{"id":"0c789fbc-041e-403d-bde6-b47e41f6fc78","arxiv_id":"1908.08063","paper_version":6,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Assuming the Beilinson-Bloch conjecture, generating series of special cycles of codimension er on these Shimura varieties are Hilbert-Siegel modular forms of genus r and weight 1+n/2.","lead":"Under the Beilinson-Bloch conjecture on algebraic cycles, special cycles on mixed-signature orthogonal Shimura varieties over totally real fields generate Hilbert-Siegel modular forms. The paper extends a one-place result to several split places, but for higher genus it also assumes an unproved absolute convergence.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Section 4.2 induction applies the genus-one theorem to intermediate Shimura varieties M_{Kf,y} whose hypotheses are not among the assumptions of Theorem 1.5; the r>=2 claim is therefore not established as written.","rationale":"After reading the proof, the central risk is not the explicitly unproved absolute convergence (which the paper states in Remark 1.9) but the induction step, because the proof uses the genus-one theorem for a family of smaller Shimura varieties without verifying their hypotheses. The reader's weakest_assumption already flags the deferred use of Theorem 1.5(2) in the induction; I agree and make it more specific: the dimension of the orthogonal complement can drop below 3, and for such y the needed variant of the r=1 theorem is neither Theorem 1.5(2) nor covered by the paper's assumptions. This is an internal proof gap, not a disagreement with the consensus. It is potentially repairable by adding an explicit hypothesis, e.g. Conjecture 1.3 for the Shimura varieties attached to all admissible orthogonal complements (or for a fixed sufficiently large ambient variety containing all of them), or by supplying the missing argument from [22]. Because this is a defect in the theorem as stated rather than a falsification of the expected result, the reader's conditional verdict stands; the conditions should be strengthened. I do not see an independent reason to doubt the r=1 cases or the unconditional cohomological input. The absolute convergence issue is real but explicitly assumed, so it is secondary.","tokens_in":12742,"tokens_out":17710,"duration_ms":173941,"concrete_test":"Instantiate Section 4.2 at r=2, n=3, e>1. Fix a nonzero admissible y in V. The inner sum over x1 in K_{f,y}\\z y^K is the genus-one generating series on M_{Kf,y}, whose quadratic space y^\\perp has dimension 4, i.e. n'=2. Check which theorem can justify the w1-identity displayed after 'by Theorem 1.5(2) and Theorem 1.6(2)': Theorem 1.5(2) fails because n'<3; Theorem 1.6(2) would require Conjecture 1.3 for an enlarged M'_{K'_f,y} built from y^\\perp, which is not an assumption of Theorem 1.5. If [22] contains a different argument for this identity that uses only the original assumptions, exhibit it; otherwise Theorem 1.5(1) has a genuine gap for r>=2.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing gap is the induction step in Section 4.2. After writing Z_{phi_f}(g1) as a sum over admissible y in V^{r-1} and x1 in K_{f,y}\\z y^K, the proof asserts 'by Theorem 1.5(2) and Theorem 1.6(2)' the identity under w1. For fixed y, this identity is the r=1 modularity statement for the Shimura variety M_{Kf,y} attached to the orthogonal complement y^\\perp, not for the originally assumed MKf. If m = dim U(y), then y^\\perp has dimension n+2-m, so the r=1 theorem for M_{Kf,y} requires either n-m >= 3 (the hypothesis of Theorem 1.5(2)) or, when n-m <= 2, Conjecture 1.3 for a further enlarged variety M'_{K'_f,y} as in Theorem 1.6(2). Neither condition is part of the assumptions of Theorem 1.5: only Conjecture 1.3 for the original MKf (or for the fixed M'_K'_f in Theorem 1.6) is assumed. The deferred reference [22] is the e=1 case, where the Abel-Jacobi map in degree 1 is automatically injective, so it cannot supply the missing hypothesis for e>1. Thus the induction is incomplete unless hypotheses are added.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies Kudla's generating series of special cycles in the Chow groups of orthogonal Shimura varieties over a totally real field F, in the situation where the quadratic space has signature (n,2) at e real places and is definite at the remaining d−e places, with 1 ≤ e < d. The main theorems state that, assuming the Beilinson–Bloch injectivity of the higher Abel–Jacobi map (Conjecture 1.3) at m = e, every C-linear functional of the Chow-valued generating series is a Hilbert–Siegel modular form of genus r and weight 1 + n/2, unconditionally for r = 1 and for r ≥ 2 under an additional absolute-convergence hypothesis. The proof for r = 1 reduces to the cohomological modularity theorems of Kudla and Rosu–Yott via injectivity of the cycle map; for r ≥ 2 the paper uses induction on r, following Yuan–Zhang–Zhang, and reduces w1-invariance to the genus-one case for certain smaller Shimura varieties.","tokens_in":13033,"tokens_out":7930,"duration_ms":69593,"significance":"If the main theorems were established as stated, they would give a conditional generalization of Yuan–Zhang–Zhang's theorem from e = 1 to arbitrary e < d, and would show that Kudla's modularity conjecture for higher Chow groups follows from a standard conjecture on algebraic cycles. The r = 1 reduction is elegant: the vanishing of H^{2e−1} for n ≥ 3 combines Matsushima's formula with a Vogan–Zuckerman/Kumaresan vanishing argument, and the use of Conjecture 1.3 to extend linear functionals from Chow groups to cohomology is a clean idea. The paper is also transparent about its limitations, explicitly stating in Remark 1.9 that absolute convergence is not known for r ≥ 2. However, the induction step for r ≥ 2 contains a serious gap that affects the central claim of the paper.","major_comments":[{"comment":"The step 'by Theorem 1.5 (2) and Theorem 1.6 (2)' is applied to the Shimura variety M_{K_{f,y}} attached to the orthogonal complement y^⊥, not to the original M_{K_f}. For a fixed admissible y with dim U(y) = m, the quadratic space y^⊥ has dimension n+2−m, so the genus-one theorem for M_{K_{f,y}} requires either n−m ≥ 3 together with Conjecture 1.3 for M_{K_{f,y}} at m = e (Theorem 1.5(2)) or, when n−m ≤ 2, Conjecture 1.3 for a further enlarged variety M'_{K'_{f,y}} attached to y^⊥ ⊕ W (Theorem 1.6(2)). Neither condition is among the assumptions of Theorems 1.5 and 1.6, which only assume Conjecture 1.3 for the original M_{K_f} (or for the fixed M'_{K'_f} in Theorem 1.6). The reference to Yuan–Zhang–Zhang [22] covers e = 1, where m = 1 and the Abel–Jacobi map is automatically injective, so it cannot supply the missing hypothesis for e > 1. Consequently the Poisson-summation argument establishes w1-invariance only under additional assumptions that are not stated, and Theorems 1.5(1) and 1.6(1) for r ≥ 2 are not proven as written.","section":"§4.2, invariance under w1"},{"comment":"The abstract omits the absolute-convergence hypothesis that the theorems require. The abstract claims that 'assuming the Beilinson–Bloch conjecture' the generating series is a Hilbert–Siegel modular form, but the theorems for r ≥ 2 have the additional condition that ℓ(Z_{φ_f})(τ) is absolutely convergent, which Remark 1.9 explicitly says is not known. This mismatch between the advertised result and the actual statement should be corrected, either by adding the hypothesis in the abstract or by stating the result as conditional on absolute convergence.","section":"Abstract and Theorems 1.5(1), 1.6(1)"}],"minor_comments":[{"comment":"In the paragraph after the m(a) calculation, the sentence 'On the other hand. we have Upxq “ Upxaq, so Zφf pxq “ Zφf pxaq' appears to contain a notational slip: it should express equality of the cycles Z_{x a} and Z_x, not equality of the functions Z_{φ_f}(x) and Z_{φ_f}(xa).","section":"§4.1"},{"comment":"The deduction that Z_{φ_f}(g1) is absolutely convergent from the identity i^*(Z_{φ_f⊗φ'_f}(g1)) = Z_{φ_f}(g1) θ_{φ'_f}(g1) is not fully justified in the text; the manuscript should cite or provide the argument showing that the theta factor does not obstruct the conclusion.","section":"§3.2"},{"comment":"The proof of Proposition 2.4 refers to 'the same way as [22, Proposition 3.1]' but the non-trivial intersection-theoretic input from Proposition 2.3 is only sketched; a more detailed reference to the specific steps in [22] would improve readability.","section":"§2.2, Proposition 2.4"},{"comment":"There are several typos and formatting issues, including 'Bruhat-Schwartz' for 'Bruhat–Schwartz' (or 'Schwartz–Bruhat'), missing parentheses in 'SO pV qpALq' in Lemma 3.2, and an undeclared tensor-product symbol in 'π – b vπv'.","section":"Throughout"},{"comment":"The proof that all archimedean components π_i are trivial for i = 1,...,e relies on 'See [21, Section 4.3.2, Example 4]' together with Lemma 3.2; this is plausible but the deduction that π_i is trivial from the vanishing of H^0 and H^1 cohomology for non-trivial representations would benefit from one additional sentence explaining the role of the trivial representation in the Künneth decomposition.","section":"§3.1"}],"recommendation":"major_revision","confidential_remarks":"The author notes that Kudla independently obtained similar results in [17] and that the proof strategies differ; the editor may wish to consider the novelty and overlap. The main technical issue is the induction gap in §4.2: as written, the r ≥ 2 theorems are not established because the w1-invariance step requires unstated hypotheses for intermediate Shimura varieties. If the author can fix this by adding the missing assumptions or by proving that Conjecture 1.3 is inherited by the intermediate varieties, the paper could be acceptable; as it stands, the central claim for r ≥ 2 is not supported by the proof."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The r=1 half is clean and genuinely worth having; the r≥2 half has a real gap in the induction that the author's summary glosses over.\n\nThe r=1 part is a genuine, clean contribution: assuming Conjecture 1.3 for the original Shimura variety, it deduces Chow-valued modularity from the cohomological modularity of Kudla and Rosu-Yott via the injectivity of the cycle map, with the vanishing of H^{2e-1} proved by Matsushima plus a vanishing theorem. The n≤2 case handled by embedding into a larger variety is also plausible. The author deserves credit for being honest that absolute convergence is unknown for r>1 (Remark 1.9) and for acknowledging Kudla's independent, different proof under a stronger BB hypothesis.\n\nThe problem is the r≥2 induction. In Section 4.2, the proof writes the genus-r series as a sum over y ∈ V^{r-1} and then applies the r=1 theorem to the intermediate Shimura variety M_{K_{f,y}} attached to y^⊥. That theorem requires either dim(y^⊥) - 2 ≥ 3 or, if not, Conjecture 1.3 for a further enlarged variety attached to y^⊥. Neither hypothesis is among the assumptions of Theorem 1.5 or 1.6: the BB conjecture is assumed only for the original M_{K_f} (or the fixed M'_{K'_f} in the n≤2 case). The deferred reference [22] does not help because in [22] the codimension is 1, where Abel-Jacobi is automatically injective. So for r≥2 the main theorem is not established as written. This is not a minor omission; it is load-bearing for half the paper. It may be repairable by strengthening the BB assumption to cover all sub-Shimura varieties arising in the induction, but as it stands the theorem needs that extra hypothesis.\n\nTwo smaller points: the abstract says modularity holds under the BB conjecture without mentioning the absolute convergence assumption, which appears only in the theorems and Remark 1.9. And the proof of the intersection formula Proposition 2.3 is terse, though that looks like a routine detail.\n\nOverall, this is for specialists in arithmetic geometry working on Kudla's program. The r=1 part deserves to see the light; the r≥2 claim needs a fix or at least a clear statement of the stronger hypothesis. I would send it to a referee; a good referee will catch the induction gap. I wouldn't cite the r≥2 result in its current form, but I'd keep the paper around for the r=1 argument and the survey of the mixed-signature setup.","headline":"The r=1 half is a clean conditional contribution; the r≥2 induction has a real, fixable gap that the current hypotheses do not cover.","tokens_in":13595,"tokens_out":7151,"would_cite":false,"duration_ms":61410,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11G18","11F46","14C17"],"pacs":[],"model":"deepseek-v4-flash","headline":"Under the higher Abel-Jacobi injectivity conjecture, the generating series of special cycles is a Hilbert-Siegel modular form of weight $1+n/2$.","keywords":["Shimura varieties","special cycles","Hilbert-Siegel modular forms","higher Chow groups","higher Abel-Jacobi map","totally real fields","modularity of generating series","orthogonal groups"],"falsifier":"For an explicit totally real field and quadratic space with $n\\ge 3$ and $r=2$, take a linear functional $\\ell$ on $CH^{2e}$ not induced by cohomology, compute the first Fourier coefficients of $\\ell(Z_{\\varphi_f})(\\tau)$, and check whether the series obeys the Hilbert-Siegel transformation under $\\tau\\mapsto-\\tau^{-1}$ after analytic continuation. A single violation of that functional equation would falsify the theorem's assertion, while divergence of the series in that example would falsify its absolute-convergence hypothesis.","tokens_in":12499,"feed_emoji":"📐","tokens_out":12053,"duration_ms":113128,"temperature":0.7,"pith_summary":"On Shimura varieties attached to quadratic spaces that are indefinite of signature $(n,2)$ at $e$ real places and definite at the remaining places, one can assemble special cycles of codimension $er$ into a formal power series. This paper proves that, assuming the higher Abel-Jacobi injectivity conjecture, every linear evaluation of this Chow-valued series is a Hilbert-Siegel modular form of genus $r$ and weight $1+n/2$. For genus one the absolute convergence is proved, so the modularity is unconditional there; for higher genus the paper must assume absolute convergence. The interest is that modularity of cycle generating series is the geometric input to identities that relate cycle intersections to Fourier coefficients of modular forms and to the arithmetic of special $L$-values.","feed_headline":"Generating series of special cycles are Hilbert-Siegel modular","feed_subtitle":"The result, conditional on the higher Abel-Jacobi injectivity conjecture, extends modularity over totally real fields to mixed signatures.","key_machinery":"The load-bearing object is the Chow-valued generating series $Z_{\\varphi_f}(\\tau)=\\sum_x \\varphi_f(g^{-1}x)Z(x,g)q^{T(x)}$, with coefficients special cycles in $CH^{er}(M_{K_f})_{\\mathbb{C}}$. The argument runs through four linked mechanisms: a vanishing theorem for the odd cohomology group, obtained from the standard representation-theoretic cohomology formula and relative Lie-algebra cohomology, which under the Abel-Jacobi injectivity conjecture makes the cycle map injective; the pull-back formula $i_W^*Z_{\\varphi_f}=Z_{\\varphi_{1,f}}\\theta_{\\varphi_{2,f}}$, which passes modularity from a larger variety down to a smaller one; degenerate Whittaker functions and the oscillator representation, which convert the requirement of being a Hilbert-Siegel modular form into invariance under the Siegel parabolic subgroup and the Weyl element $w_1$; and induction on $r$, which uses the summation identity to reduce the $w_1$-invariance at genus $r$ to the already-proved genus-one case.","core_discovery":"The paper's central claim is a conditional transfer: if the Abel-Jacobi injectivity conjecture holds for the Shimura variety at $m=e$, so that the cycle map from $CH^e$ to $H^{2e}$ is injective, then modularity of special cycles, already known in cohomology, descends to the Chow group. The proof first shows $H^{2e-1}(M_{K_f},\\mathbb{C})=0$ for $n\\ge 3$ by the standard cohomological formula for locally symmetric spaces together with relative Lie-algebra cohomology vanishing; with the conjecture, the cycle map $CH^e_{\\mathbb{C}}\\to H^{2e}_{\\mathbb{C}}$ is injective. Therefore any $\\mathbb{C}$-linear functional on $CH^{er}$ for $r=1$ factors through cohomology, where unconditional modularity results apply. For $r\\ge 2$, induction on $r$, the pull-back formula, and the standard summation identity extend the genus-one statement, with absolute convergence of the evaluated series as an extra hypothesis. For $n\\le 2$ the same conclusion is obtained by embedding into a larger Shimura variety and assuming the conjecture there.","pith_inferences":["If absolute convergence for $r\\ge 2$ turns out to be automatic in these settings, for instance because the relevant Chow groups are finite-dimensional or the cycle growth is slow enough, then the paper's induction would prove modularity without any extra convergence hypothesis; checking this on an explicit two-dimensional family would be a direct test.","The modularity statement is a natural ingredient for height-pairing identities: pairing the generating series with a fixed cycle should produce Fourier coefficients of an Eisenstein series or a derivative of an $L$-function, an arithmetic application the paper does not pursue.","The method is likely to transplant to other groups, such as unitary or symplectic Shimura varieties, wherever the same odd-cohomology vanishing and an Abel-Jacobi injectivity statement are available, though the paper does not discuss such generalizations.","A failure of the Abel-Jacobi injectivity conjecture would not necessarily destroy modularity: the conjecture is a sufficient route through cohomology, not a necessary condition, so the modularity phenomenon could survive even if cycle-map injectivity fails."],"forward_implications":["For $n\\ge 3$ and $r=1$, the modularity theorem is unconditional once the Abel-Jacobi injectivity conjecture at $m=e$ is granted, because the paper proves the needed absolute convergence in this case.","For $n\\le 2$, the modularity statement for the smaller Shimura variety follows from the same conjecture imposed on a larger ambient Shimura variety, so small-dimensional orthogonal cases are covered.","When $e=1$, the required injectivity is known, so the result recovers the classical modularity conjecture for special cycles over totally real fields as a special case.","For every $r\\ge 2$, the higher-genus statement is reduced to the genus-one statement; the full Hilbert-Siegel modularity is therefore governed by cycle-map injectivity at codimension $e$ together with absolute convergence.","If the Abel-Jacobi injectivity conjecture is eventually proved in the relevant cases, the same argument would produce unconditional Hilbert-Siegel modular forms for all $r$."],"supporting_citations":[{"why":"Supplies the intersection formula, pull-back formula, and induction-by-$r$ strategy that the paper adapts to mixed signatures.","marker":"[22]"},{"why":"Provides the unconditional modularity of cohomology-valued generating series that the main theorem reduces to after cycle-map injectivity.","marker":"[17]"},{"why":"Constructs the special cycles in the mixed-signature setting and proves their cohomological modularity.","marker":"[19]"},{"why":"Formulates the higher Abel-Jacobi injectivity conjecture that the main theorem assumes.","marker":"[1]"},{"why":"Gives the vanishing theorem for relative Lie-algebra cohomology used to prove the odd cohomology group vanishes.","marker":"[20]"},{"why":"Provides the vanishing criterion for irreducible unitary modules with nonzero relative cohomology.","marker":"[18]"},{"why":"Shows that a trivial archimedean component forces all local components to be characters, completing the cohomology vanishing.","marker":"[7]"},{"why":"Establishes the original modularity of intersection numbers on locally symmetric spaces that underpins the generating series.","marker":"[12]"}],"fun_headline_variants":["Modularity of special cycles in Chow under Beilinson-Bloch","Conditional modularity: special cycles in Chow are Hilbert-Siegel forms","Hilbert-Siegel modularity of special cycles in Chow, assuming ABJ","Special cycles in Chow: modular generating series assuming Beilinson-Bloch"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the higher Abel-Jacobi injectivity conjecture in the relevant codimension, and, for $r\\ge 2$, the additional assumption that the evaluated generating series converges absolutely.","fun_headline_variants_meta":{"raw":{"variants":["Modularity of special cycles in Chow under Beilinson-Bloch","Conditional modularity: special cycles in Chow are Hilbert-Siegel forms","Hilbert-Siegel modularity of special cycles in Chow, assuming ABJ","Special cycles in Chow: modular generating series assuming Beilinson-Bloch"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001435,"raw_usage":{"total_tokens":5805,"prompt_tokens":985,"completion_tokens":4820,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":601,"completion_tokens_details":{"reasoning_tokens":4737}},"tokens_in":601,"tokens_out":4820,"duration_ms":31879,"temperature":1.0,"reasoning_tokens":4737,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:52:08.345215+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For an explicit totally real field and quadratic space with $n\\ge 3$ and $r=2$, take a linear functional $\\ell$ on $CH^{2e}$ not induced by cohomology, compute the first Fourier coefficients of $\\ell(Z_{\\varphi_f})(\\tau)$, and check whether the series obeys the Hilbert-Siegel transformation under $\\tau\\mapsto-\\tau^{-1}$ after analytic continuation. A single violation of that functional equation would falsify the theorem's assertion, while divergence of the series in that example would falsify its absolute-convergence hypothesis.","supporting_citations":[{"cited_title":"145 (2009), no","cited_arxiv_id":null,"evidence_quote":"Supplies the intersection formula, pull-back formula, and induction-by-$r$ strategy that the paper adapts to mixed signatures."},{"cited_title":"Remarks on generating series for special cycles","cited_arxiv_id":"1908.08390","evidence_quote":"Provides the unconditional modularity of cohomology-valued generating series that the main theorem reduces to after cycle-map injectivity."},{"cited_title":"Generating series of a new class of orthogonal Shimura varieties","cited_arxiv_id":"1812.05183","evidence_quote":"Constructs the special cycles in the mixed-signature setting and proves their cohomological modularity."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Formulates the higher Abel-Jacobi injectivity conjecture that the main theorem assumes."},{"cited_title":"53 (1984), no","cited_arxiv_id":null,"evidence_quote":"Gives the vanishing theorem for relative Lie-algebra cohomology used to prove the odd cohomology group vanishes."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the vanishing criterion for irreducible unitary modules with nonzero relative cohomology."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Shows that a trivial archimedean component forces all local components to be characters, completing the cohomology vanishing."},{"cited_title":"Hautes ´Etudes Sci","cited_arxiv_id":null,"evidence_quote":"Establishes the original modularity of intersection numbers on locally symmetric spaces that underpins the generating series."}],"review_version":1}