{"id":"01a1e8e9-00f9-4b5f-bfba-0f8a7d91eb10","arxiv_id":"2507.13299","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The authors construct explicit boundary corrections making the Kudla-Millson generating series of special cycles on compactified unitary Shimura varieties a holomorphic Hermitian modular form for codimensions g ≤ n/2.","lead":"This paper proves the cohomological version of the Kudla conjecture for unitary Shimura varieties of signature (n+1,1): it constructs boundary corrections to the generating series of special cycles so the corrected series is a holomorphic Hermitian modular form, for codimensions up to the middle. It also introduces Hermitian quasi-modular forms and shows the generating series of Zariski closures of special cycles is one such form.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central modularity claim is conditional on Theorem 2.12, which is stated without proof and deferred to Howard; a normalization or weight error there would propagate through every completion and correction in Sections 4–5.","rationale":"I read the paper as proving the cohomological Kudla conjecture for unitary Shimura varieties of signature (n+1,1) by reducing modularity to a boundary computation. The architecture is coherent: Section 2 develops the weighted theta formalism and an sl2 action, Section 3 proves the splitting lemma and identifies boundary divisors with abelian varieties, Section 4 computes the boundary theta lift and applies the sl2 projectors, and Section 5 assembles the proof. The weakest point is not internal circularity or an explicit formula I can disprove; it is that the most heavily used external theorem, Theorem 2.12, is unproved in this manuscript and deferred to a forthcoming paper of Howard. The reader identified the same assumption, and the manuscript itself is transparent about this. I also considered possible internal issues: the H-eigenvalue sign in Theorem 4.14, the homology-versus-cohomology notation in Section 5, and the assertion that all Hodge classes on E⊗M are generated by the cycles Z(λ). These look like presentation issues or standard supporting facts rather than errors I can demonstrate. The authors also honestly restrict the main theorem to g ≤ n/2 because of Lemma 3.3. The verdict should therefore remain conditional: the central claim stands if and only if Theorem 2.12 holds with the stated weight, normalization, and Weil representation. A direct small-case verification of Theorem 2.12 is the most efficient way to settle the remaining uncertainty.","tokens_in":29515,"tokens_out":23330,"duration_ms":279770,"concrete_test":"Verify Theorem 2.12 by direct Poisson summation for g = 1 and a rank-one lattice: take M = O_k with the standard Hermitian form, P(λ) = h(λ,λ), and compute ϑ_P(τ+1) and ϑ_P(−τ^{-1}) for several τ; check that the transformation factor is det(Cτ+D)^{n+2} times the Weil representation, with no extra |det|^2 factor and no additional central character. Then repeat for a rank-two lattice with P(λ) = det(h(λ_i,λ_j)) to probe the determinant factor and the exp(−Δ/4π) completion. If the predicted automorphy holds, the main theorem's weight and corrections are internally consistent; if not, the weight in Theorems 1.2–1.5 shifts and the corrections in Section 4.3 need revision.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The single most load-bearing assumption is Theorem 2.12 (Section 2.5): for every P in F_{n,g}, the weighted theta series ϑ_P(τ) = det(Y)^{-1} Σ_λ exp(−Δ/4π)(P)(λ·Y^{1/2}) q^{h(λ)} e_λ transforms as a Hermitian modular form of weight n+2 with representation ρ_{M,g}. The paper says only that a detailed proof will appear in forthcoming work of Ben Howard. This theorem is not peripheral: it is used to assert modularity of the boundary theta lifts in Theorem 4.12, to construct the non-holomorphic completion of Φ^g_M, and, together with the sl2 projectors, to conclude in Theorem 4.19 that the corrected boundary series is modular. Section 5 then transfers this to X^tor via Lemma 3.3. Because the formula involves both a determinant factor and an exponentiated Laplacian acting on polynomials, a small change in normalization—an extra |det| power, a central character twist, or a restriction on the level/discriminant of M—would alter the weight or representation in Theorems 1.2–1.5. The paper provides no internal check of this criterion. The other ingredients (sl2 decomposition, boundary computation, excess intersection formula) are at least sketched, but they are sound only if this unproved modularity theorem is correct. This is an external-validation concern, not an internal contradiction; the reader's conditional verdict is appropriate.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves a cohomological version of the Kudla conjecture for unitary Shimura varieties of signature (n+1,1) in codimensions g ≤ n/2. For a neat arithmetic group and toroidal compactification, the authors construct boundary-supported correction classes to the Kudla–Millson generating series of special cycles and prove that the corrected series is a holomorphic Hermitian modular form of weight n+2 with respect to the genus-g Weil representation. They also construct non-holomorphic completions, exhibiting the uncorrected series as a Hermitian quasi-modular form. The proof combines a splitting lemma for homology, an explicit analysis of the boundary divisors as abelian varieties, a new sl2-action on spaces of homogeneous polynomials, and weighted theta series with completions. The main results are stated as Theorems 1.2–1.4 and Corollaries 1.5–1.6.","tokens_in":29702,"tokens_out":11370,"duration_ms":131623,"significance":"If the proof is completed, this is a substantial advance: it provides the first general modularity result for compactified special cycles of arbitrary codimension in unitary Shimura varieties, going beyond the codimension-one and zero-cycle cases in the literature. The development of Hermitian quasi-modular forms and the explicit sl2-projector formalism are valuable technical contributions. The argument is not circular: the correction terms are constructed from sl2-projections and an external modularity criterion, not fitted to force the theorem. However, the central modularity criterion is deferred to unpublished work, and there are notation mismatches in the main formula that must be resolved before the paper can be accepted.","major_comments":[{"comment":"The paper's central modularity criterion is stated without proof and deferred to forthcoming work of Ben Howard. This theorem is used as the basis for the modularity of the boundary theta lift (Theorem 4.12), for the corrected boundary series (Theorem 4.19), and for the final transfer to X^tor in Section 5; a normalization or weight error in the determinant factor or the exponentiated Laplacian would change the weight or representation in Theorems 1.2–1.5. As written, the main theorems are conditional on an unpublished external result, and no internal consistency check (e.g., a low-genus or low-weight example) is supplied.","section":"§2.5, Theorem 2.12"},{"comment":"The notation in the main correction formula is internally inconsistent. Theorem 1.4 and Corollary 1.5 write the boundary correction as ι_{J,*}[W_i^ℓ ∪ L^{g−ℓ−1}], whereas Definition 5.1 uses D_J^{g−ℓ−1}; Corollary 4.5 gives c1(N∨_{B_J}) = d_k/r_J D_J, so if L denotes the conormal bundle, the two formulas differ by powers of d_k/r_J. In addition, Theorem 1.4 states the summation condition [λ] = p_M^L(ν), but p_M^L maps from (M∨/M)^g to (L∨/L)^g; the correct projection of ν is p^M_L(ν). These mismatches make the theorem statement not directly usable.","section":"§1.1, Theorem 1.4; §2.2; §5, Definition 5.1"},{"comment":"The symbol \\tilde Z is used for two different corrections: in Section 4.3 (before Theorem 4.19) it denotes [Z(λ)] − Σ ... [W_i^ℓ ∪ D_J^{g−ℓ}], while Definition 5.1 defines the boundary correction for the compactified cycle as [Z(ν,N)] + Σ (r_J/d_k) ... [W_i^ℓ ∪ D_J^{g−ℓ−1}]. The sign and the exponent of D_J change between these formulas, and a reader cannot tell whether Theorem 4.19 is compatible with Theorem 1.4. Please state explicitly how the boundary correction of Section 4.3 is converted, via the Gysin map and excess intersection, into the correction in X^tor.","section":"§4.3 vs §5, Definition 5.1"}],"minor_comments":[{"comment":"The proof refers to “Theorem 3.3” for the splitting of homology; the statement is Lemma 3.3.","section":"§5, proof of Theorems 1.3–1.4"},{"comment":"The text says the pairings are Hermitian modular forms of “weight 1+n”; the correct weight is n+2, as stated in Theorems 1.2 and 3.1.","section":"§5, proof of Theorems 1.3–1.4"},{"comment":"The introduction cites “Theorem 3.6” for the Hodge conjecture for the boundary abelian variety, but the actual statement is Remark 3.6.","section":"§1.1 and §3.5"},{"comment":"“Thethetaseries” is a typo for “The theta series”.","section":"§1.1, Corollary 1.6 paragraph"},{"comment":"The proof of Theorem 1.2 asserts the existence of cycles, but the proof in Section 5 defines cohomology classes; please clarify that the corrections are algebraic cycles with Q-coefficients and specify the coefficient field of the polynomials P_i^ℓ(ν,N).","section":"§4.3, definition of \\tilde Z"}],"recommendation":"major_revision","confidential_remarks":"The central dependence on unpublished work of Ben Howard is a publication-risk issue. I recommend that the editors require a proof of Theorem 2.12 or a precise public reference before acceptance, and that the notation mismatches in Theorem 1.4 be corrected in the revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's the short version: this is the first real push on the cohomological Kudla conjecture for unitary Shimura varieties in codimension up to the middle, and the machinery—Hermitian quasi-modular forms, the sl2 projectors, the boundary corrections—looks right. But the entire modularity result depends on Theorem 2.12, a weighted theta series criterion stated without proof and deferred to Ben Howard's forthcoming paper. That is the thing to check.\n\nWhat's new: Bruinier–Zemel and EGT23 handled divisors, BRZ24 handled zero-cycles; this paper handles arbitrary g ≤ n/2 for U(n+1,1). The generating series of corrected Zariski closures is shown to be a holomorphic Hermitian modular form, and the Zariski closure series is a Hermitian quasi-modular form with explicit non-holomorphic completion. That is a genuine advance, and the explicit correction formula with boundary cycles and the sl2-action on polynomial weight functions is a nice piece of work.\n\nWhat the paper does well: the strategy is coherent—reduce to the boundary via Greer's splitting lemma, compute harmonic representatives of boundary cycles, then correct the theta series using projectors onto primitive sl2 components. The paper is honest about the range restriction g ≤ n/2, and the corrections are constructed, not fitted; there is no circularity. The reliance on the Hodge conjecture for abelian varieties is fine; that's a theorem in this setting.\n\nSoft spots, in order of seriousness. First, Theorem 2.12 is load-bearing and unproved. Every non-holomorphic completion and modular correction in Sections 4–5 passes through it. A normalization or weight error there would change the weight or representation in Theorems 1.2–1.5. The paper gives no internal check. This is an external-validation concern, not an internal contradiction, but it is real: the proof, as written, is conditional on that theorem. Second, there are notation mismatches between the introduction (Theorem 1.4, Corollary 1.5) and the proof in Section 5—the definition of P_i^ℓ(ν,N) and the φ formula differ in small but confusing ways. A final version should clean these up. Third, the Splitting Lemma (3.3) is what forces g ≤ n/2; the authors acknowledge it, and it's fine, but it does limit the result.\n\nThe reader's conditional verdict is right. If Theorem 2.12 holds, this is a major result. I didn't see any circular dependence on the main claim.\n\nBottom line: send it to a good referee. The referee's main job is to check Theorem 2.12 against Howard's work and to verify the boundary transfer. If that checks out, publish. I would cite it.","headline":"Substantial and likely correct advance on the cohomological Kudla conjecture for U(n+1,1), but the central modularity criterion is deferred to forthcoming work and must be checked by the referee.","tokens_in":30349,"tokens_out":2663,"would_cite":true,"duration_ms":27898,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11F27","11F46","11G18","14C25","14G35"],"pacs":[],"model":"deepseek-v4-flash","headline":"For unitary Shimura varieties of signature (n+1,1), boundary-supported corrections make the generating series of special cycles a holomorphic Hermitian modular form of weight n+2 in all codimensions g≤n/2.","keywords":["unitary Shimura varieties","Kudla–Millson generating series","Hermitian modular forms","Hermitian quasi-modular forms","special cycles","toroidal compactification","theta series","sl2-action"],"falsifier":"Compute the Fourier coefficients of the corrected series $eΦ^{1}$_L(τ) for the smallest nontrivial case, say a signature (3,1) lattice with k=Q(i) and g=1, using the explicit correction formula of Theorem 1.4, and compare them with the Fourier coefficients of the known weight-4 Hermitian Eisenstein series for U(1,1)(Z); a mismatch at any positive-definite matrix N would disprove the central modularity claim.","tokens_in":29207,"feed_emoji":"📐","tokens_out":8755,"duration_ms":94809,"temperature":0.7,"pith_summary":"The paper aims to prove the cohomological version of the Kudla conjecture for toroidal compactifications of unitary Shimura varieties of signature (n+1,1): the generating series of special cycles, after adding cycles supported in the boundary, should be a holomorphic Hermitian modular form valued in cohomology. The authors establish this for all codimensions g up to n/2, and they also show that the uncorrected series of Zariski closures is a Hermitian quasi-modular form with an explicit non-holomorphic completion. If correct, this resolves a conjecture of Kudla and Bruinier–Rosu–Zemel in cohomology in that range and provides the first general construction of compactified-cycle modularity for arbitrary codimension. The restriction g≤n/2 is traced to a splitting lemma, and the authors conjecture the same results hold in all codimensions.","feed_headline":"Cohomological Kudla conjecture proved up to middle codimension","feed_subtitle":"Boundary-supported corrections turn special-cycle series into holomorphic Hermitian modular forms.","key_machinery":"The central mechanism is the sl2-action on the finite-dimensional spaces F_{n,g} of polynomials P:M_{n×g}(C)→C satisfying the equivariance P(U A)=|det(A)|^2 P(U). The lowering operator Δ, raising operator Λ, and weight operator H form an sl2-triple, so each polynomial decomposes into primitive (pluriharmonic) pieces $Λ^{{g-ℓ}}$$P^{{ℓ,ℓ}}$. A modularity criterion (Theorem 2.12) turns a polynomial P into a $\\theta$ series ϑ_P(τ)=det(Y)^{-1} Σ_λ exp(-Δ/4π)(P)(λ·$Y^{{1/2}}$) $q^{{h(λ)}}$ e_λ that transforms like a Hermitian modular form after completion. On each boundary component the special cycles are represented by harmonic forms f^g(λ)=g! f(λ_1)∧⋯∧f(λ_g), and pairing these with a cohomology class produces an element of F_{n,g}; this map intertwines the sl2-actions, with Lefschetz on cohomology matching Λ/Δ on polynomials. The other load-bearing ingredient is the Splitting Lemma, which uses Hard Lefschetz on the normal bundle of each boundary divisor to express every cohomology class in degree 2g≤n as an interior class plus a boundary class, reducing modularity of the full series to a computation in the boundary abelian varieties E_M.","core_discovery":"On the paper's own terms, the central discovery is that the non-compactness of a ball quotient does not destroy the modularity of Kudla–Millson generating series; it merely moves the series into a larger space. For each isotropic boundary line J, with boundary divisor B_J ≅ E ⊗_{O_k} M, the paper computes the restriction of special cycles to B_J, represents those cycles by harmonic forms f(λ)=u_λ^*ω_E, and uses an sl2-triple (Λ, Δ, H) on the space of polynomial weight functions to interpolate between primitive classes and their Lefschetz powers. The correction term in Theorem 1.4 is the explicit boundary cycle Σ_{J} Σ_{ℓ,i} (r_J/d_k) P_i^ℓ(ν,N) ι_{J,*}[W_i^ℓ ∪ $L^{{g-ℓ-1}}$], and Corollary 1.5 packages these corrections as the statement that eΦ^g_L(τ) transforms with weight n+2 and representation ρ_{L,g} under U(g,g)(Z). Theorem 1.3 asserts the same for the uncorrected series after a non-holomorphic completion.","pith_inferences":["The sl2-projector mechanism is independent of the Splitting Lemma, so a finer boundary analysis could plausibly extend the result beyond the middle without changing the modularity criterion itself.","The explicit correction formula suggests a practical numerical test: for small n and g, pairing eΦ^g_L with test classes should reproduce Fourier coefficients of known Hermitian Eisenstein series, giving a check that does not depend on the deferred proof of Theorem 2.12.","The same Lefschetz-graded correction idea should apply to orthogonal Shimura varieties in higher codimensions, extending the known codimension-one case to a full generating series of compactified cycles.","If the deferred modularity criterion is supplied and the convergence issue for Chow-valued series is resolved, the same correction formulas would likely upgrade from cohomology to Chow groups, matching the original conjecture."],"forward_implications":["For every neat arithmetic group and every g≤n/2, the corrected generating series eΦ^g_L(τ) is a holomorphic Hermitian modular form of weight n+2 with representation ρ_{L,g}, giving the cohomological form of Conjecture 1.1 in the unitary case.","The uncorrected generating series of Zariski closures of special cycles is a Hermitian quasi-modular form, the first instance of Hermitian quasi-modularity appearing for special cycles.","The explicit correction formula makes the modularity effective: Fourier coefficients of the corrected series can be read from intersection numbers with explicit boundary cycles W_i^ℓ ∪ L^{g-ℓ-1}.","The authors conjecture the same statement holds in all codimensions, since the restriction g≤n/2 is an artifact of the splitting lemma rather than of the theta-lift mechanism."],"supporting_citations":[{"why":"Supplies the base theorem that the uncorrected generating series of special-cycle cohomology classes on the open Shimura variety is already a holomorphic Hermitian modular form.","marker":"[KM90]"},{"why":"Provides the model of toroidal compactifications of unitary Shimura varieties, including the boundary divisors, normal bundles, and the intersection of closures with the boundary.","marker":"[How15]"},{"why":"Source of the Splitting Lemma used to reduce modularity of cohomology classes to classes supported in the boundary.","marker":"[Gre19]"},{"why":"Previous work on special divisors in the orthogonal case whose completion-and-correction strategy the paper generalizes to higher codimension.","marker":"[EGT23]"},{"why":"Formulates the conjecture in the form proved here and proves the zero-cycle case on toroidal compactifications.","marker":"[BRZ24]"},{"why":"Provides the orthogonal analogue of the weighted theta series criterion that Theorem 2.12 adapts to the Hermitian setting.","marker":"[Roe21]"},{"why":"Together with Murasaki, establishes the Hodge conjecture for the boundary abelian varieties E_M, giving algebraic cycle representatives for the primitive Hodge classes W_i^ℓ.","marker":"[Tat65]"},{"why":"Completes the Hodge-class computation for abelian varieties used to justify the basis of primitive cycles in the boundary.","marker":"[Mur69]"},{"why":"Supplies the structure of pluriharmonic polynomials used in the decomposition of F_{n,g} into primitive pieces.","marker":"[KV78]"}],"fun_headline_variants":["Cohomological Kudla conjecture resolved","Kudla conjecture proven for unitary Shimura varieties","Special cycles become holomorphic Hermitian modular forms","Boundary corrections yield modularity for special cycle series"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is Theorem 2.12 — the criterion that every polynomial-weighted theta series can be completed to a Hermitian modular form — which the paper states without proof, defers to forthcoming work of Ben Howard, and uses for every completion and correction.","fun_headline_variants_meta":{"raw":{"variants":["Cohomological Kudla conjecture resolved","Kudla conjecture proven for unitary Shimura varieties","Special cycles become holomorphic Hermitian modular forms","Boundary corrections yield modularity for special cycle series"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000684,"raw_usage":{"total_tokens":3075,"prompt_tokens":887,"completion_tokens":2188,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":503,"completion_tokens_details":{"reasoning_tokens":2137}},"tokens_in":503,"tokens_out":2188,"duration_ms":19605,"temperature":1.0,"reasoning_tokens":2137,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T16:27:08.712548+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the Fourier coefficients of the corrected series $eΦ^{1}$_L(τ) for the smallest nontrivial case, say a signature (3,1) lattice with k=Q(i) and g=1, using the explicit correction formula of Theorem 1.4, and compare them with the Fourier coefficients of the known weight-4 Hermitian Eisenstein series for U(1,1)(Z); a mismatch at any positive-definite matrix N would disprove the central modularity claim.","supporting_citations":[],"review_version":1}