{"id":"6c643a34-0160-4b05-878a-a424d7bddfe1","arxiv_id":"2412.04187","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Quenches from antipodally entangled crosscap states yield a delayed linear decrease and revivals of entanglement in integrable systems, while chaotic systems show constant entropy and a vanishing mutual information.","lead":"This paper studies how entanglement changes after a sudden quench from 'crosscap' states, in which opposite points of a periodic chain are entangled. It finds opposite behavior in integrable versus chaotic systems, and adapts the standard quasiparticle and membrane pictures to capture it.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The Hamiltonian results are derived for one Gaussian component of the spin crosscap state, not the crosscap state itself; no evidence shows the superposition has the same entanglement dynamics, so Eq. (3.18) may not describe the claimed quench.","rationale":"The paper's core Hamiltonian result is the quasiparticle prediction Eq. (3.18) for the entanglement entropy after a quench from a crosscap state. That formula is derived for the Gaussian state (3.3), which the paper itself identifies as only one component of the actual spin crosscap state (1.2). The reader's weakest-assumption analysis correctly locates this as the main gap. The concern is load-bearing because the abstract and conclusions make a claim about crosscap states broadly, and the Hamiltonian section is the quantitative basis for the integrable-system behaviour. Since the reduced density matrix of a superposition of two Gaussian states is not the same as the reduced density matrix of either component, and since entropy is nonlinear, there is no a priori reason for Eq. (3.18) to describe the true crosscap quench. The paper presents no test of the full superposition. The concrete numerical test proposed would settle whether the two components contribute equivalently for entanglement. The free-fermion numerics in the paper are internally consistent and convincing for the substituted state, so the correct verdict is CONDITIONAL rather than REJECT: the single-component calculation stands on its own, but its connection to the advertised crosscap state remains unverified. This matches the reader's verdict, so no change is needed.","tokens_in":27663,"tokens_out":3910,"duration_ms":41698,"concrete_test":"For a small chain (e.g. 2L = 20, 2ℓ = 6), construct the full spin crosscap state |C> = 2^{-L/2} ⊗_{x=1}^L (|0>_x|0>_{x+L} + |1>_x|1>_{x+L}), express it in the Jordan-Wigner fermionic basis as the explicit superposition of the two Gaussian components, and evolve under the free-fermion Hamiltonian H0 by exact diagonalization of the 2L-site chain. Compute SA(t) via the reduced density matrix and compare with Eq. (3.18) and with the single-component Gaussian calculation from Eqs. (3.4)-(3.5). If the full-state entropy agrees with Eq. (3.18), the substitution is benign; if it does not, the paper's Hamiltonian claim applies only to the Gaussian component, not to the crosscap state.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative prediction for Hamiltonian dynamics is Eq. (3.18), derived from the Gaussian state |C> defined in Eq. (3.3). The paper explicitly notes, just before Eq. (3.3), that a Jordan-Wigner transformation of the spin crosscap state |C> of Eq. (1.2) yields a superposition of two fermionic Gaussian states, and then states that 'we shall study the fermionic crosscap state defined directly', i.e. one component. This substitution is load-bearing because entanglement entropy is a nonlinear functional of the reduced density matrix: the entropy of a superposition is not the average of the entropies of the components, and the paper provides no argument or numerical evidence that the second component is harmless. All free-fermion checks in Sec. 3.3 (Fig. 7, and the charge-fluctuation checks in Fig. 5) compare Eq. (3.18) against exact numerics that also start from the Gaussian state (3.3), not from the spin crosscap state (1.2). Thus the excellent agreement tests the substituted state, not the claimed physical initial state. The interacting TBA extension in Sec. 3.4 inherits the same issue. If the full superposition has different entanglement dynamics, the abstract-level claim about quenches from crosscap states does not follow, even though the single-component free-fermion result may be correct in isolation.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the quench dynamics of the bipartite Rényi entropy S_A^(n)(t) and the mutual information I_{A:A_M}(t) from initial states with long-range correlations (crosscap states) in two classes of systems. For brickwork circuits, the authors analyze swap gates, random unitary circuits, and dual unitary circuits, finding that integrable (swap) circuits show delayed linear decrease and revivals of S_A, while chaotic circuits keep S_A constant and show a linear decrease of I_{A:A_M} that permanently vanishes. For Hamiltonian dynamics, they derive a free-fermion quasiparticle prediction, Eq. (3.18), which is checked against exact numerics, and extend it to interacting integrable chains via the thermodynamic Bethe ansatz. The results are interpreted through a modified quasiparticle/membrane picture.","tokens_in":27900,"tokens_out":14187,"duration_ms":116420,"significance":"The paper extends the quasiparticle and membrane pictures to initial states with long-range entanglement and maximal initial entropy, a scenario opposite to the standard low-entanglement quench. The main strengths are the exact free-fermion formulas (3.13), (3.18), and (3.21), which are derived from the correlation matrix and verified numerically (Figs. 5, 7, 8), and the explicit solvable circuit results. If the identified caveats are resolved, the paper would provide a useful reference for entanglement dynamics from crosscap-type states, including falsifiable predictions for quantum simulators.","major_comments":[{"comment":"The Hamiltonian free-fermion results are derived for the Gaussian state |C⟩ of Eq. (3.3), which the authors note is one component of the Jordan-Wigner image of the spin crosscap state (1.2); all numerical checks in Figs. 5, 7, and 8 start from this Gaussian state rather than the actual spin crosscap state. Since the entanglement entropy is not additive over a superposition, the central claim about quenches from crosscap states under Hamiltonian dynamics is not established; the authors should prove that the second Gaussian component does not affect S_A(t) and I_{A:A_M}(t) or provide numerical evidence for the full spin state.","section":"Section 3, before Eq. (3.3)"},{"comment":"The stationary-phase derivation in Appendix A is algebraically inconsistent with the main text. Using Eq. (3.12) and the relation ⟨N_A^2⟩ = σ_A^2 + ℓ^2, one obtains ⟨N_A^2⟩ = ℓ^2 + ℓ/2 − (1/2) Σ ..., whereas Eq. (A.2) has ℓ^2 − ℓ/2 − ℓ^2 ∫ ... . The final expression (A.13) also does not reduce to Eq. (3.13): the counting function in (A.13), after multiplying by ℓ, is max(0, ℓ − |2 τ_k v(k) − L/2|), while Eq. (3.13) contains max(0, 2ℓ − |2 τ_k v(k) − L|). These factors of two and the sign of the ℓ/2 term need to be corrected, since the derivation as written does not reproduce the numerically verified result (3.13).","section":"Appendix A, Eqs. (A.2), (A.13)"}],"minor_comments":[{"comment":"The condition '2ℓ ≤ mt > 0' is likely a typo; it should probably read '0 < mt ≤ 2ℓ'. Also, the explanation of the parity-dependent minimum at t_min would benefit from an explicit example.","section":"Sec. 2.1, Eq. (2.30)"},{"comment":"The interacting TBA results (3.23) and the curves in Figs. 9 and 10 are not compared to any independent numerical simulation; a small-system exact diagonalization or TEBD benchmark would support the extension to interacting integrable models.","section":"Sec. 3.4"},{"comment":"The statement that the annealed average result implies the constant entropy 'for all realizations' is terse; since ⟨tr ρ^2⟩ = q^{−2ℓ} and tr ρ^2 ≥ q^{−2ℓ}, the argument is valid, but it would benefit from stating this inequality explicitly.","section":"Sec. 2.2"},{"comment":"The variables τ_k and τ'_k are used in Eqs. (3.16) and (3.18) before their formal definitions in (3.14) and (3.17); consider defining them earlier to improve readability.","section":"Sec. 3.2"}],"recommendation":"major_revision","confidential_remarks":"The paper is well-written and the topic is timely. The main concern is the mismatch between the claimed initial state (the spin crosscap state) and the one actually used in the Hamiltonian calculations (a single Gaussian component). This needs to be resolved either by additional evidence or by reframing the claims. The appendix derivation also needs to be corrected."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know before you read it. First, the free-fermion quasiparticle calculation is solid and the agreement with exact numerics is genuine. Second, the Hamiltonian section does not actually study the spin crosscap state: after noting that its Jordan-Wigner image is a superposition of two Gaussian states, the authors switch to a single Gaussian component and never come back. The title and abstract promise crosscap-state quenches; Eq. (3.18) is derived and tested only for that component.\n\nWhat the paper does well. The crosscap state was known as an integrable boundary condition and through Bethe overlaps, but not as a quench initial state. The circuit section works with the actual qudit crosscap state and gets clean results: constant entropy under random and dual-unitary circuits, linear mutual information decay, and the swap-gate even-odd effect. The membrane-picture argument for constant entropy is plausible. The free-fermion section is the strongest part: the modified counting functions with the L-shift and periodic tau_k are non-obvious, and Figures 5, 7, 8 match exact numerics well, including for small subsystems. That is a real contribution.\n\nWhere it is soft. The Gaussian-substitution issue is load-bearing, not cosmetic. Entanglement entropy is nonlinear in the state, so the entropy of the superposition need not equal that of one component, and the paper gives no argument—numerical or analytic—that the second component is harmless. Since all free-fermion checks start from the same Gaussian state, they do not test the claimed initial state. The interacting TBA result of Eq. (3.23) inherits the same issue and is, in addition, an unverified ansatz: no exact numerics are shown for U != 0. Appendix A also has sign and coefficient mismatches with Eq. (3.12); the final result matches numerics, so probably typos, but they should be fixed. Finally, 'all integrable systems' overstates the evidence, which is one interacting model studied through an extrapolated quasiparticle picture.\n\nVerdict: the free-fermion and circuit results deserve publication, but the Hamiltonian claims need to be either proved for the superposition or explicitly restricted. This is refereeable, not desk-rejectable. I would send it out with a request to resolve the Gaussian issue. Reading group: maybe—worth seeing the modified QPP once, but the caveat makes it hard to take the abstract at face value.","headline":"The free-fermion core is solid and the circuit results are genuinely new, but the Hamiltonian section substitutes a single Gaussian component for the advertised crosscap state, leaving the central quantitative claim about crosscap-state quenches unsupported.","tokens_in":28471,"tokens_out":4367,"would_cite":true,"duration_ms":44791,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81P40","82B23","82B10"],"pacs":["05.30.-d","03.67.Mn","05.45.Mt"],"model":"deepseek-v4-flash","headline":"Quenches from crosscap states give opposite entanglement signatures in integrable and chaotic systems.","keywords":["crosscap states","quench dynamics","entanglement entropy","mutual information","quasiparticle picture","entanglement membrane","integrable systems","XXZ model"],"falsifier":"Evolve the exact spin-chain crosscap state $|C\\rangle=2^{-L/2}\\bigotimes_{x=1}^{L}(|0\\rangle_x|0\\rangle_{x+L}+|1\\rangle_x|1\\rangle_{x+L})$ under the free-fermion Hamiltonian without replacing it by a single Gaussian component, and compare the computed block entropy with Eq. (3.18) at sizes such as $2L=400$, $2\\ell=140$; any deviation beyond the stationary-phase error would falsify the claim that the Gaussian-component dynamics describes the crosscap quench.","tokens_in":27410,"feed_emoji":"🔗","tokens_out":11970,"duration_ms":105687,"temperature":0.7,"pith_summary":"This paper asks what happens to entanglement after a quench from a state with long-range correlations, specifically crosscap states in which antipodal points of a periodic chain are prepared in maximally entangled pairs. It establishes that the answer divides cleanly by integrability: in integrable systems the block entanglement stays maximal for a delay time, then falls linearly and revives periodically, while in chaotic systems it stays constant; the mutual information between a block and its antipodal mirror falls linearly in both and revives only in integrable systems. The paper derives a quantitative quasiparticle formula for the free-fermion entropy, checks it against exact numerics, and extends the picture to the interacting integrable chain using thermodynamic Bethe ansatz data, while circuit results are explained through the entanglement membrane picture. If the paper is right, crosscap states turn entanglement dynamics into a sharp integrability detector and show how long-range initial correlations alter the standard growth rules.","feed_headline":"Crosscap-state quench entropy: integrable dips, chaotic stays flat","feed_subtitle":"Maximally entangled antipodal pairs delay, then revive entanglement only in integrable systems.","key_machinery":"The central object is the crosscap state $|C\\rangle$, a translation-invariant state of $2L$ qudits in which each site $x$ is maximally entangled with its antipode $x+L$, so that any local block looks like the infinite-temperature state while global correlations are maximal. The argument runs through two effective descriptions modified for long-range initial correlations: the quasiparticle picture, where the counting function $\\max(0,2\\ell-|2\\tau_k v(k)-L|)$ with folded time $\\tau_k=t\\bmod L/|v(k)|$ counts antipodally correlated pairs entering a block only after a delay, and the entanglement membrane picture, where the maximally entangled initial state makes membrane termination on the lower boundary costly, with the cost cancelled exactly for membranes ending at diametrically opposite points. In the interacting integrable model these inputs are replaced by species-resolved Bethe-ansatz data: each bound-state species $m$ contributes its own counting function with velocity $v_m(\\lambda)$ and entropy weight $s_m$.","core_discovery":"The paper's central claim is that a quench from a crosscap state produces two qualitatively different entanglement histories depending on whether the dynamics is integrable or chaotic. In integrable systems the Rényi entanglement entropy $S_A^{(n)}(t)$ of a contiguous block of length $2\\ell$ remains pinned at its maximal value $2\\ell\\log q$ until the delay time $(L-2\\ell)/4$, then decreases linearly in time and goes through periodic revivals; in chaotic systems it remains constant at the maximal value. The mutual information $I_{A:A_M}(t)$ between the block and its mirror-image block behaves as a complementary probe: it decreases linearly from the start in both classes, vanishes permanently under chaotic dynamics, and undergoes revivals under integrable dynamics. For free fermions the paper derives the explicit prediction $$S_A(t)=2\\ell\\log 2-2\\log2\\int_{-\\pi}^{\\pi}\\frac{dk}{2\\pi}\\max\\left(0,2\\ell-|2\\tau_k v(k)-L|\\right),\\qquad \\tau_k=t \\bmod \\frac{L}{|v(k)|},$$ and verifies it against exact numerics, then generalizes the counting-function structure to interacting integrable models through a sum over Bethe-ansatz quasiparticle species.","pith_inferences":["Editorial inference: the membrane-picture mechanism implies that in chaotic systems the antipodal mutual information is a sharper scrambling diagnostic than single-block entropy, because entropy is frozen at its maximal value while mutual information decays and stays zero; the paper does not frame it this way.","Editorial inference: because the spin-chain crosscap state is a superposition of two Gaussian components, interference between them could alter the free-fermion prediction; testing the full state numerically is a direct way to see whether the discarded component is genuinely harmless.","Editorial inference: the delay time $(L-2\\ell)/4$ and revival period $L/|v(k)|$ are concrete signatures that could be sought in quantum simulators that prepare antipodal Bell pairs, and the revival pattern would serve as an experimental integrability marker.","Editorial inference: generalizing the construction to entanglement between more than two distant sites would replace pair counting with multiplet counting and likely produce multiple delay times rather than one; the paper lists this as a natural next step."],"forward_implications":["For integrable dynamics, $S_A^{(n)}(t)$ stays at its maximal value $2\\ell\\log q$ until $t=(L-2\\ell)/4$, then decreases linearly and revives periodically; for chaotic circuits it remains constant.","The mutual information between a block and its antipodal mirror decreases linearly at early times in both classes, vanishes permanently under chaotic dynamics, and revives under integrable dynamics.","The free-fermion entropy is fixed quantitatively by Eq. (3.18), and the same quasiparticle counting structure, summed over Bethe-ansatz species, controls the interacting integrable chain.","Time-averaged entanglement from these states has a Page-curve-like dependence on subsystem size, given by Eq. (3.20).","The modified membrane picture predicts constant entropy and vanishing mutual information for chaotic circuits at large local Hilbert space dimension."],"supporting_citations":[{"why":"supplies the correlation-matrix method used to compute exact free-fermion entanglement entropies.","marker":"[86]"},{"why":"provides the standard quasiparticle picture of entanglement growth that the paper modifies for long-range initial correlations.","marker":"[11]"},{"why":"extends the quasiparticle picture to generic integrable models, the template for the interacting case here.","marker":"[12]"},{"why":"supplies the thermodynamic Bethe ansatz formalism and occupation functions used for the interacting chain.","marker":"[84]"},{"why":"establishes crosscap states as integrable initial states, giving the overlap data used in the Bethe ansatz treatment.","marker":"[52]"},{"why":"provides the solvable-state and dual-unitary circuit rules that underpin the diagrammatic reductions.","marker":"[64]"},{"why":"supplies the entanglement membrane picture used to interpret the chaotic circuit results.","marker":"[33]"},{"why":"gives the random-unitary entanglement growth baseline against which the flat entropy result is contrasted.","marker":"[55]"},{"why":"provides the stationary-phase and disjoint-interval techniques used to derive the quasiparticle formula.","marker":"[85]"},{"why":"gives exact disjoint-interval entanglement results in dual-unitary circuits, used for the mutual information calculation.","marker":"[56]"}],"fun_headline_variants":["Crosscap quench: integrable entropy dips and revives, chaotic stays constant","Crosscap-state quench: integrable revivals, chaotic constant entropy","Mutual info from crosscap quench: revivals in integrable, vanish in chaotic","Entanglement quench from crosscap states: integrable seesaw, chaotic flatline","Crosscap-state quench: entropy flat in chaos, revives under integrable dynamics"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that the full spin-chain crosscap state, which after the Jordan-Wigner transformation is a superposition of two free-fermion components, behaves like the single one of those components that the paper actually evolves; if the discarded component changes the entanglement, the paper's quantitative spin-chain predictions do not follow.","fun_headline_variants_meta":{"raw":{"variants":["Crosscap quench: integrable entropy dips and revives, chaotic stays constant","Crosscap-state quench: integrable revivals, chaotic constant entropy","Mutual info from crosscap quench: revivals in integrable, vanish in chaotic","Entanglement quench from crosscap states: integrable seesaw, chaotic flatline","Crosscap-state quench: entropy flat in chaos, revives under integrable dynamics"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000596,"raw_usage":{"total_tokens":2859,"prompt_tokens":1082,"completion_tokens":1777,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":698,"completion_tokens_details":{"reasoning_tokens":1669}},"tokens_in":698,"tokens_out":1777,"duration_ms":13127,"temperature":1.0,"reasoning_tokens":1669,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T21:40:40.603786+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Evolve the exact spin-chain crosscap state $|C\\rangle=2^{-L/2}\\bigotimes_{x=1}^{L}(|0\\rangle_x|0\\rangle_{x+L}+|1\\rangle_x|1\\rangle_{x+L})$ under the free-fermion Hamiltonian without replacing it by a single Gaussian component, and compare the computed block entropy with Eq. (3.18) at sizes such as $2L=400$, $2\\ell=140$; any deviation beyond the stationary-phase error would falsify the claim that the Gaussian-component dynamics describes the crosscap quench.","supporting_citations":[{"cited_title":"Peschel, Calculation of reduced density matrices from correlation functions, J","cited_arxiv_id":null,"evidence_quote":"supplies the correlation-matrix method used to compute exact free-fermion entanglement entropies."},{"cited_title":"Takahashi, Thermodynamics of One-Dimensional Solvable Models","cited_arxiv_id":null,"evidence_quote":"supplies the thermodynamic Bethe ansatz formalism and occupation functions used for the interacting chain."},{"cited_title":"He and Y","cited_arxiv_id":null,"evidence_quote":"establishes crosscap states as integrable initial states, giving the overlap data used in the Bethe ansatz treatment."},{"cited_title":"Piroli, B","cited_arxiv_id":null,"evidence_quote":"provides the solvable-state and dual-unitary circuit rules that underpin the diagrammatic reductions."},{"cited_title":"Zhou and A","cited_arxiv_id":null,"evidence_quote":"supplies the entanglement membrane picture used to interpret the chaotic circuit results."},{"cited_title":"Nahum, J","cited_arxiv_id":null,"evidence_quote":"gives the random-unitary entanglement growth baseline against which the flat entropy result is contrasted."},{"cited_title":"Fagotti and P","cited_arxiv_id":null,"evidence_quote":"provides the stationary-phase and disjoint-interval techniques used to derive the quasiparticle formula."},{"cited_title":"Entanglement of Disjoint Intervals in Dual-Unitary Circuits: Exact Results","cited_arxiv_id":"2408.16750","evidence_quote":"gives exact disjoint-interval entanglement results in dual-unitary circuits, used for the mutual information calculation."}],"review_version":1}