{"id":"a0a82ba9-237e-4533-9a44-79205f3c7ccd","arxiv_id":"2602.01473","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A geometric theta lift maps modular symbols to products of two weight-one Eisenstein series and modular caps to weight-two Eisenstein series, yielding a geometric proof of Borisov-Gunnells-style relations.","lead":"This paper uses a geometric theta lift to evaluate cycles on modular curves, showing that modular symbols map to products of two weight-one Eisenstein series and modular caps to weight-two Eisenstein series. This yields a geometric proof of known relations among Eisenstein series and recovers spanning results of Borisov-Gunnells and Li.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Singular-vector analytic continuation in Prop. 4.1 is load-bearing: no remainder estimate is given for the s→0 limit that yields the cap period; a Fourier-coefficient consistency check would settle it.","rationale":"The reader's weakest assumption identifies the same load-bearing point: the analytic continuation of the singular-vector contribution at s=0 in the cap-period computation. I agree that this is the step on which Theorem 1.3 and therefore Theorem 1.5 depend. The paper is candid about the fact that the sum-integral interchange is only justified for Re(s) large, but no remainder estimate or independence-of-continuation argument is provided. Since the non-singular Fourier coefficients are fixed by independent structure (Theorem 3.6 and Hecke operators), an internal consistency check on the cap period is both feasible and decisive. I do not see a fatal flaw in the strategy: the singular contribution is a Kronecker-Eisenstein series of the sort for which meromorphic continuation is classical, and the surrounding cap/symbol decomposition is coherent. The smaller Theorem 4.7 reduction slip is a presentational error in the proof as written, but the intended correction is straightforward and does not change the verdict. The appropriate disposition remains conditional acceptance pending a rigorous justification of the s=0 limit and a corrected reduction in Theorem 4.7.","tokens_in":28660,"tokens_out":33441,"duration_ms":310863,"concrete_test":"Take N=5, p=1, q=0. Compute the q-expansion of H^{(2)}_{1,0} to order q^{10} from its definition. Independently compute ∫_{C_∞} E_{1,0} from the τ-Fourier expansion of Theorem 3.6: a_0(E^{(2)}_{0,1}) + Σ_{n≥1} ⟨C_∞, T^n{0,∞}⟩ q^n, where the intersection numbers are evaluated by expressing T^n{0,∞} in the cap/symbol basis via Theorem 2.3. If the two expansions disagree, the singular-vector continuation in Prop. 4.1 is wrong; agreement to q^{10} would confirm the cap-period formula.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central theorem 1.3, and hence the Eisenstein relation theorem 1.5, rests on Proposition 4.1's cap-period computation. In that proof the (2,0)-component is split as δ_{q0}A(τ,v,s) plus regular terms; the singular part A is obtained by integrating the theta series term-by-term and then taking s=0. The paper states in Section 3.6 that for singular vectors 'the exchange can only be done for Re(s) large enough' and then proceeds via the functional equation to evaluate A at s=0; no bound on the remainder, nor a proof that the two analytic continuations (term-wise integral and the full integral) coincide at s=0, is supplied. This is not cosmetic: the resulting equality lim_{v→∞} E_{p,q}(z,τ)=H^{(2)}_{p,q}(τ)du is exactly what Proposition 4.2 uses to get ∫_{C_r}E=H^{(2)}_{j,-n}. If the continuation limit picked up a spurious constant term or a residual contribution, the triangle relation would fail. The non-singular part of E is independently controlled by Theorem 3.6 and the Hecke/Poincaré-duality structure of the Fourier coefficients, so the singular-vector assertion is the one unverified link. A second, smaller gap is the reduction in Theorem 4.7: after replacing n_2 by n_2-k_2N, n_3' should be -n_1-n_2', not n_3-k_3N; this is easily repaired and does not affect the analytic issue.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs a geometric theta lift E from H_1(Y_1(N);Z) to weight-2 modular forms for Γ_1(N), using the author's previous construction in [Bra26b]. It evaluates the lift on the two families of generators supplied by Stevens' splitting of the homology: modular caps are sent to weight-2 Eisenstein series H^{(2)}_{j,-n} (Theorem 1.3(1)), and unimodular symbols are sent to products of two weight-1 Eisenstein series (Theorem 1.3(2)). From the decomposition of a hyperbolic cycle into caps and symbols (Theorem 2.3), it derives formulae for the images of parabolic and hyperbolic cycles (Theorem 4.4) and, by applying the lift to a hyperbolic triangle closed by caps, obtains the Borisov–Gunnells relation G_aG_b+G_bG_c+G_cG_a=G_a^{(2)}+G_b^{(2)}+G_c^{(2)} for a+b+c≡0 mod N (Theorem 1.5). The paper also identifies the hyperbolic-cycle image with a diagonal restriction of a Hilbert–Eisenstein series (Theorem 1.2), thereby connecting Li's spanning theorem with the Borisov–Gunnells picture. The exposition is detailed in §4.1–4.2, and §2.7 provides an independent constant-term verification of the main Eisenstein relation.","tokens_in":29075,"tokens_out":27357,"duration_ms":264651,"significance":"If the central computations are correct, the paper gives a genuinely geometric proof of the Borisov–Gunnells relations and explains how Li's result and the Borisov–Gunnells result arise from the same theta correspondence. The construction is parameter-free and the main relation is checked independently by a constant-term identity in §2.7, which is a real strength. The explicit formulae for the images of caps and symbols are concrete and falsifiable. However, acceptance depends on closing a nontrivial analytic-continuation gap in the singular-vector contribution (Proposition 4.1), and on supplying the deferred integral evaluation in Proposition 4.6. These are load-bearing rather than cosmetic issues.","major_comments":[{"comment":"In the proof of Theorem 3.6 the author explicitly states that, for singular vectors, the interchange of the sum over the lattice and the integral over A_R is only valid for Re(s) large enough. In Proposition 4.1, however, the (2,0)-component is decomposed as δ_{q0}A(τ,v,s) plus regular terms, and A(τ,v,s) is then evaluated at s=0 after a term-by-term integration and the use of the functional equation. No remainder estimate is supplied, and no argument is given that the analytic continuation of the term-wise integral agrees with the analytic continuation of the full integral at s=0. This is not a cosmetic omission: Proposition 4.2 uses the resulting limit lim_{v→∞}E_{p,q}=H^{(2)}_{p,q}du to compute cap periods, and Theorem 1.5 depends on those periods. The gap could be closed by a uniform estimate justifying the interchange at s=0, or by an independent Fourier-coefficient consistency chec","section":"§3.6 and §4.1, Proposition 4.1"},{"comment":"The key integral evaluation for hyperbolic cycles is deferred with the phrase 'the computation is similar to the computation in [Bra26b]' and is not actually carried out in the paper. Since Theorem 1.2 and the connection to Li's result rest on this proposition, the manuscript should include the full computation or at least a precise reduction to the cited result, including the unfolding step, the convergence, and the treatment of the quotient by units. In its current form this is an unverified load-bearing step.","section":"§4.4, Proposition 4.6"},{"comment":"The proof of the main relation contains two local sign/index slips. First, after replacing n_2 by n_2-k_2N, the residue n_3' needed to maintain n_1+n_2'+n_3'=0 is n_3-(k_3-k_2)N, not n_3-k_3N. Second, for the caps in the triangle, applying Proposition 4.2 to C_1=γ_{31}[0,1]_∞ (with γ_{31}=(m_1 m_3; -n_1 -n_3)) gives +G_{n_1}^{(2)}, and similarly for the other two caps; the displayed minus signs in ∫_{C_i}E=-G_{n_i}^{(2)} are inconsistent with the preceding equality ∫_M E = -∫_C E and, if taken literally, would prove the negative of the stated relation. Since the theorem's conclusion is confirmed by the independent constant-term check in §2.7, this is repairable, but the proof as printed needs correction.","section":"§4.5, Theorem 4.7"}],"minor_comments":[{"comment":"Small typos: 'group homorphism' should be 'group homomorphism', and the heading 'Pairing and and cohomology' has a duplicated 'and'.","section":"§2.1 and §2.4"},{"comment":"The proof concludes ∫_{γ{0,∞}}E=G_{-c}G_d before the stated formula -G_dG_c; the identity G_{-c}^{(1)}=-G_c^{(1)} is used implicitly. This identity should be stated explicitly, since without it the sign jump is hard to follow.","section":"§4.2, Proposition 4.3"},{"comment":"The notation C_1=[r_3,r_2]_{r_1}=γ_{31}[0,1]_∞ conflicts with the definition of [x,y]_r in §2.2, where [x,y]_r is oriented from π_r(y) to π_r(x). The figures and the text would be much clearer if the orientation conventions for caps were reconciled explicitly.","section":"§4.5, Fig. 5 and cap notation"}],"recommendation":"major_revision","confidential_remarks":"The manuscript relies heavily on the author's prior work [Bra26b], including the construction of the lift and part of the hyperbolic-cycle computation; the referee should have access to that preprint when assessing Proposition 4.6. The main obstacle is the singular-vector analytic continuation in Proposition 4.1: the paper itself flags the problem in §3.6 but does not resolve it. If that gap cannot be closed, the cap-period formula and hence the central theorem are unsupported. The sign/index slips in Theorem 4.7 are local and easily fixed, but they should be corrected before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper is worth engaging with. It evaluates a geometric theta lift on Stevens' generators of homology: caps go to weight-2 Eisenstein series, unimodular symbols to products of weight-1 Eisenstein series. From the boundary of hyperbolic triangles it then derives Borisov-Gunnells relations. If the evaluations are correct, this unifies Li's diagonal-restriction spanning result and the BG spanning result, and gives a natural SL_n generalization. The explicit period formulas in Theorem 1.3 are new in this geometric form, and the paper is candid about where it stands relative to BCG and Xu.\n\nThe central computation is mostly spelled out. Proposition 4.1 computes the constant term of E_{p,q} at the cusp, and Proposition 4.3 computes the integral over {0,∞}; both are detailed, and the constant-term check in §2.7 is an independent sanity check. The relation in Theorem 1.5 is a clean consequence of the triangle boundary.\n\nNow the soft spots. The main one is exactly what the stress-test flags: the singular-vector contribution to the constant term in Prop. 4.1 is handled by term-wise integration valid only for Re(s) large, then continued to s=0 with no remainder estimate. The paper says this in §3.6 and then does it anyway. That is a real gap, and it's load-bearing because the cap period formula depends on it. It may be fixable with standard estimates, but right now it's a missing proof, not a matter of taste.\n\nSmaller issues: Proposition 4.6 defers an integral evaluation to [Bra26b] with 'similar' — that should be spelled out or given a reference to a precise lemma. The identity (3.8) (E(ω_f) proportional to L(f,1) f) is used without proof; I'd want a pointer to BG or a few lines. And in Theorem 4.7 the reduction step has the index error noted in the stress-test: after replacing n_2, the n_3 replacement should preserve the zero sum, which the current text doesn't do. That's easily repaired.\n\nNone of this kills the strategy. The inclusion results in Corollary 4.4.1 are honest and the paper distinguishes inclusion from equality. The citation pattern looks appropriate; the self-citations to [Bra26b] are to the construction, which is the right place.\n\nWho is this for? People working on Eisenstein cohomology, theta lifts, and modular symbols. It deserves a serious referee, but the referee should push on Prop. 4.1's analytic continuation and Prop. 4.6 before signing off.","headline":"A genuinely useful theta-lift computation, with one load-bearing analytic-continuation step that needs a written remainder estimate before it's fully rigorous.","tokens_in":29523,"tokens_out":2595,"would_cite":true,"duration_ms":25843,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11F27","11F67","11F75"],"pacs":[],"model":"deepseek-v4-flash","headline":"A geometric theta lift maps modular caps and symbols to explicit Eisenstein series, yielding a geometric proof of the Borisov-Gunnells relations.","keywords":["theta correspondence","modular symbols","modular caps","Eisenstein series","Borisov-Gunnells relations","Borel-Serre compactification","Hilbert-Eisenstein series","weight-2 modular forms"],"falsifier":"For a small level N and a specific cusp, compute numerically the integral of the theta kernel over a modular cap and compare its q-expansion with H^{(2)}_{d,-c}(τ); a single coefficient mismatch would indicate the analytic continuation at s=0 is invalid and the relation would not follow.","tokens_in":28576,"feed_emoji":"📐","tokens_out":8738,"duration_ms":81828,"temperature":0.7,"pith_summary":"The paper constructs a theta lift from the first homology of the modular curve Y1(N) to weight-2 modular forms, and evaluates it on Stevens' generators of the homology: modular caps and unimodular symbols. It finds that the period of the theta form over a cap is a weight-two Eisenstein series, and over a unimodular symbol it is a product of two weight-one Eisenstein series. Because the boundary of a hyperbolic triangle is trivial in homology, the sum of these periods gives a linear relation among Eisenstein series—the Borisov-Gunnells relations. The same theta lift also sends hyperbolic cycles to diagonal restrictions of Hilbert-Eisenstein series, recovering Li's spanning theorem, so the two known spanning results arise from one geometric mechanism. A sympathetic reader would care because the relations are proved by geometry rather than by Fourier-coefficient manipulation, and the construction carries a natural generalization to higher rank.","feed_headline":"Theta lift proves the Borisov-Gunnells Eisenstein relations","feed_subtitle":"Integrating a theta kernel over modular-curve cycles recovers both Borisov-Gunnells and Li spanning results.","key_machinery":"The central object is the theta lift E: H1(Y1(N);Z) → M2(Γ1(N)), induced by a closed differential form E(z,τ) on the modular curve that transforms as a weight-2 modular form in τ; its Fourier coefficients are Poincaré duals of Hecke translates of the modular symbol {0,∞}. The form is built from a Mathai-Quillen Thom form on a rank-two bundle over the symmetric space, pulled back by sections and paired with a theta series of the Weil representation. The other load-bearing piece is Stevens' description of the homology, which gives a splitting H1(Y1(N);Z) ≅ C(Z) ⊕ MS0(Z) into modular caps (loops around cusps) and degree-zero modular symbols; the paper evaluates the theta lift on these generator","core_discovery":"On the paper's own terms, the central discovery is that the theta lift E, defined by periods of a closed differential form, takes explicit values on the generators of H1(Y1(N);Z) that Stevens' Borel-Serre description provides. Over a modular cap Cr at a cusp r = γr∞, the period is the weight-two Eisenstein series H^{(2)}_{d,-c}(τ); over a unimodular symbol γ{0,∞}, it is -G^{(1)}_d(τ)G^{(1)}_c(τ) (Theorem 1.3). Since a hyperbolic triangle closed by three caps and three unimodular sides is a boundary, its total period is zero, which yields the relation G_a G_b + G_b G_c + G_c G_a = G_a^{(2)} + G_b^{(2)} + G_c^{(2)} for a+b+c≡0 mod N (Theorem 1.5). The author presents this as a geometric proof","pith_inferences":["The paper states that the construction generalizes to SL_n(R)/SO(n) and weight-n forms; a natural test is whether the d-gon relation becomes a polytope relation in the Borel-Serre boundary of the higher-rank symmetric space.","The explicit formulas invite a coefficient-by-coefficient check of the triangle relation from the known q-expansions of G^{(1)} and G^{(2)}; the geometric proof here does not by itself show the relations hold beyond the constant term.","One could seek new relations among higher-weight Eisenstein series by integrating the higher-weight analogue of the theta form over higher-dimensional geodesic cycles, which the present weight-1/weight-2 framework does not cover."],"forward_implications":["The Borisov-Gunnells relations follow from the vanishing of the boundary of a hyperbolic triangle in Borel-Serre homology, giving a geometric proof rather than a coefficient-wise one.","The d-gon version (Corollary 4.7.1) shows the same argument produces relations for any geodesic polygon with unimodular sides: the sum of products of weight-one Eisenstein series equals the sum of weight-two Eisenstein series at the cusps.","Both known spanning results—Borisov-Gunnells' spanning by Eisenstein spaces and Li's spanning by diagonal restrictions of Hilbert-Eisenstein series—arise from a single theta lift, with hyperbolic cycles giving the diagonal restrictions.","The image of the theta lift contains H^(2) ⊕ S^{new}_{2,rk=0}; for prime N this containment is an equality, so the geometric construction accounts for exactly the Eisenstein and rank-zero newform parts of the weight-2 space."],"fun_headline_variants":["Theta lift yields geometric proof of Eisenstein relations","Periods of theta kernel prove Borisov-Gunnells","Modular cycles give geometric proof of Eisenstein identities","Theta correspondence explains Borisov-Gunnells relations","Homology cycles and theta lift prove Eisenstein relations"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The proof evaluates the singular-vector contribution to the theta lift's constant term at s=0 by analytic continuation, and the paper notes the sum-integral interchange is only valid for Re(s) large without giving a full remainder estimate; if this continuation were wrong, the cap and symbol periods, and with them the Eisenstein relation, would fail.","fun_headline_variants_meta":{"raw":{"variants":["Theta lift yields geometric proof of Eisenstein relations","Periods of theta kernel prove Borisov-Gunnells","Modular cycles give geometric proof of Eisenstein identities","Theta correspondence explains Borisov-Gunnells relations","Homology cycles and theta lift prove Eisenstein relations"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000137,"raw_usage":{"total_tokens":951,"prompt_tokens":676,"completion_tokens":275,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":420,"completion_tokens_details":{"reasoning_tokens":200}},"tokens_in":420,"tokens_out":275,"duration_ms":3574,"temperature":1.0,"reasoning_tokens":200,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T05:38:04.284328+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For a small level N and a specific cusp, compute numerically the integral of the theta kernel over a modular cap and compare its q-expansion with H^{(2)}_{d,-c}(τ); a single coefficient mismatch would indicate the analytic continuation at s=0 is invalid and the relation would not follow.","supporting_citations":[],"review_version":1}