloch-Beilinson hypothesis, which is an explicit assumption rather than an internal flaw.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies generating series for special algebraic cycles on orthogonal Shimura varieties S associated to a quadratic space V over a totally real field F of degree d, with signatures ((m,2)^{d_+},(m+2,0)^{d-d_+}). For each n with 1≤n≤m it defines weighted special cycles Z(T,ϕ) of codimension nd_+ and their formal generating series valued in CH^{nd_+}(S). It proves a product formula for these cycles (Proposition 5.2), a pullback formula to Shimura subvarieties (Propositions 6.2 and 6.3), and uses these to implement an embedding trick: a larger Shimura variety \\tilde S containing S is chosen so that, by Vogan-Zuckerman vanishing (Corollary 9.4), all relevant intermediate Jacobians of \\tilde S vanish. Assuming the Bloch-Beilinson conjecture on the Abel-Jacobi map, the Chow-valued generating series on \\tilde S is modular; the pullback identity (7.1) and a formal Fourier series argument (Section 8) then transfer modularity to S. Theorem 1.1 states that, under Bloch-Beilinson, the CH^{nd_+}(S)-valued generating series (1.3) is modular for all n.","tokens_in":30310,"tokens_out":8684,"duration_ms":93284,"significance":"If the proof is completed as written, the paper gives a conditional proof of the conjectured modularity of Chow-valued special cycle generating series in the cases d_+>1, where no evident geometric source of relations among the cycles exists. The paper is explicit and detailed about the main geometric ingredients: the intersection product formula via excess bundles (Theorem 4.15), the pullback formula (Proposition 6.2), and the Vogan-Zuckerman Hodge-number computation (Propositions 9.1 and 9.2, Corollary 9.4) are all worked out. The dependence on the Bloch-Beilinson conjecture is stated honestly and is an explicit hypothesis rather than a hidden assumption. However, the central transfer argument in Section 8 is written only for integral weight, while Theorem 1.1 also covers m odd and hence half-integral weight; the key integral-domain lemma (Lemma 8.3) is asserted without proof. These gaps affect the advertised scope of the main theorem and require attention.","major_comments":[{"comment":"The proof of the transfer statement Proposition 7.1 is carried out only for integral weight. The text says: 'The case of half-integral weight can be formulated in exactly the same way using the metaplectic group. We leave this to the reader.' This is load-bearing, because Theorem 1.1 is stated for all m and the case m odd involves parallel weight m/2+1, which is half-integral and requires the metaplectic cover and multiplier systems; the m=1, d_+=2 example of Section 3 is precisely such a case. The integral-domain property of FFS^•_\\Lambda and the injectivity of Q(φ) used in Proposition 8.2 are established only for the integral-weight graded ring M_*(Γ). Please supply the metaplectic version of Propositions 8.1 and 8.2, or restrict the statement of Theorem 1.1 to m even.","section":"Section 8, paragraph after Eq. (8.2)"},{"comment":"Lemma 8.3 asserts that FFS^•_\\Lambda = \\lim R/I_k, and the identification of FFS^•_\\Lambda with the completed local ring \\hat R depends on it. The proof of this lemma is not given; the sentence 'The following result is the analogue of the Hilfsatz ... and is proved using standard facts about Poincaré series' is not a proof. Since the integral-domain conclusion of Proposition 8.1 is the key input to Proposition 8.2, this step needs either a complete proof or a precise reference that covers the Hilbert-Siegel setting with the Λ-invariance and the filtration I_k.","section":"Section 8, proof of Proposition 8.1"}],"minor_comments":[{"comment":"The text refers to 'Proposition 2.2 asserts the modularity of φ_1(τ,S)', but there is no Proposition 2.2; the intended reference is Theorem 2.2.","section":"Section 3, paragraph after Problem 1"},{"comment":"Reference [13] spells 'Monatshefte' as 'Montashefte', and reference [10] spells 'Kiehl' as 'Keihl'; please correct these.","section":"References"},{"comment":"It is not explicitly checked that, for a fixed K-invariant φ, the scalar series c = λ(φ_n(τ,ϕ)) obtained from the generating series (5.9) has support in S^•_F ∪ {0} for some level ν, as required for c to lie in FFS^•_\\Lambda. This is likely true because the possible T are constrained by the dual lattice, but the support and level compatibility should be stated in the application of Proposition 8.2.","section":"Section 8, application to (7.1)"},{"comment":"The set S^•_F depends on ν but the notation suppresses this dependence; a brief remark would avoid confusion in the proof of Proposition 8.1.","section":"Section 8, definition of S^•_F"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Stephen,\n\nMy read on Kudla's note (arXiv:1908.08390). The headline: there is a genuinely new conditional theorem here — assume Bloch–Beilinson, and the CH^{nd_+}-valued generating series of special cycles is modular for every n, including d_+ > 1, which had been stuck. The architecture is the embedding trick plus Vogan–Zuckerman: enlarge V by a 4ℓ-dimensional totally positive U_0, kill H^{2nd_+−1}(~S), invoke BB for Chow-valued modularity upstairs, then pull back through identity (7.1). The transfer step (Prop. 7.1) is a formal-Fourier-series argument credited to Bruinier, using Knöller's theorem that the ring of symmetric formal series is an integral domain.\n\nWhat is genuinely good and complete: the intersection theory. Theorem 4.15 gives the Chow product formula for special cycles in general d_+, with the excess bundle computed as C_Γ ⊗ (U_1 ∩ U_2) and a careful cover argument via the Jaffee lemma. The weighted-cycle product formula (Prop. 5.2) and the pullback formula (Prop. 6.2) are new in this generality and are essentially self-contained. Section 9 works out the Vogan–Zuckerman Hodge diamond in detail, which is useful on its own. The paper is honest: the main theorem is explicitly conditional, and the sketchy points are labeled as sketches.\n\nSoft spots, in proportion. The largest is the half-integral weight gap, and the stress-test note is right about it. Section 8 works in the integral-weight graded ring M_*(Γ). The sentence that the half-integral case 'can be formulated in exactly the same way... we leave this to the reader' is not a proof. This is not a corner case: for m odd, the weight m/2+1 is half-integral, including the m=1 quaternionic surface that motivates the paper. The transfer needs the metaplectic analogue of Props. 8.1–8.2 with multiplier systems, and Knöller's Satz 3.1.3 is quoted for the integral-weight Hilbert–Siegel setting. I think the gap is fillable, but as it stands Theorem 1.1 is not fully proved for m odd. Second, Prop. 8.1 and Lemma 8.3 are sketches: the identification of FFS^•_Λ with the completion of the local ring at the Baily–Borel cusp, and the equivalence of the λ-filter with the m-adic filter, rest on Knöller's paper; a referee should verify that the Λ-symmetry condition matches his hypotheses. Third, minor: Props. 4.8 and Lemma 4.14 are quoted without proof, but those are standard Fulton facts; the new computation, the excess bundle, is done in full. Bloch–Beilinson is an explicit assumption, not a hidden flaw; there is no circularity and no fitted parameter.\n\nWho is this for: anyone working on special cycles, Shimura varieties, and Kudla-style generating series. It deserves a serious referee and publication after revision; I would ask for the half-integral case written out or a precise reference, and for the Knöller details. My verdict is conditional accept.\n\nCheers.","headline":"Kudla proves Chow-valued modularity of special-cycle generating series for all n and general d_+ conditionally on Bloch–Beilinson; the new content is real, but the half-integral-weight case needed for m odd is left to the reader, so Theorem 1.1 has a gap as stated.","tokens_in":30873,"tokens_out":11429,"would_cite":true,"duration_ms":104573,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14C25","11F27","11F46","14G35"],"pacs":[],"model":"deepseek-v4-flash","headline":"Assuming the Bloch–Beilinson conjecture, the Chow group-valued generating series for special cycles is a Hilbert–Siegel modular form.","keywords":["orthogonal Shimura varieties","special cycles","generating series","Chow groups","Hilbert-Siegel modular forms","theta correspondence","Abel-Jacobi map","Bloch-Beilinson conjecture"],"falsifier":"Since the theorem is conditional, a decisive test would be to find a smooth projective variety, ideally one of the orthogonal Shimura varieties treated here, carrying a cohomologically trivial special cycle whose Abel–Jacobi invariant in the intermediate Jacobian is nonzero; such a cycle would falsify the Bloch–Beilinson hypothesis under which the theorem is proved. A less radical test is computational: in the quaternionic real-quadratic example with $d_+=2$ and $m=1$, compute finitely many Fourier coefficients of $\\varphi_n(\\tau,\\phi,S,\\lambda)$ for a linear functional $\\lambda$ that vanishes on the image of the cycle class map and check whether they satisfy the Fourier-coefficient recurrences forced by Hilbert modularity; a violation would contradict the theorem's conclusion under its stated assumptions.","tokens_in":29734,"feed_emoji":"🧮","tokens_out":11287,"duration_ms":109792,"temperature":0.7,"pith_summary":"The paper proves that the formal generating series packaging all special algebraic cycles of codimension $nd_+$ on an orthogonal Shimura variety over a totally real field is a Hilbert–Siegel modular form of parallel weight $m/2+1$, for every $1\\le n\\le m$, conditional on the Bloch–Beilinson conjecture. The new content is that this holds for every $d_+$, including the previously intractable cases $d_+>1$ where no evident geometric source of relations among cycles exists. The proof shows that the conjecture supplies the missing relations indirectly: after enlarging the variety by a totally positive space of dimension $4\\ell$, low odd-degree cohomology vanishes, the Abel–Jacobi intermediate Jacobian vanishes, and the cycle class map becomes injective on the enlarged variety. The paper also establishes an intersection product formula and a pullback formula for special cycles that are needed to transport modularity back down to the original variety.","feed_headline":"Special-cycle generating series are modular, assuming a key conjecture","feed_subtitle":"Extends the result to every codimension on orthogonal Shimura varieties, conditional on a major conjecture.","key_machinery":"The load-bearing object is the weighted special-cycle generating series $\\varphi_n(\\tau,\\phi)=\\sum_{T\\in \\mathrm{Sym}_n(F)_{\\ge 0}} [Z(T,\\phi)]\\,q^T$, whose coefficients live in the Chow group $CH^{nd_+}(S)$ and whose constant term is a power of the class $c_S=c_{d_+}(\\mathcal{C}_S)$, the top Chern class of the co-tautological bundle. The mechanism that carries the proof is the embedding trick: forming $\\widetilde{V}=U_0\\oplus V$ with $\\dim U_0=4\\ell>nd_+$ produces a larger Shimura variety with vanishing low odd-degree Betti cohomology, so that, assuming Bloch–Beilinson, its Chow-valued series is modular. The transfer back to $S$ is delivered by two identities: the product formula for intersections of special cycles, proved with standard intersection-theoretic tools (normal cones, Segre classes, excess bundles), and the pullback formula $\\rho^*\\varphi_n(\\tau;\\phi_0\\otimes\\phi)=\\theta(\\tau,\\phi_0)\\varphi_n(\\tau,\\phi)$. A lemma on formal Fourier series — that the ring of symmetric formal Fourier series is an integral domain, so quotients by nonvanishing $\\theta$ series are legitimate — completes the descent.","core_discovery":"On its own terms, the central claim is Theorem 1.1: assume the Bloch–Beilinson conjecture. Then, for a quadratic space $V$ over a totally real field of degree $d$ with signature $((m,2)^{d_+},(m+2,0)^{d-d_+})$ and $1\\le d_+<d$, the formal series $\\varphi_n(\\tau,\\phi,S)=\\sum_T [Z(T,\\phi)]\\,q^T$ with coefficients in $CH^{nd_+}(S)$ is a Hilbert–Siegel modular form for all $n$, $1\\le n\\le m$. The image of this series under the cycle class map was already known to be modular by $\\theta$-correspondence results; the difficulty is that the cycle class map can have a kernel. The paper's strategy is to embed $S$ in a larger Shimura variety $\\widetilde{S}$ obtained by adding a totally positive definite space of dimension $4\\ell>nd_+$, where a representation-theoretic vanishing theorem forces $H^{2nd_+-1}(\\widetilde{S})=0$ and hence $J_{nd_+}(\\widetilde{S})=0$. Under the Bloch–Beilinson conjecture, Abel–Jacobi is injective up to torsion, so the cycle class map is injective on $\\widetilde{S}$ and the Chow-valued series there is modular; a pullback formula expresses the pulled-back series as a product of a $\\theta$ series and the original series, and a result on formal Fourier series shows the $\\theta$ factor can be cancelled.","pith_inferences":["Because the proof requires only vanishing of $H^{2nd_+-1}$ plus Abel–Jacobi injectivity, the same recipe could prove modularity for other families of Shimura varieties whenever a suitable Hodge-diamond vanishing is found; the paper notes no such argument is currently available for unitary Shimura varieties.","A computational check is conceivable in the quaternionic real-quadratic example: modularity predicts specific linear recurrences among the weighted degrees of the special $0$-cycles, and these recurrences could be tested numerically even though the underlying relations are invisible geometrically.","The product formula suggests a constructive route to the missing relations: intersecting special cycles with powers of the class $c_S$ and comparing with known modular Fourier coefficients may generate explicit candidate relations among special cycles.","If the Bloch–Beilinson conjecture were later found to fail in this range, the modularity statement would remain plausible but would require a different, non-abelian source of relations among special cycles."],"forward_implications":["For every complex-valued linear functional $\\lambda$ on $CH^{nd_+}(S)$, the series $\\varphi_n(\\tau,\\phi,S,\\lambda)$ is absolutely convergent and is a Hilbert–Siegel modular form of parallel weight $m/2+1$.","When $d_+>1$, special cycles occupy only codimensions that are multiples of $d_+$, with no lower-codimension cycles to generate relations; the theorem shows the Bloch–Beilinson conjecture supplies those relations indirectly.","The product formula and pullback formula give the algebraic identities among special cycles that are needed to transfer modularity between different Shimura varieties; they generalize the divisor-case identities used for $d_+=1$.","The weighted adèlic formulation makes the statement equivariant under finite-adèlic changes of level, so modularity holds on the direct limit $CH^{nd_+}(S)$, not just on individual level structures."],"supporting_citations":[{"why":"Produces harmonic theta forms whose pushforwards give the cohomology classes of special cycles, establishing cohomological modularity.","marker":"[16]"},{"why":"Completes the theta-correspondence construction used for the cohomology-valued generating series.","marker":"[17]"},{"why":"Identifies Fourier coefficients of holomorphic modular forms with intersection numbers of special cycles, yielding the known modularity of the cycle-class image.","marker":"[18]"},{"why":"Proves the $d_+=1$ case of Chow-valued modularity and introduces the embedding trick that this paper adapts.","marker":"[24]"},{"why":"Shows, under a convergence assumption, that higher-codimension Chow-valued generating series are Siegel modular forms, providing the inductive template.","marker":"[25]"},{"why":"Supplies the convergence argument for Fourier–Jacobi series that completes the modularity proof in the $F=\\mathbb{Q}$ case.","marker":"[7]"},{"why":"Provides the intersection-theoretic machinery (normal cones, Segre classes, excess bundles) used to prove the product formula.","marker":"[11]"},{"why":"Proves that the ring of formal Fourier series at the boundary is an integral domain, used to cancel theta factors in the descent argument.","marker":"[13]"},{"why":"Gives the vanishing theorem for unitary representations with nonzero cohomology that yields the low odd-degree Betti-number vanishing.","marker":"[23]"},{"why":"Sets up special cycles, weighted cycles, and generating series in the $d_+=1$ case; its definitions and properties are extended here.","marker":"[14]"}],"fun_headline_variants":["Chow-valued special cycle series modular, assuming BB","Assuming Bloch-Beilinson, cycle series become modular","All codimension cycle series modular on Shimura varieties","Modularity of Chow series for cycles, under a conjecture","Conditional proof: cycle series are Hilbert-Siegel modular"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the Bloch–Beilinson conjecture, the unproved statement that the Abel–Jacobi map from cohomologically trivial cycles to the intermediate Jacobian is injective up to torsion; if this fails, the proof cannot upgrade cohomological modularity to Chow-valued modularity on the enlarged variety.","fun_headline_variants_meta":{"raw":{"variants":["Chow-valued special cycle series modular, assuming BB","Assuming Bloch-Beilinson, cycle series become modular","All codimension cycle series modular on Shimura varieties","Modularity of Chow series for cycles, under a conjecture","Conditional proof: cycle series are Hilbert-Siegel modular"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000202,"raw_usage":{"total_tokens":1454,"prompt_tokens":1091,"completion_tokens":363,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":707,"completion_tokens_details":{"reasoning_tokens":284}},"tokens_in":707,"tokens_out":363,"duration_ms":4231,"temperature":1.0,"reasoning_tokens":284,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:40:18.351058+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Since the theorem is conditional, a decisive test would be to find a smooth projective variety, ideally one of the orthogonal Shimura varieties treated here, carrying a cohomologically trivial special cycle whose Abel–Jacobi invariant in the intermediate Jacobian is nonzero; such a cycle would falsify the Bloch–Beilinson hypothesis under which the theorem is proved. A less radical test is computational: in the quaternionic real-quadratic example with $d_+=2$ and $m=1$, compute finitely many Fourier coefficients of $\\varphi_n(\\tau,\\phi,S,\\lambda)$ for a linear functional $\\lambda$ that vanishes on the image of the cycle class map and check whether they satisfy the Fourier-coefficient recurrences forced by Hilbert modularity; a violation would contradict the theorem's conclusion under its stated assumptions.","supporting_citations":[{"cited_title":"Kudla and J","cited_arxiv_id":null,"evidence_quote":"Produces harmonic theta forms whose pushforwards give the cohomology classes of special cycles, establishing cohomological modularity."},{"cited_title":"Annalen, 277 (1987), 267–314","cited_arxiv_id":null,"evidence_quote":"Completes the theta-correspondence construction used for the cohomology-valued generating series."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Identifies Fourier coefficients of holomorphic modular forms with intersection numbers of special cycles, yielding the known modularity of the cycle-class image."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Proves the $d_+=1$ case of Chow-valued modularity and introduces the embedding trick that this paper adapts."},{"cited_title":"thesis, Columbia University (2009)","cited_arxiv_id":null,"evidence_quote":"Shows, under a convergence assumption, that higher-codimension Chow-valued generating series are Siegel modular forms, providing the inductive template."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the convergence argument for Fourier–Jacobi series that completes the modularity proof in the $F=\\mathbb{Q}$ case."},{"cited_title":"Fulton, Intersection Theory, Ergebnisse der Mathem atik und ihrer Grenzgebiete 2, Springer-Verlag, Berlin, 1984","cited_arxiv_id":null,"evidence_quote":"Provides the intersection-theoretic machinery (normal cones, Segre classes, excess bundles) used to prove the product formula."},{"cited_title":"unendlich-ferner","cited_arxiv_id":null,"evidence_quote":"Proves that the ring of formal Fourier series at the boundary is an integral domain, used to cancel theta factors in the descent argument."},{"cited_title":"Vogan and G","cited_arxiv_id":null,"evidence_quote":"Gives the vanishing theorem for unitary representations with nonzero cohomology that yields the low odd-degree Betti-number vanishing."},{"cited_title":"Kudla, Algebraic cycles on Shimura varieties of orthogonal type , Duke Math","cited_arxiv_id":null,"evidence_quote":"Sets up special cycles, weighted cycles, and generating series in the $d_+=1$ case; its definitions and properties are extended here."}],"review_version":1}