{"id":"5a0a6208-c653-4516-aaf8-81aa4ed8c6ad","arxiv_id":"2509.09640","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For sudden quenches in translation-invariant quadratic fermionic chains, the work distribution maps to a linear statistic of Haar-distributed unitary traces and is Gaussian in the large-system limit with variance 1/2 Σ r(a_r^2+b_r^2).","lead":"This paper connects the work distribution of sudden quantum quenches to the statistics of random unitary matrices, showing that it becomes Gaussian in the thermodynamic limit with a variance set by the post-quench Hamiltonian's Fourier coefficients. It also identifies mechanisms, such as long-range interactions and Fisher-Hartwig singularities, that produce non-Gaussian tails.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The physical Loschmidt amplitude is a product over Brillouin-zone modes, not the Toeplitz/CUE average; Theorem 1's premise fails for generic PBC translation-invariant initial states.","rationale":"The reader identified the Toeplitz representation as the weakest assumption, and I agree that it is the load-bearing point. My concern is sharper: the representation is not merely underived; for generic translation-invariant Slater/BCS initial states under PBC it is false. The physical Loschmidt amplitude is a deterministic product over occupied momentum modes (or over k>0 in the BdG case), whereas D_N(e^{-it\\varepsilon}) is a Toeplitz determinant equal to a Haar expectation over random eigenvalues with a Vandermonde weight. These are different objects already at finite N: the Toeplitz determinant has Szegő corrections giving an O(1) Gaussian variance, while the deterministic product has no such fluctuation. A simple N=2 example with symbol e^{-it\\cos\\theta} shows G(t)=1 versus D_2=J_0(t)^2+J_1(t)^2. Therefore the central claim that sudden-quench work statistics reduce to Haar trace statistics does not hold for the stated class of translation-invariant PBC quenches. The paper's mathematical results on traces of Haar unitaries and the numerical diagnostics of the abstract linear statistic may be correct, but they do not validate the physical quench model. The verdict should be rejection of the central claim as stated, or at minimum a major rewrite that narrows the scope to initial states for which the Toeplitz/circulant distinction is handled correctly (e.g., domain-wall states) and re-derives the reduction. I recommend REJECT because the simple counterexample is decisive and not a matter of missing estimates.","tokens_in":17260,"tokens_out":14015,"duration_ms":177382,"concrete_test":"Take L=4 (N=4) with H_i=-sum_j(c^\\dagger_{j+1}c_j+h.c.) and ground state filling the two modes with \\varepsilon_i(k)<0; take H_f=J sum_j(c^\\dagger_{j+1}c_j+h.c.). Compute the exact Loschmidt amplitude G_ex(t)=prod_{k in F} e^{-it\\varepsilon_f(k)} and the exact TPM work distribution by diagonalizing e^{-iH_f t}|\\psi_0>. Also compute D_4(e^{-it\\varepsilon_f}) from its Fourier coefficients c_n=(1/2\\pi)\\int e^{-it 2J\\cos\\theta} e^{-in\\theta}d\\theta. If |G_ex(t)| neq |D_4(e^{-it\\varepsilon_f})| for t neq 0, or if Var_TPM(W) neq 2 sum_r r J_r^2, the central reduction fails. Repeat for the XY chain with L=4 using (57) versus the claimed block-Toeplitz determinant; verify whether |G(t)|=1 in the \\gamma=0 limit.","verdict_should_be":"REJECT","load_bearing_attack":"Theorem 1 and Corollary 1 rest on the assertion (Sec. II) that for a translation-invariant Slater/BCS eigenstate under PBC, G_N(t)=D_N(e^{-it\\varepsilon}). For a number-conserving quench, H_i and H_f are both diagonal in momentum, so an eigenstate of H_i has occupations n_k in {0,1}; the exact amplitude is G(t)=prod_{k:n_k=1} e^{-it\\varepsilon_f(k)}, a deterministic product over the occupied set, not the CUE average D_N(e^{-it\\varepsilon}). Even in the full-filling case n_k=1, G(t)=e^{-it\\sum_k \\varepsilon_f(k)}, while D_N(e^{-it\\varepsilon}) contains Szegő fluctuation corrections. For example, take N=2 and \\varepsilon(\\theta)=\\cos\\theta. Then G(t)=e^{-it(1-1)}=1, whereas D_2(e^{-it\\cos\\theta})=J_0(t)^2+J_1(t)^2, which is not identically 1. Thus |G(t)| differs from |D_N(e^{-it\\varepsilon})|, and the claimed equality P(W)=Law(S_N(U)-E_0) is not established; in the XX limit P(W) is a delta, not Gaussian. The XY product formula (57) is also a product over k>0, not a Toeplitz average. The Toeplitz determinant representation is more plausibly valid for domain-wall Slater states (Toeplitz minors), so the theorem's scope is narrower than stated.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims that for sudden quenches in translation-invariant quadratic fermionic chains with periodic boundary conditions, the work characteristic function is the characteristic function of a linear statistic of traces of Haar-distributed unitary matrices. The argument proceeds through a Toeplitz determinant representation of the Loschmidt amplitude, an exact Heine–Szegő/Toeplitz identity, and multivariate central limit theorems for traces of Haar unitaries. The paper further claims a Gaussian core for the work distribution with variance Var(W) = (1/2) Σ_r r(a_r² + b_r²), identifies non-Gaussian deviations (many harmonics, slow Fourier decay, Fisher–Hartwig singularities), and illustrates the mechanism with the XY chain and with numerical diagnostics based on Haar-distributed unitary traces.","tokens_in":17631,"tokens_out":11103,"duration_ms":126405,"significance":"If the central mapping were correct, the paper would provide an elegant and potentially powerful bridge between sudden-quench work statistics and random matrix theory, with sharp variance formulas and a clear criterion for Gaussianity. The paper correctly reviews the relevant CLTs for traces of Haar unitaries, and Lemma 1 is a correct mathematical identity. However, the load-bearing physical premise—that translation-invariant Slater/BCS eigenstates under PBC produce a Toeplitz determinant with symbol e^{-itε(θ)}—is false. The exact Loschmidt amplitude in such settings is a deterministic product over momentum modes, not a Toeplitz determinant of the Heine–Szegő type. This is not a missing proof but an incorrect claim, with an explicit finite-N counterexample. The XY-chain section itself uses a product formula that is not a Toeplitz determinant, and the numerical diagnostics simulate random matrix traces rather than physical quenches. The advertised physical conclusion is therefore not established.","major_comments":[{"comment":"The asserted Toeplitz representation is false for the stated physical initial states. For a number-conserving translation-invariant quench, H_i and H_f are both diagonal in momentum, so an eigenstate of H_i has occupation numbers n_k ∈ {0,1} and the exact amplitude is G(t) = ∏_{k: n_k=1} e^{-it ε_f(k)}, a deterministic product. This is a determinant of a circulant matrix, not the Toeplitz determinant D_N(e^{-itε}) used in the Heine–Szegő identity. Concrete counterexample: N=2, ε(θ)=cosθ, full filling: G(t)=e^{-it(cos0+cosπ)}=1, whereas D_2(e^{-it cosθ}) = J_0(t)² + J_1(t)², which is not identically 1. Hence Eq. (24) and Theorem 1(i) do not hold for translation-invariant Slater/BCS eigenstates, and the equality P(W) = Law(S_N(U) − E_0) is not established. The Toeplitz-minor representation mentioned for domain-wall states is a different and narrower setting.","section":"Sec. II and Theorem 1"},{"comment":"The XY-chain exact amplitude is the product over k>0 in Eq. (57), not a Toeplitz determinant. The passage claiming 'G(t)=√det T_N(Φ)=D_N(f_t)^{1/2}' is unjustified: for a block Toeplitz operator, det T_N(Φ) is not generally equal to ∏_{k>0} det Φ(k), and the physical G(t) is the latter product. Consequently the linear-statistics form in Eq. (65) does not follow; log G(−u) is a Riemann sum over k, and the O(u²) error in Eq. (63) is uncontrolled and can affect the variance and higher cumulants. Formula (66) is therefore not a derived consequence for the physical XY quench.","section":"Sec. V.C.1, Eqs. (57)–(63)"},{"comment":"The numerical diagnostics simulate Haar-distributed unitary traces, or Gaussian surrogates for them, not the work distribution of any of the advertised quenches. They verify the known RMT CLT for traces, not the mapping from Hamiltonians to traces. Without a numerical implementation of, e.g., the XY product formula (57), the plots do not support the paper's physical Gaussian-core claim.","section":"Sec. VI"}],"minor_comments":[{"comment":"The statement contains a duplicated sentence: 'Let |ψ0⟩ be an eigenstate of Hi with Hi|ψ0⟩ = E0|ψ0⟩. |ψ0⟩ is an eigenstate of Hi with Hi|ψ0⟩ = E0|ψ0⟩.'","section":"Theorem 1"},{"comment":"The title 'FOURIER TRANSFORM OF LOCSHMIDT AMPLITUDE' misspells 'Loschmidt'.","section":"Sec. III title"},{"comment":"The notation alternates between L and N for the system size; this should be fixed, since the Toeplitz dimension N is also the unitary-group dimension.","section":"Sec. II"},{"comment":"The text should explicitly clarify that D_N(f_t) in the Toeplitz convention is not the momentum-space product; otherwise Eq. (57) and the claimed equality G(t)=D_N(f_t)^{1/2} are contradictory.","section":"Eqs. (57) and (60)"},{"comment":"The phrase 'representation-invariant' is undefined and should be removed or explained.","section":"Sec. V.A"}],"recommendation":"reject","confidential_remarks":"The counterexample in Major Comment 1 is decisive: the central mapping is false for the physical setup advertised in the abstract. The domain-wall case is a different and much narrower claim, and the XY product-formula section would need a completely different derivation. I do not see how a local revision can repair the paper's central assertion within its current scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear X,\n\nThe paper has a real gap at the load-bearing point, and it is not a minor one. The author correctly assembles the multivariate CLT for traces of Haar unitaries and derives a clean variance formula Var(W)=1/2 Σ r(a_r^2+b_r^2), with a sensible taxonomy of non-Gaussian mechanisms. The numerical work is careful and the summary of Johansson–Lambert bounds is useful. If the physical premise were right, this would be a nice unification.\n\nBut the premise is not right for the systems the paper claims. For a translation-invariant quadratic chain with PBC and a Slater or BCS initial eigenstate, the Loschmidt amplitude is a deterministic product over occupied modes: G(t)=∏_k e^{-it ε_f(k)} (or the standard XY product over k>0). It is not a Toeplitz determinant D_N(e^{-it ε}) in the sense of an average over Haar-distributed eigenvalues. A concrete check: N=2, ε(θ)=cosθ, full filling gives G(t)=1, while D_2(e^{-it cosθ}) is not identically 1. The paper's own Eq. (57) for the XY chain is a product over k>0, not a Toeplitz average; the subsequent identification of a 'scalar symbol' controlling sqrt(det T_N(Φ)) appears to be a reparametrization of the same product, not an equality of determinants. So Theorem 1's conclusion that P(W) is the law of S_N(U)-E_0 does not follow for the stated class.\n\nThe random-matrix model can be correct for other settings—domain-wall initial states give Toeplitz minors, and the trace-CLT statements hold for any linear statistic of CUE eigenangles. The paper would be on solid ground if it restricted its claims accordingly and dropped the PBC translation-invariant Slater/BCS story. As written, the central identification is an overreach.\n\nThe numerical diagnostics mostly validate the abstract trace CLT, not the physical quench model. The small-u expansion in Sec. V.d also has uncontrolled O(u^2) terms, though that is secondary to the main issue.\n\nMy take: this deserves a serious referee because the RMT content is mostly correct and the idea of connecting quench work to trace statistics is worth pursuing. But the present version conflates a circulant/product structure with a Toeplitz average, and the theorems should not be trusted until that is fixed. I would condition acceptance on a substantial revision: either correct the physical scope or show explicitly when a Toeplitz representation does hold.\n\nBest.","headline":"The RMT trace-CLT part is solid, but the paper misidentifies the physical Loschmidt amplitude as a Toeplitz determinant; the central theorem fails for generic PBC translation-invariant quenches.","tokens_in":18088,"tokens_out":4027,"would_cite":false,"duration_ms":46192,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60B20","82B10","60F05"],"pacs":[],"model":"deepseek-v4-flash","headline":"The work distribution of a sudden quench in a translation-invariant quadratic fermionic chain is the law of a linear statistic of Haar-distributed unitary traces, becoming Gaussian in the large-system limit with a variance fixed by the post","keywords":["sudden quantum quench","work distribution","Loschmidt amplitude","Toeplitz determinant","random matrix theory","Haar unitary traces","central limit theorem","Fisher-Hartwig singularities"],"falsifier":"Compute, for a finite-length XY chain (or any quadratic chain) with a BCS ground-state initial state, the exact Loschmidt amplitude from the product formula over Bogoliubov angles, and compare its logarithm to the logarithm of the Toeplitz determinant D_N(e^{-itε_f}) at fixed N and moderate t. If the two disagree beyond O(t^2) terms, the exact Toeplitz representation assumed for paired initial states is false. Alternatively, measure (or compute) the excess kurtosis of the work distribution for a fixed, small set of harmonics and verify it tends to 0 as N→∞; if it does not vanish, the Gaussian-","tokens_in":17113,"feed_emoji":"⚛️","tokens_out":4898,"duration_ms":51303,"temperature":0.7,"pith_summary":"The paper establishes that for sudden quenches in translation-invariant quadratic fermionic chains with periodic boundary conditions, the full work distribution is governed by the statistics of traces of Haar-distributed random unitary matrices: the work characteristic function is exactly the characteristic function of a linear statistic S_N(U)=Σ_{r≤m}(a_r Re Tr U^r + b_r Im Tr U^r), shifted by the initial energy. Because traces of different powers become independent complex Gaussians with variance r in the large-N limit, the work distribution acquires a Gaussian core with variance (1/2)Σ r(a_r^2+b_r^2), set entirely by the Fourier coefficients of the post-quench dispersion. The paper also identifies the mechanisms that produce non-Gaussian tails: many active harmonics, slow Fourier decay, Fisher–Hartwig singularities in the Toeplitz symbol, and Fisher zeros near the real time axis from critical quenches. The result makes precise the general expectation that sudden-quench work statistics are generically Gaussian, while showing exactly where and why deviations appear.","feed_headline":"Sudden quench work becomes Gaussian via random-matrix law","feed_subtitle":"For translation-invariant fermion chains, the work variable equals a linear statistic of Haar unitary traces, whose central limit theorem se","key_machinery":"The key identity is the Heine–Szegő (Toeplitz/unitary matrix-model) identity, which equates the Toeplitz determinant D_N(e^{iuε(θ)}) with the expectation over Haar-distributed U of exp(iuΣ_r(a_r Re Tr U^r + b_r Im Tr U^r)). This turns the Loschmidt amplitude into the moment-generating function of a linear statistic of the eigenangles of a random unitary, and the work law into a pushforward of the Haar measure. The second pillar is the classical multivariate central limit theorem for traces of powers of Haar-distributed unitaries: the normalized traces Tr U^r/√r are asymptotically independent standard complex Gaussians for fixed r as N→∞, with quantitative extensions allowing the number of ha","core_discovery":"The central claim is that, for a sudden quench from an eigenstate of the initial Hamiltonian, when the Loschmidt amplitude admits a Toeplitz determinant representation with symbol e^{iuε(θ)}, the work random variable is distributionally equal to S_N(U)−E_0, where U is Haar-distributed on U(N) and ε(θ) is the post-quench single-particle dispersion. By the multivariate central limit theorem for traces of Haar unitaries, as N→∞ the work distribution converges to a Gaussian with variance (1/2)Σ_{r≥1} r(a_r^2+b_r^2) whenever the Szegő regularity condition Σ r(a_r^2+b_r^2)<∞ holds; deviations are controlled by the number of harmonics, the decay of the Fourier coefficients, and possible Fisher–Hart","pith_inferences":["If the Toeplitz representation extends to open boundary conditions or to non-translation-invariant initial states, the same trace-statistics mechanism should produce a similar Gaussian core with boundary-induced Fisher–Hartwig corrections; the paper notes the open-boundary case as future work, and the machinery suggests explicit Painlevé-type tail asymptotics.","The identification P(W)=law of S_N(U)−E_0 suggests that the large-deviation properties of W could be probed through known phase transitions in unitary matrix models (e.g., Gross–Witten–Wadia-type transitions), giving universal tail reorganization without changing the bulk Gaussian behavior.","Since each harmonic contributes two independent Gaussian quadratures (Re and Im), the leading-order variance depends only on the magnitudes |a_r| and |b_r|, not their phases; phases would only enter at subleading, non-Gaussian order, a testable prediction for anisotropic chains.","The small-u expansion used in the XY-chain reduction implies that for long times (large u) the Gaussian core may break down even within the formally Gaussian regime; studying the large-u asymptotics of the characteristic function through Toeplitz/Painlevé asymptotics could probe the full tail structure experimentally."],"forward_implications":["For any finite-range, translation-invariant quadratic chain under periodic boundary conditions, the work distribution from a sudden quench is asymptotically Gaussian with variance given by the weighted Fourier sum; this gives a closed-form prediction for two-projective-measurement work experiments.","The Gaussian core is robust: it holds for exponentially decaying Fourier coefficients (gapped models) and even for mild power-law decay satisfying the Szegő condition; deviations are quantitatively controlled by the growth of the number of harmonics.","Non-Gaussian tails appear in identifiable, model-specific regimes: quenches with slowly decaying (long-range) couplings, at critical points where Fisher–Hartwig singularities or Fisher zeros develop, and at finite N where higher cumulants are visible; each mechanism leaves a distinct signature in histograms and Q–Q plots.","Because the Gaussian variance is additive over neighbor ranges and pairing terms, tuning the coefficients {a_r,b_r} (by changing interaction range or anisotropy) can engineer or suppress the Gaussian core and tails.","The random-matrix normalization (variance of Tr U^r of order r, not N) means the work variance is O(1) rather than O(N): the thermodynamic limit enters through the unitary group dimension, not through summing N i.i.d. variables."],"fun_headline_variants":["Sudden quench work turns Gaussian via Haar unitary traces","Quench work statistics: Gaussian via random matrix law","Random matrices make quench work Gaussian in fermion chains","Work distribution becomes Gaussian in quenches: random matrix theory","Haar unitaries yield Gaussian work for sudden quenches"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing assumption is that the Loschmidt amplitude G(t)=⟨ψ0|e^{-iH_f t}|ψ0⟩ is exactly equal to a Toeplitz (or block-Toeplitz) determinant whose symbol is e^{-itε(θ)} (extended by Bogoliubov factors when pairing is present); if this representation fails for a given initial state or boundary condition, the entire reduction of work to a linear statistic of Haar-unitary traces collapses.","fun_headline_variants_meta":{"raw":{"variants":["Sudden quench work turns Gaussian via Haar unitary traces","Quench work statistics: Gaussian via random matrix law","Random matrices make quench work Gaussian in fermion chains","Work distribution becomes Gaussian in quenches: random matrix theory","Haar unitaries yield Gaussian work for sudden quenches"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000492,"raw_usage":{"total_tokens":2279,"prompt_tokens":796,"completion_tokens":1483,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":540,"completion_tokens_details":{"reasoning_tokens":1401}},"tokens_in":540,"tokens_out":1483,"duration_ms":11900,"temperature":1.0,"reasoning_tokens":1401,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-04T18:44:20.121789+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute, for a finite-length XY chain (or any quadratic chain) with a BCS ground-state initial state, the exact Loschmidt amplitude from the product formula over Bogoliubov angles, and compare its logarithm to the logarithm of the Toeplitz determinant D_N(e^{-itε_f}) at fixed N and moderate t. If the two disagree beyond O(t^2) terms, the exact Toeplitz representation assumed for paired initial states is false. Alternatively, measure (or compute) the excess kurtosis of the work distribution for a fixed, small set of harmonics and verify it tends to 0 as N→∞; if it does not vanish, the Gaussian-","supporting_citations":[],"review_version":1}