{"id":"1714281b-f25b-4aee-ae5d-d6739d7d6feb","arxiv_id":"2601.13467","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"Witness-filtered Berry curvature, quantum geometric tensor, and quantum Fisher information obey exact lattice identities in the spinless Haldane model, with gap-crossing jumps glossed as Hecke modifications under a Langlands-style duality.","lead":"This paper filters the geometric response of the Haldane topological-insulator model through an entanglement-witness operator, splitting Berry curvature, the quantum metric, and quantum Fisher information into sectors that carry sublattice coherence. It derives exact lattice identities for those sectors and frames phase-transition jumps in the language of Hecke modifications and a Langlands-style duality.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Hecke formula (Eq. 10) is applied to a quantity that is not an index: in the Haldane embedding S is not an endomorphism of the valence bundle, so ν_S = -2 Re(e^{iθ} J_F) is a continuous real number, not an integer; the claimed quantized Hecke jumps of ν_S are unsupported.","rationale":"The reader correctly identified the unproven integrality of the ν_S jumps as a weak point. My stress test strengthens this from 'unproven' to 'not applicable as stated': the Hecke index (6) requires an S-invariant splitting of the vector bundle, but in the single-orbit Haldane construction S acts on a two-mode Fock space and the valence line is not an eigenline of S. The lattice quantity ν_S is therefore a scalar curvature weighted by an expectation value, not a trace of S against a matrix curvature; its values are generically continuous. This undermines the central claim that Eq. (10) describes the jump of the entanglement response. The algebraic identities (19), (25), and the QFI bounds (34)-(40) appear correct and are not the target of this objection; they are consistent with an interpretation as a coherence filter. The paper should be re-scoped to either construct a nontrivial S-invariant bundle or drop the Hecke/Langlands claims. Because the headline claim as stated is internally inconsistent with the definitions, I recommend REJECT for the current version, with the possibility of a revised conditional version after removing or repairing the Hecke identification.","tokens_in":14278,"tokens_out":19805,"duration_ms":193695,"concrete_test":"Run an FHS computation for the spinless Haldane model with t2 = 1/3, φ = π/2. For θ = 0, compute J_F = (1/2π) Σ_□ F(k) v_A(k) v_B(k)* and hence ν_S = -2 Re(J_F) for M = 1.0 and M = 2.0 (on either side of the gap-closing value M_c = √3), using N×N meshes with N = 50, 100, 200. If the extrapolated ν_S values are non-integer, or the difference Δν_S across the jump is not an integer (and not merely equal to the Chern-number jump Δμ = ±1), then Eq. (10) does not describe ν_S. As an analytic pre-check, verify at a generic k that ⟨u_+|S'|u_-⟩ ≠ 0; if so, the valence bundle is not S-invariant and the Sec. II index theorem cannot be invoked for this model.","verdict_should_be":"REJECT","load_bearing_attack":"The central load-bearing step is the transition from the exact FHS identities (19)/(25) to the Hecke-jump formula (10). In Sec. II, Ind_S is defined as (1/2πi)∫ Tr(S F_A) for an A-parallel endomorphism S of the vector bundle E; because S splits E into ± eigensubbundles, Ind_S is an integer (a difference of Chern numbers). In the Haldane application, however, S in Eq. (17) is a constant 2×2 operator on the ambient single-excitation Fock space, not an endomorphism of the rank-1 valence bundle. The valence line is not S-invariant: generically [S, |u_-⟩⟨u_-|] ≠ 0, and indeed Eq. (31) gives P_⊥ S P_⊥ = η P_⊥ with η = -⟨S⟩, which is ±1 only in the maximally entangled case. Consequently the quantity computed in Eq. (19), ν_S = (1/2π)Σ F ⟨Ψ|S|Ψ⟩ = -2 Re(e^{iθ} J_F), is not of the form (6); it is a continuous real number, not an integer index. There is no argument that the jump of ν_S across the gap-closing stratum is an integer, and the phrase 'quantized jumps' for ν_±, ν in Sec. III.B is not supported by Eq. (19). The Hecke formula (10) would describe changes of c_1(det E_+) - c_1(det E_-) for a true S-invariant splitting; it does not apply to a witness expectation value against a single scalar Berry curvature. Thus the paper's headline claim that the entanglement response itself undergoes Hecke jumps conflates two distinct objects.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops 'witness-filtered' geometric response diagnostics for the spinless Haldane model on a Fukui–Hatsugai–Suzuki (FHS) discretized Brillouin torus. The occupied Bloch spinor at each k is embedded via an isometry J into the single-excitation sector of a two-mode Fock space, and an entanglement witness sign operator S = sgn(W) defines sector weights. The authors derive exact lattice identities ν±(W) = μ/2 ± Re(e^{iθ}J_F^{AB}) and ν_S = ν+ − ν− (Eq. (19)), with a multi-orbital matrix generalization (Eq. (25)), verified numerically to machine precision. They introduce S-filtered QGT and QFI and prove bounds |F^{Q,(S)}| ≤ F^Q and |Im Q^{(S)}| ≤ (C/2)|F|, Eqs. (34)–(40). The paper further claims that across the gap-closing stratum the quantized response changes realize Hecke modifications and support an 'entanglement-sensitive Langlands correspondence,' thereby validating a proposed picture of entanglement as a cohomological obstruction.","tokens_in":14800,"tokens_out":30600,"duration_ms":310824,"significance":"Concrete results: (i) identities (19)/(25) are exact algebraic consequences of the definitions, free of fitting parameters, and confirmed at machine precision on FHS meshes; (ii) the two-witness tomography reconstructs J_F without ambiguity; (iii) the filtered-QGT/QFI bounds (34)–(40) are elementary, correct, and numerically confirmed. As a coherence-filter diagnostic for two-band Chern insulators these are solid, reproducible contributions. What is not established is the interpretive superstructure: ν_S is not shown to be an integer index, and the Hecke jump formula (10) does not apply to it (Major comment 1). The paper's headline claims about Hecke modifications and an entanglement-sensitive Langlands correspondence therefore outrun the mathematics, even though the underlying lattice identities are correct.","major_comments":[{"comment":"Eq. (10) applies to Ind_S = (2πi)^{-1}∫Tr(SF_A), an integer only when S is an A-parallel endomorphism of E splitting E into ± eigenbundles. In the Haldane application, S|_{S_1} in (17) is a constant 2×2 matrix on the ambient two-qubit space; the valence line L_k = J|u_-(k)⟩ is generically not S-invariant ([S, |u_-⟩⟨u_-|] ≠ 0; Eq. (31): P_⊥S′P_⊥ = ηP_⊥ with η = −⟨S⟩, ±1 only for maximally entangled states). Hence S is not an endomorphism of the rank-1 valence bundle, D_AS ≠ 0, and ν_S in (19) — equal to −2 Re(e^{iθ}J_F^{AB}) — is a continuous real number, not the index (6). The Fig. 1(d)/§III.B claim that μ, ν±, ν exhibit 'quantized jumps' realizing ΔInd_S is unsupported: (10) gives integer ΔInd_S, while Δν_S = −2 Re(e^{iθ}ΔJ_F^{AB}) has no demonstrated integrality (for an isotropic Dirac point the local ΔJ_F vanishes by angular symmetry, so a nonzero integer jump is not automatic). The H","section":"§III.B (19) vs §II (6),(10)"},{"comment":"The sentence 'The boundary contribution supported on ∂N_ε(Σ) implements the middle extension j_{!*} that restores quantization' is stated without proof or reference. For a smooth bundle on the punctured surface C°, the limit ε→0 of ∫_{X_ε}Tr(SF_A) is not obviously an integer; integrality depends on the extension data at the punctures, and the link between a boundary term and the sheaf-theoretic middle extension requires argument. Since this excision claim underlies the Hecke interpretation, give a precise statement with hypotheses and proof (or citation), or remove it and present the Hecke picture explicitly as an analogy.","section":"§II (after Eq. (2))"},{"comment":"All entanglement statements pass through the isometry J into the single-excitation two-mode Fock space; 'mode entanglement' 2|v_Av_B| is a modeling choice, not a computed many-body entanglement measure, and the witness W lives in the same local two-qubit model. Consequently the §V conclusion that the numerics 'validate ... entanglement, viewed as a geometric obstruction' tests the obstruction picture only within this embedding and is partly self-referential. The lattice identities are unaffected, but the paper should state this restriction prominently (the abstract's phrase 'single-particle mode entanglement' carries the load) or add an independent entanglement probe.","section":"§III.A (15), §V"}],"minor_comments":[{"comment":"The witness phase θ is first defined as arg⟨v_Bv_A^*⟩ (a fixed reference) but later swept as a free parameter in the two-witness tomography. Clarify that θ is a tunable witness phase and that the definition is just one particular choice.","section":"§III.B (17)–(19)"},{"comment":"F^{Q,(S)} = ηF^Q can be negative since η∈[−1,1]; calling it a Fisher information is misleading, as QFI is positive semidefinite by definition. Suggest 'signed filtered QFI' or use |η|, noting that (34) already employs absolute values.","section":"§IV.A (30)–(32)"},{"comment":"Panel (d) is asserted to show 'quantized jumps' of μ, ν±, ν, but the numerical values of the jumps at the two critical M are not reported. In light of Major comment 1, report the jump sizes and check integrality explicitly.","section":"§III.B / Fig. 1(d)"},{"comment":"Equation (7) is written for the twist torus T²_Φ, while the FHS computation in §III.A is performed on the Brillouin-zone torus T²_k with U_α(k) = ⟨u_-(k)|u_-(k+α̂)⟩. The two tori are different objects; please align the notation, also in (33).","section":"§II (7) / §III.A"},{"comment":"Reference [9] has a duplicated DOI URL, and the arXiv title ('Towards Entanglement-Sensitive Langlands Data') differs from the in-text title ('Towards an Entanglement-Sensitive Langlands Correspondence'). Choose one.","section":"Refs. [9], title"}],"recommendation":"major_revision","confidential_remarks":"The core FHS identities and QFI bounds are correct and reproducible; the main risk is the interpretive layer. The Hecke/Langlands framing is currently not supported by the mathematics because ν_S is not an index; I recommend major revision with an emphasis on either proving an index property for a suitably constructed S or explicitly demoting the Hecke/Langlands content to analogy restricted to the Chern number. The machine-precision residuals are algebraic consistency checks and should not be presented as empirical validation of the entanglement-obstruction picture. With an honest reframing, the filtered-coherence diagnostic is publishable."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"There are two papers bolted together here. The first is an exact algebraic result about witness-filtered Berry curvature in the Haldane model; the second is an attempt to dress that result in Hecke modifications and a Langlands correspondence. The first is solid, genuinely new, and useful. The second does not hold up, and I think the stress-test note is right about why.\n\nThe core content: Eqs. (19) and (25) are exact on the FHS lattice, verified to machine precision against a standard one-particle Chern insulator with no fitting. The multi-orbital matrix generalization and two-witness tomography of J_F are clean and, as far as I can tell from the cited literature, original. The QFI/QGT relations in Sec. IV are elementary but correct: the Schwarz bound and the |η| ≤ C(k) inequality follow from the two-state Bloch representation, and the saturation analysis for the maximally entangled case is right. Anyone working on coherence-resolved topological diagnostics will want these identities. That part of the paper survives removing every Langlands word.\n\nWhere it falls apart: the Hecke formula Eq. (10) is applied to a quantity that is not an index. In Sec. II, Ind_S is defined for an endomorphism S that preserves the valence bundle splitting, guaranteeing an integer. In the Haldane application, S from Eq. (17) is a constant operator on the ambient Fock space, not an endomorphism of the valence line bundle. Eq. (31) makes this explicit: P_⊥ S P_⊥ = η P_⊥ with η = −⟨S⟩, which is generally not ±1. So ν_S = −2 Re(e^{iθ} J_F) is a continuous real function of θ and M. The jump of ν_S across the gap-closing stratum is not argued to be integer, and the phrase “quantized jumps” in Sec. III.B is unsupported by the paper’s own equations. The Hecke modification story could plausibly apply to a bundle that actually splits under S; here it does not.\n\nOther soft spots are minor by comparison: no code or data shipped, mesh sizes and reference phase unstated, so the figures cannot be independently regenerated. The conclusion’s claim about sheafification preserving within-patch structure is overreach from what is, at bottom, a single-particle two-band calculation.\n\nThe authors seem aware of the gap between the algebra and the conceptual overlay, but they do not flag the non-integrality of ν_S. That is the load-bearing problem. The fix is straightforward: rescope the Hecke/Langlands discussion as speculation, prove or explicitly qualify the jump integrality, and ship the numerics. Then the exact sector identities stand on their own.\n\nWho is this for? Condensed-matter theorists interested in Berry curvature diagnostics and QFI bounds. The Langlands layer will not convince that community and is not needed. I would send it to a serious referee, but with the expectation of major revision.","headline":"The witness-filtered sector identities are exact and new, but the Hecke/Langlands layer is unsupported: ν_S is a continuous, θ-dependent quantity, not the integer index that Eq. (10) would require.","tokens_in":15247,"tokens_out":1584,"would_cite":true,"duration_ms":26567,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A coherence filter exactly decomposes the Chern number into two sectors in a two-band Chern insulator.","keywords":["quantum entanglement","witness filter","Berry curvature","Chern number","Haldane model","quantum Fisher information","Hecke modification","stratified spaces"],"falsifier":"Compute the jump of ν_S across the gap-closing stratum for several witness phases θ; if the jump is not an integer independent of θ, the Hecke jump formula ΔInd_S = ⟨χ_S, λ⟩ fails for the graded response. Or, on an interacting version of the model, compute a genuine many-body entanglement measure (e.g., negativity of the reduced density matrix) and check whether its discontinuities coincide with the witness-filtered curvature response; a mismatch would indicate the single-particle mode entanglement is not a cohomological obstruction.","tokens_in":14059,"feed_emoji":"⚛️","tokens_out":7123,"duration_ms":70206,"temperature":0.7,"pith_summary":"The paper claims that in the canonical two-band Chern insulator (the spinless Haldane model), the Berry-curvature response can be split exactly, on a lattice-discretized Brillouin torus, into two sector responses governed by a witness-filtered coherence quantity. The split yields the closed identities ν± = μ/2 ± Re(e^{iθ} J_F), so the Chern number μ is recovered as ν+ + ν− and the witness-graded response ν_S = ν+ − ν− is a new gauge-invariant diagnostic of sublattice coherence. The same construction extends to multi-orbital embeddings, where the coherence becomes a matrix and is reconstructed by two phase settings. The paper interprets crossings of the gap-closing stratum as Hecke modifications, giving a Langlands-style description of the quantized jumps. If correct, this gives a direct, computable link between single-particle mode entanglement and topological response.","feed_headline":"Witness filter splits Chern response exactly","feed_subtitle":"In a two-band Chern insulator, filtered Berry curvature obeys exact identities and ties topology to mode entanglement.","key_machinery":"The central object is the witness sign operator S = sgn(W), a spectral splitting of the valence bundle that reduces the structure group to a Levi subgroup, together with the single-excitation embedding J that maps the occupied Bloch spinor |u_-(k)⟩ = v_A|A⟩+v_B|B⟩ into a two-mode Fock space as |Ψ_k⟩ = v_A|10⟩+v_B|01⟩. This embedding makes the product |v_A||v_B| the single-particle mode entanglement (concurrence) at each k. The identity doing the work is the exact lattice relation between the sector integrals ν± and the curvature-weighted coherence J_F = (1/2π)Σ F v_A v_B*, which shows that the witness filter isolates the coherence-carrying part of the Berry curvature. The Hecke jump formula","core_discovery":"On a standard plaquette discretization of the Brillouin torus for the spinless Haldane model, the witness-filtered sector responses satisfy exactly ν−(W)=μ/2+Re(e^{iθ}J_F^{AB}) and ν+(W)=μ/2−Re(e^{iθ}J_F^{AB}), with residuals at machine precision, where J_F is the curvature-weighted A/B coherence and θ is the witness phase. The graded response ν_S = ν+ − ν− equals the integrated trace of the witness sign operator S against Berry curvature, and the paper verifies these identities across the phase diagram. In the multi-orbital case the scalar J_F becomes a matrix x†J_F y, and two- or three-witness tomography recovers the full matrix. The paper also proves that the S-filtered quantum Fisher inf","pith_inferences":["A natural extension the authors gesture toward but do not carry out is replacing the pure-state Berry connection with the Uhlmann connection for mixed states; if the same identities survive, the witness filter would give a thermal entanglement diagnostic.","The paper's 'entanglement' is single-particle mode entanglement by construction; whether genuine many-body entanglement (e.g., in interacting or multi-particle states) tracks these filtered responses is an open question that could be tested with exact diagonalization or tensor-network methods.","Because the witness phase θ enters the jump of ν_S, the integer Hecke jumps asserted in Eq. (10) may depend on the choice of witness; checking whether the jump across Σ remains an integer independent of θ would either sharpen or challenge the Langlands interpretation.","The reconstruction of J_F could be implemented in cold-atom or photonic simulators of the Haldane model by measuring filtered Berry curvature via Bloch oscillations, making the coherence diagnostic experimentally testable."],"forward_implications":["The sector identities provide an exact, gauge-invariant decomposition of the topological response into coherence-carrying sectors, so μ = ν+ + ν− holds to machine precision on the lattice.","The S-filtered QFI never exceeds the conventional QFI; in the two-band case saturation occurs exactly at maximal single-particle entanglement, giving an operational test of entanglement via metrology.","Two witness phases fully reconstruct the curvature-weighted coherence J_F, and multi-orbital matrices are recovered by basis-probe scans.","Quantized jumps of the entanglement index across the gap-closing stratum are described by the Hecke jump formula ΔInd_S = ⟨χ_S, λ⟩, connecting topological phase transitions to Hecke modifications."],"fun_headline_variants":["Exact Chern split from witness-filtered curvature","Filtered geometry separates topology and coherence","Witness filter exposes exact topological response","Entanglement-sensitive splitting of Chern invariant","Hecke modifications appear in quantized jumps"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that modeling the occupied Bloch spinor at each momentum as a single-excitation two-mode Fock state |Ψ_k⟩ = v_A|10⟩+v_B|01⟩ makes single-particle mode entanglement, |v_A||v_B|, the correct notion of entanglement for the problem; if this identification is refused, the results reduce to a coherence filter on a 2×2 Bloch model without evidence about entanglement as a cohomological obstruction.","fun_headline_variants_meta":{"raw":{"variants":["Exact Chern split from witness-filtered curvature","Filtered geometry separates topology and coherence","Witness filter exposes exact topological response","Entanglement-sensitive splitting of Chern invariant","Hecke modifications appear in quantized jumps"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000233,"raw_usage":{"total_tokens":1314,"prompt_tokens":714,"completion_tokens":600,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":458,"completion_tokens_details":{"reasoning_tokens":536}},"tokens_in":458,"tokens_out":600,"duration_ms":6182,"temperature":1.0,"reasoning_tokens":536,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T09:33:53.207311+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the jump of ν_S across the gap-closing stratum for several witness phases θ; if the jump is not an integer independent of θ, the Hecke jump formula ΔInd_S = ⟨χ_S, λ⟩ fails for the graded response. Or, on an interacting version of the model, compute a genuine many-body entanglement measure (e.g., negativity of the reduced density matrix) and check whether its discontinuities coincide with the witness-filtered curvature response; a mismatch would indicate the single-particle mode entanglement is not a cohomological obstruction.","supporting_citations":[],"review_version":1}