{"id":"ab0f1dc7-096e-4029-8670-e062c43760e9","arxiv_id":"2507.12526","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"Non-Gaussian doping of matchgate circuits restores generic entanglement growth in unitary evolution and, at extensive per-time injection rates, stabilizes a volume-law phase under measurements.","lead":"Adding non-Gaussian Clifford gates to matchgate circuits restores generic ballistic entanglement growth and Kardar-Parisi-Zhang fluctuations, while monitored versions show a transition from area-law to power-law entanglement. The paper quantifies how much non-Gaussianity is required to escape free-fermion behavior.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The monitored phase diagram is fit over N=64–512 with doping density q(N) that vanishes as N grows; a convergence check of α(β,N) at larger N is needed before the 'extensive per-layer rate required' claim is secure.","rationale":"The paper's central claim has two parts: in unitary circuits, an extensive total number of non-Gaussian gates restores ballistic growth and KPZ fluctuations; in monitored circuits, an extensive per-layer rate is needed for a volume-law phase, with a power-law phase S∼N^α otherwise. The unitary part is well supported by the analytic arc model and is not my main concern. The monitored part rests on the scaling of S_N/2 versus N in Fig. 4(b), where q=η/N^β. The key problem is that for fixed β>0 the model itself changes with N: q(N)→0, so the extrapolation to N→∞ is not at fixed parameters. The reported α(β) is a finite-size effective exponent over only a factor of 8 in N, and the authors provide no collapse or convergence evidence for β>0, unlike the β=0 case. Because the headline distinction (volume law only for β=0) depends on the N→∞ limit of α, a direct large-N check is the decisive test. If α for β=1 drifts upward at larger N, the minimal-rate claim is not correct; if it is stable, the concern is resolved. I agree with the reader that the Clifford restriction and the missing code/data are relevant caveats, but the finite-size convergence of α is a more immediate internal correctness risk and should be added to the conditions for acceptance.","tokens_in":15229,"tokens_out":19378,"duration_ms":257040,"concrete_test":"Run the paper's stabilizer simulation for p=0.01, η=1, β=1 and β=0.5 at N=1024, 2048, and 4096 if feasible, with at least 500 trajectories, and extract α by the same fitting procedure (power law plus logarithmic corrections) over the largest decade of N. Plot α(β,N) versus 1/N and test whether it is constant within error bars. If α for β=1 trends upward toward 1 as N increases, the 'extensive per-layer rate necessary' conclusion is a finite-size artifact; if α is stable for both β values, the concern is resolved and the phase diagram is supported.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Concern: the monitored phase diagram for β>0 is extracted from a sequence of circuits in which the non-Gaussian doping probability q=η/N^β depends on N. For fixed η=1 and β>0, q(N)→0 as N→∞, so the thermodynamic limit is not taken at fixed Hamiltonian parameters. Over the measurement correlation time τ~1/p≈100 layers, the expected number of non-Gaussian gates touching a given bond is τ q(N)→0, so locally the dynamics becomes Gaussian as N grows; the observed S∼N^α with α<1 must then arise from rare events whose statistics need not be converged at N=512. The exponent α(β) in Fig. 4(b) is inferred from N=64–512, only a factor of 8 in system size, and no convergence check or scaling collapse is shown for β>0; the reported collapse and ν≈1.3 are for β=0, where q is fixed. If α drifts with N—upward toward the volume-law value α=1 or downward toward the Gaussian log phase—the central claim that volume-law requires an extensive per-layer rate is not established. This concern is internal to the Clifford model and is therefore more immediate than the separate question of whether Clifford gates faithfully represent Haar non-Gaussian gates under monitoring.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies random matchgate (fermionic Gaussian) Clifford circuits doped with non-Gaussian Clifford gates, both without and with projective measurements. In the unitary case, it claims that injecting an extensive total number of non-Gaussian gates restores ballistic entanglement growth S(t) ~ t and KPZ-type fluctuations delta S ~ t^{1/3}, and that the late-time Page curve crosses over from the Gaussian-stabilizer curve to the generic stabilizer curve. In the monitored case, it reports a measurement-induced transition between an area-law phase and a power-law phase S ~ N^alpha with alpha controlled by the doping exponent beta, with a genuine volume-law phase only when the doping rate is extensive per unit time (beta=0). The paper provides an analytical arc-model description for the Gaussian unitary and monitored dynamics, including a master equation for the arc-length distribution. All doped numerical results are obtained from stabilizer-circuit simulations of Clifford gates for systems up to a few hundred qubits.","tokens_in":15479,"tokens_out":8593,"duration_ms":105149,"significance":"If correct, the paper identifies a sharp resource threshold: a total of O(N) non-Gaussian gates restores generic unitary entanglement dynamics, while an extensive per-layer injection rate is needed to stabilize a volume-law phase under monitoring. This would quantify the minimal non-Gaussianity needed to escape integrable fermionic behavior and would connect the fermionic magic resource theory to entanglement phase diagrams. The analytic arc-model derivations for the Gaussian case are clean and self-contained, and the exact Page-curve formula for Gaussian stabilizer states is a useful contribution. The numerical program is ambitious and the qualitative distinction between intensive and extensive doping is falsifiable. However, the monitored phase diagram rests on finite-size fits over a narrow range of N, and the Clifford restriction leaves open the question of whether the results apply to generic non-Gaussian (Haar) circuits.","major_comments":[{"comment":"The central monitored claim—that volume-law entanglement requires an extensive per-layer doping rate (beta=0)—is supported only by fitting S_N/2 versus N for N=64-512 at p=0.01. For beta>0, q=eta/N^beta tends to zero as N grows, so the thermodynamic limit is not taken at fixed coupling; over the measurement correlation time tau ~ 1/p, the expected number of non-Gaussian gates per bond is tau q(N) -> 0. The exponent alpha(beta) in the inset is extracted without a convergence check (no alpha(beta,N) versus 1/N, no scaling collapse for beta>0) and without reported uncertainties. If alpha drifts with N, the distinction between alpha=1 and alpha<1 phases is not established. Please provide a systematic finite-size analysis for each beta, including the fitted functional form, the treatment of logarithmic corrections, and error bars.","section":"Fig. 4(b) and inset; 'Doped monitored dynamics'"},{"comment":"The numerical evidence is entirely restricted to Clifford gates, and the justification for transferring conclusions to generic (Haar) non-Gaussian circuits relies on the unitary 3-design property. That property controls averages of few-replica observables over the ensemble; it does not automatically control phase boundaries and critical exponents of postselected monitored dynamics, which are properties of individual trajectories. The abstract and conclusions state results for 'generic entanglement structure' rather than for the Clifford ensemble. Please either add small-N exact comparisons with non-Clifford non-Gaussian gates (e.g., matchgates combined with T-like or other non-Clifford gates) to test universality, or explicitly delimit the claims to the Clifford ensemble.","section":"Setup and footnote [92]"},{"comment":"The single-arc master equation closes the dynamics with an ad hoc rule assigning probability 1/2 to resetting the arc length to zero and probability 1/2 to composing two arc lengths. The text itself acknowledges that this description does not capture the CPLC log^2 N contribution and that RG/field-theoretic corrections exist. Without a quantitative estimate of the closure error, the analytic predictions for the Gaussian monitored phase (in particular pc|CG ~ 0.36 and the log-law coefficient) should be presented as an approximate analytical model rather than as an exact derivation. Please state the expected range of validity or provide a controlled justification for the closure.","section":"End Matter, 'Master equation for the monitored Clifford Gaussian dynamics'"},{"comment":"The claims of ballistic growth S ~ t and KPZ-like fluctuations delta S ~ t^{1/3} are central, but the evidence is a visual comparison at a single system size N=600, with no exponent fit, error bars, or finite-size collapse. Please provide quantitative local-exponent fits and, ideally, a collapse in the variable that encodes the crossover at N_NG ~ N (e.g., t/N), or at least report the statistical uncertainty of the extracted exponents.","section":"Fig. 2(a) and 'Unitary dynamics and Page Curve'"}],"minor_comments":[{"comment":"The caption says the crossing identifies 'a MIPT between a region with a super-logarithmic entanglement growth'; the sentence is incomplete and should end with 'and an area-law region.'","section":"Fig. 4(a) caption"},{"comment":"The footnote about magic states and Clifford universality appears as a numbered entry in the reference list; it should be formatted as a footnote or integrated into the text.","section":"Reference/footnote [84]"},{"comment":"No data or code availability statement is provided; for a Letter whose central claims depend on stabilizer simulations, please add a statement on data/code availability and report the number of trajectories used in each figure.","section":"Reproducibility"},{"comment":"The caption states 'Data obtained by averaging over 500 quantum trajectories' but does not specify the number of realizations for the other panels; please make the averaging consistent and explicit across all figures.","section":"Fig. 4(b) caption"}],"recommendation":"major_revision","confidential_remarks":"The paper addresses a timely and interesting question and contains a clean analytic treatment of the Gaussian arc model. My main reservation is the uncontrolled finite-size extrapolation in the monitored phase diagram for beta>0, which is load-bearing for the extensive-per-layer claim. I also think the Clifford-only numerics should be either supplemented by non-Clifford tests or accompanied by carefully qualified wording. None of these concerns suggest dishonesty or circularity; they are standard convergence and universality checks."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper is a solid step on a question worth asking: how much non-Gaussianity do you need to turn free-fermion dynamics into generic dynamics? The controlled interpolation via doping density q = eta/N^beta is the real contribution, and it pays off in concrete thresholds: O(N) total gates for unitary circuits, O(N) per layer for monitored ones. The Gaussian limit is treated properly. The arc-model master equation is a genuine analytic result, internally consistent, and the authors honestly flag that it misses the log^2 N correction from loop models. The Page-curve crossover and the KPZ fluctuation recovery are nice supporting evidence.\n\nThe soft spot is the monitored phase diagram for beta > 0. The stress-test note is on target: with eta fixed and beta > 0, q(N) -> 0 as N grows, so the thermodynamic limit is not taken at fixed Hamiltonian parameters. Over the measurement correlation time tau ~ 1/p ~ 100 layers, the expected number of non-Gaussian gates touching a bond is tau q(N) -> 0, so locally the dynamics becomes Gaussian as N grows. The exponent alpha in Fig. 4(b) is inferred from N = 64 to 512, only a factor of 8, with no scaling collapse or convergence check shown for beta > 0. The reported collapse with nu ~ 1.3 is for beta = 0, where q is fixed. If alpha drifts with N, the claim that volume-law strictly requires an extensive per-layer rate is not yet established. That said, the beta = 0 result is secure, and the beta > 0 claim is plausible; it just needs larger sizes or a different construction where q stays fixed.\n\nThe Clifford restriction is a real caveat — 3-design guarantees for averaged observables do not automatically transfer to postselected trajectories — but it is a standard working assumption in this subfield, not a unique flaw. The absence of released code and documented exponent extraction is a practical barrier to verification, and the reader's conditional verdict is reasonable on that basis.\n\nFor anyone working on quantum circuits or measurement-induced transitions, this is a useful and honest paper. It deserves a serious referee. I would send it to peer review with the request that the authors either supply a convergence check of alpha(beta,N) at larger N or soften the extensive-rate claim accordingly.","headline":"A clean analytic arc-model result for the Gaussian limit plus a plausible but not yet fully converged monitored phase diagram; the extensive-per-layer-rate claim needs a larger-N scaling check before it is secure.","tokens_in":16010,"tokens_out":2410,"would_cite":true,"duration_ms":27195,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Doping matchgate circuits with an extensive number of non-Gaussian gates restores the generic entanglement structure of random quantum circuits, and under measurements a genuine volume-law phase requires an extensive per-layer doping rate.","keywords":["matchgate circuits","non-Gaussianity","fermionic magic","entanglement growth","measurement-induced phase transition","Kardar-Parisi-Zhang universality","Clifford circuits","arc model"],"falsifier":"Simulate monitored doped matchgate dynamics with a genuinely non-Clifford non-Gaussian gate ensemble, such as Haar-random two-qubit gates in place of the non-Gaussian Clifford gates, and check whether a volume-law phase still appears only for an extensive per-layer doping rate of $\\mathcal{O}(N)$; if the threshold moves or the power-law phase changes, the Clifford restriction is the reason.","tokens_in":15002,"feed_emoji":"⚛️","tokens_out":12997,"duration_ms":122672,"temperature":0.7,"pith_summary":"This paper asks how much non-Gaussianity must be injected into free-fermion matchgate circuits before their entanglement dynamics becomes generic. In purely unitary circuits, an intensive doping rate per unit time suffices: once the total number of injected non-Gaussian gates reaches $\\mathcal{O}(N)$, entanglement growth crosses from diffusive $\\sqrt{t}$ to ballistic $t$ and its fluctuations enter the Kardar-Parisi-Zhang universality class. Under monitoring, the same doping only produces a power-law entangled phase $S \\sim N^{\\alpha}$, and a genuine volume-law phase requires injecting non-Gaussian gates at an extensive rate, $\\mathcal{O}(N)$ per layer. The paper also gives an exact arc-model description of the purely Gaussian limit, explaining the diffusive growth and the logarithmic entanglement phase of monitored free fermions. These results identify non-Gaussianity as a resource whose amount and per-layer injection rate control the emergence of generic entanglement structure.","feed_headline":"An extensive dose of non-Gaussian gates restores generic entanglement","feed_subtitle":"Ballistic growth and KPZ fluctuations return; under monitoring, a volume law needs an extensive per-layer doping rate.","key_machinery":"The load-bearing object is the arc model: under Clifford-Gaussian matchgate dynamics, stabilizer generators are represented as arcs pairing $2N$ Majorana points on a line, and each gate randomly permutes arc endpoints. The entanglement entropy across a bipartition equals half the number of arcs crossing the cut, so the Gaussian dynamics reduces to a classical stochastic process. In the unitary case the endpoint distribution evolves by a binomial recursion that converges to a Gaussian with variance $4t$, yielding $S_{N/2}(t) \\approx \\sqrt{t/\\pi}$; in the monitored case the measurement rule glues arcs together, and a nonlinear master equation for the arc-length distribution has a steady-state solution $P(\\ell) \\sim 1/\\ell^2$, which produces $S \\sim \\log N$. Doping with non-Gaussian gates breaks this free-fermion description, and the paper uses stabilizer-formalism simulations of Clifford circuits to reach system sizes of several hundred qubits.","core_discovery":"The central claim is that the free-fermion matchgate class of random circuits can be pushed into the generic regime of random quantum circuits by doping with non-Gaussian gates, and that the resource count is controlled by the exponent $\\beta$ in the per-layer doping rate $q = \\eta / N^{\\beta}$. For unitary evolution, ballistic entanglement growth $S(t) \\sim t$ and Kardar-Parisi-Zhang fluctuations $\\sim t^{1/3}$ are recovered once the total number of injected non-Gaussian gates $N_{\\mathrm{NG}} = \\eta t$ becomes extensive, $\\mathcal{O}(N)$, and the late-time Page curve crosses from the Gaussian-stabilizer curve to the universal stabilizer curve with the deviation decaying exponentially in the doping density. Under measurements, pure Gaussian circuits have no volume law at any $p>0$, showing a logarithmic phase for weak monitoring and an area law above $p_c \\approx 0.36$; doping produces a power-law entangled phase $S \\sim N^{\\alpha}$ with $\\alpha$ controlled by the doping rate, and only an extensive per-layer doping rate ($\\beta = 0$) restores a true volume law, with critical exponent $\\nu \\approx 1.3$ matching the Clifford measurement-induced transition universality class.","pith_inferences":["If the Clifford 3-design assumption survives postselection, the $\\mathcal{O}(N)$-per-layer threshold should be unchanged for Haar-random non-Gaussian gates, making it a genuine resource bound rather than a Clifford-sampling artifact.","The continuously tunable exponent $\\alpha$ in the power-law phase suggests a one-parameter family of states that an effective field theory might describe; the paper explicitly leaves an analytic framework for the doped dynamics open.","The exact arc model may admit a generalized stochastic description in which non-Gaussian gates act as additional moves, yielding a testable prediction for how $\\alpha$ depends on the doping density.","The exponential decay of the Page-curve deviation with doping density implies that only a few $\\mathcal{O}(N)$ non-Gaussian gates already produce near-generic late-time entanglement, relevant for fermionic linear-optics experiments with limited magic resources."],"forward_implications":["An intensive doping rate (a fixed fraction of gates per layer) is enough to restore generic unitary entanglement dynamics, but not enough under monitoring.","The minimal non-Gaussian resource for a stable volume-law phase in monitored circuits is an extensive rate of $\\mathcal{O}(N)$ non-Gaussian gates per layer.","Between area law and volume law, monitored doped matchgate circuits host a power-law entangled phase $S \\sim N^{\\alpha}$ with continuously tunable $\\alpha$.","The purely Gaussian monitored transition is captured analytically by the arc model, yielding logarithmic entanglement for weak measurements and a transition at $p_c \\approx 0.36$.","Once extensive doping is present, the monitored critical behavior belongs to the Clifford measurement-induced transition universality class, with $\\nu \\approx 1.3$."],"supporting_citations":[{"why":"Supplies the generic random-unitary baseline of ballistic growth and KPZ fluctuations that doping is shown to restore.","marker":"[5]"},{"why":"Defines the measurement-induced entanglement transition in hybrid circuits, including the volume-law phase that monitored stabilizer circuits exhibit.","marker":"[11]"},{"why":"Provides the stabilizer-circuit MIPT universality class and critical exponent used to identify the extensive-doping limit.","marker":"[22]"},{"why":"Establishes the fermionic-Gaussian-state formalism and Majorana correlation description on which the arc model and efficient simulation rest.","marker":"[40, 41]"},{"why":"Shows the Clifford group is a unitary 3-design, the stated justification for approximating Haar-random circuits with Clifford gates.","marker":"[87, 88]"},{"why":"Shows the Clifford-Gaussian set is a 3-design for the Gaussian ensemble, justifying the Clifford restriction in the free-fermion case.","marker":"[90, 91]"},{"why":"Provides the diffusive Majorana-defect model whose arc statistics the monitored Gaussian phase is compared with.","marker":"[95]"}],"fun_headline_variants":["Doping matchgates with non-Gaussian gates recovers generic entanglement","Non-Gaussian doping restores ballistic growth and KPZ fluctuations","Extensive non-Gaussian gates revive volume law in monitored matchgates","Matchgate circuits become generic with enough non-Gaussian resources"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that drawing every gate from the Clifford group reproduces the entanglement dynamics of fully random gates, a statistical equivalence known as a 3-design; under postselected monitored dynamics that equivalence is not guaranteed to hold.","fun_headline_variants_meta":{"raw":{"variants":["Doping matchgates with non-Gaussian gates recovers generic entanglement","Non-Gaussian doping restores ballistic growth and KPZ fluctuations","Extensive non-Gaussian gates revive volume law in monitored matchgates","Matchgate circuits become generic with enough non-Gaussian resources"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000239,"raw_usage":{"total_tokens":1549,"prompt_tokens":1012,"completion_tokens":537,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":628,"completion_tokens_details":{"reasoning_tokens":463}},"tokens_in":628,"tokens_out":537,"duration_ms":6413,"temperature":1.0,"reasoning_tokens":463,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T16:45:34.523370+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate monitored doped matchgate dynamics with a genuinely non-Clifford non-Gaussian gate ensemble, such as Haar-random two-qubit gates in place of the non-Gaussian Clifford gates, and check whether a volume-law phase still appears only for an extensive per-layer doping rate of $\\mathcal{O}(N)$; if the threshold moves or the power-law phase changes, the Clifford restriction is the reason.","supporting_citations":[{"cited_title":"Nahum and B","cited_arxiv_id":null,"evidence_quote":"Provides the diffusive Majorana-defect model whose arc statistics the monitored Gaussian phase is compared with."}],"review_version":1}