{"id":"d6ed929f-ae26-4ceb-89b1-3443bc5c87b7","arxiv_id":"1908.11253","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":8,"one_line_summary":"In a 1D Bose-Hubbard quench, increasing the random local measurement rate drives a volume-to-area law entanglement phase transition with critical exponent nu≈2.","lead":"Random local measurements switch how quantum entanglement spreads in a one-dimensional Bose-Hubbard model: below a critical measurement rate entanglement grows with volume, above it only with area. This is one of the first demonstrations of a measurement-induced entanglement transition in a non-integrable lattice Hamiltonian, extending earlier results from random circuits.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"MPS truncation error is never quantified, so the volume-law phase and the extracted critical point could be bond-dimension artifacts; a χ-convergence check is required.","rationale":"The reader's weakest assumption identifies two premises: MPS truncation fidelity and the validity of the finite-size scaling ansatz. I focus on the former because it is the more load-bearing for the existence of the transition itself. If the MPS bond dimension is insufficient, the apparent volume-law phase may simply be the entanglement ceiling of the tensor network, and the transition emerges where the entropy falls below that ceiling—a numerical artifact, not a physical phase transition. The paper states in Section II that χ=2048 was \"test to be fine\" but provides no supporting data, no truncation error estimates, and no convergence in χ. The L0=48 test is also insufficient because truncation error grows with time and projection count, not with system size. The paper itself flags another limitation in Section III: the mutual-information critical exponent shifts from Δ≈1.29 at L0=36 to Δ≈1.56 at L0=48, described as \"seems non-universal\" and attributed to finite-size effects. That is a real concern for the universality claim, but it does not directly threaten the existence of the transition. The truncation concern, by contrast, could invalidate the phase diagram entirely. Because the reader's verdict is already CONDITIONAL, and the concern is precisely one of the conditions that needs to be checked, the verdict should remain unchanged pending a convergence test.","tokens_in":10645,"tokens_out":5838,"duration_ms":58552,"concrete_test":"Recompute the Nt=50 data at Px=0.02, 0.04, 0.06, and 0.08 for L0=36 and T=30 with bond dimensions χ=1024, 2048, 4096, and 8192, recording S(L=8) and S(L=16) at final time and the discarded weight per time step. If S(L) at any Px below the claimed critical point increases by more than 5% when χ is doubled, or if the data-collapse value of Px,c shifts by more than the quoted uncertainty (±0.004), the volume-law phase and transition parameters are not converged. A simpler diagnostic is to check whether the largest Schmidt values for the central cut at T=30 approach the χ cutoff (entropy near log2 χ) for Px=0.02–0.05.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim, a volume-to-area-law transition at Px,c≈0.060 with ν≈2.00, depends on the assumption that MPS/TEBD with χ≤2048 and T=30 faithfully represents the post-measurement dynamics (Section II). The only support is the statement that χ=2048 was \"test to be fine\", with no convergence data, discarded-weight curves, or entropy-versus-χ comparisons. This is especially dangerous on the volume-law side of the transition: for Px<Px,c the state becomes highly entangled, and a rank-2048 MPS can carry at most log2(2048)≈11 ebits of von Neumann entropy across any cut. If the true volume-law entropy exceeds this ceiling, the computed entropy saturates at the bond-dimension bound, so the volume-law phase could actually be a truncation-limited plateau. Since the critical point and exponent are extracted from these entropy curves (Figs. 3 and 5), the phase boundary and ν could be numerical artifacts rather than physical. The L0=48 test reported in Fig. 6 does not rule this out, because truncation error accumulates with both time and the number of projections; it is a function of the simulation protocol, not of total system size alone.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies the quench dynamics of a one-dimensional Bose-Hubbard model subject to random local projective measurements, using matrix product states and the time-evolving block decimation algorithm. The authors introduce two tuning parameters, the spatial measurement rate Px and the temporal measurement rate Nt, and map out a two-dimensional phase diagram. They report a volume-to-area-law entanglement phase transition at Px,c≈0.060±0.004 with correlation-length exponent ν≈2.00±0.15 for Nt=50, a logarithmic growth of entanglement in time and in subsystem size at criticality, a power-law decay of mutual information, and a single-site entropy distribution that distinguishes the two phases. The paper interprets these results as evidence for a universal measurement-induced entanglement transition in a non-integrable lattice model, possibly describable by a conformal field theory.","tokens_in":10911,"tokens_out":4692,"duration_ms":44696,"significance":"If the numerical results are correct, this is an important extension of measurement-induced entanglement transitions from random circuit models to a concrete, non-integrable lattice Hamiltonian. The paper provides a useful demonstration that a stable volume-law phase survives finite measurement rates in a bosonic lattice model, and the data collapse with stated parameters is a credible first step. The L0=48 test, the focus on a microscopic lattice model, and the comparison with percolation/CFT expectations are valuable. However, the central quantitative claims rest on numerical simulations whose truncation error is never quantified, and the scaling analysis is used both to locate and to characterize the critical point. These issues must be addressed before the paper can be considered fully convincing.","major_comments":[{"comment":"The central numerical premise — that TEBD with bond dimension up to 2048 faithfully simulates the post-measurement dynamics up to T=30 — is asserted but not demonstrated. The text states that χ=2048 was \"test to be fine\", yet no discarded-weight curves, no S versus χ comparisons, and no dependence of the extracted Px,c on χ are shown. This is particularly serious on the volume-law side of the transition (e.g., Px=0.02–0.04 in Fig. 3), where a rank-2048 MPS can carry at most about 11 ebits of von Neumann entropy across a cut; if the true entropy exceeds this ceiling, the computed curves would saturate at the bond-dimension bound and could mimic the observed area-law behavior. The L0=48 check in Fig. 6 does not rule this out, because truncation error accumulates with time and with the number of projections. I ask the authors to add a systematic convergence test for representative Px on both sides and at the putative critical point, showing, for example, S(L=8,t=T) versus χ, discarded weight versus time, or a comparison of the phase boundary at two or more bond dimensions.","section":"Section II, bond-dimension statement"},{"comment":"The finite-size scaling analysis that yields Px,c≈0.060±0.004 and ν≈2.00±0.15 is not described in sufficient detail. The same data are used to determine both the critical point and the exponent, but the paper does not state how the values and uncertainties were obtained, how many subsystem sizes were included, or whether the collapse is stable under excluding the smallest L. Without a defined collapse metric or a bootstrap over disorder realizations, the quoted error bars are not reproducible. In addition, the inference of a \"single universality class\" from one model and one exponent is stronger than the evidence supports; the agreement with ν≈2 from random circuits is suggestive but not a demonstration of universality. At minimum, I request a more robust fitting procedure and a more cautious interpretation, or a collapse of additional observables such as the mutual information.","section":"Section III, Eq. (2) and Fig. 5"},{"comment":"The mutual information at the critical point yields an exponent Δ≈1.29 for L0=36 and Δ≈1.56 for L0=48, while the CFT argument quoted in the paper predicts Δ=2. The paper attributes the discrepancy to finite-size effects, but with only two system sizes there is no evidence of convergence, and the trend with L0 is in the right direction but still far from the predicted value. To support the CFT description, the authors should either provide a finite-size extrapolation of Δ (e.g., Δ versus 1/L0) or explicitly state that the current data are consistent with a power-law decay but do not yet determine the universal exponent.","section":"Section III, Fig. 6"},{"comment":"The phase diagram reports critical spatial rates Px,c≈0.06, 0.08, 0.12, 0.24 for Nt=50, 40, 30, 20, but no error bars or extraction method are given for the Nt<50 values. Since the paper uses the Nt-dependence of Px,c to argue for the robustness of the transition and the role of information scrambling, please specify how these estimates were obtained and provide uncertainties. The current presentation makes it difficult to assess whether the claimed trend is significant.","section":"Section III, Fig. 1(b)"}],"minor_comments":[{"comment":"The phrase \"test to be fine\" should read \"tested to be fine\". In addition, the statement about the maximum boson number per site being 5 should be accompanied by a check that this truncation does not affect the results.","section":"Section II"},{"comment":"The number of random realizations is stated as \"up to 900\" in the text, but the figure does not show error bars or standard errors. Please specify the number of realizations used for each Px and include error estimates on the averaged entropy curves.","section":"Section III, Fig. 2"},{"comment":"The probability distributions of the single-site entropy are described only qualitatively. Please specify how the distributions are computed (binning, number of samples, whether they are averaged over sites and realizations) and provide the sample size for each curve.","section":"Section III, Fig. 4"},{"comment":"The growth in the volume-law phase is described as \"ballistically\" in one place and as \"linear dependence (with possible logarithmic correlations)\" in another; please use consistent terminology and clarify which quantity is being fit.","section":"Section III, Fig. 2 and text"},{"comment":"The quantity S(Px,c) in Eq. (2) is not precisely defined: is it the measured entropy at the estimated critical point for each L, or a fitting parameter? Please clarify the definition and how it is evaluated.","section":"Section III, Eq. (2)"}],"recommendation":"major_revision","confidential_remarks":"The topic is timely and the central observation is plausible, but the quantitative claims are not yet fully supported without a truncation-error benchmark and a more transparent scaling analysis. The stress-test concern about MPS bond-dimension saturation is valid and lands directly on the paper's central claim. If the authors provide the requested convergence tests and a more conservative discussion of the universality and CFT claims, the paper would be suitable for publication. I see no scope or citation-practice problems requiring editorial action."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear —,\n\nThe one thing to know: this is the first non-integrable lattice-model demonstration of the measurement-induced entanglement transition, and the qualitative picture — stable volume-law phase at low measurement rate, area law at high rate, logarithmic entropy at the crossover — is probably correct. The quantitative case is shakier. The paper quotes Px,c ≈ 0.060±0.004 and ν≈2.00±0.15, but those numbers rest on MPS data with no truncation benchmark, and the mutual-information exponent drifts from 1.29 to 1.56 with system size. I would describe the transition as established in broad strokes and the universality class as not.\n\nWhat is actually new: the Bose-Hubbard model with projective measurements, simulated by TEBD up to χ=2048. That extends the transition beyond Clifford/random circuits and non-interacting models, which matters because the field has been worried about whether the effect survives generic interactions. The two-parameter (Nt, Px) phase diagram with the information-scrambling dependence is a nice addition, and the single-site entropy distribution as a transition indicator is a useful, cheap diagnostic. The paper is also honest about its own weak points — it flags the non-universal mutual-information exponent and blames finite size. Citation pattern is appropriate; no sign of missing key refs.\n\nThe soft spots, in proportion. The biggest is the missing truncation-error analysis. The stress-test note is right: on the volume-law side, a rank-2048 MPS can carry at most log2(2048) ≈ 11 ebits across a cut, and a 36-site chain with local dimension 6 has plenty of room to exceed that. If the true entropy does, the 'volume-law' curves are a bond-dimension plateau, and Px,c extracted from them is a numerical artifact. Simply asserting that χ up to 2048 was tested to be fine is not enough, especially since the error accumulates with time and with each projection. A single discarded-weight or S(χ) plot would answer this. The L0=48 check does not help, because total size is not the relevant variable. Second: Eq. 2 is used both to locate Px,c and to extract ν from the same data, so the collapse partly supports itself; the quoted error bars look like fitting errors, not statistical ones. Third: the mutual-information exponent changes by about 20% between L0=36 and 48, which is more than a footnote. It does not sink the paper, but it undercuts the 'single universality class' conclusion as written.\n\nWho gets value: tensor-network people and anyone working on measurement-induced transitions, as a case study with a different Hamiltonian class. I would send this out for review, but with the expectation that the authors add a χ-convergence check and dial back the universality claim. As it stands, I would not cite the precise exponents, only the qualitative observation.\n\nRegards.","headline":"First non-integrable lattice demonstration of the measurement-induced transition, but quantitative claims are undercut by a missing truncation benchmark and a drifting mutual-information exponent.","tokens_in":11473,"tokens_out":3230,"would_cite":true,"duration_ms":31087,"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":"Random measurements switch entanglement from volume to area law","keywords":["measurement-induced phase transition","entanglement entropy","Bose-Hubbard model","matrix product state","projective measurement","volume law","area law","universality class"],"falsifier":"Compute the same quench with successively larger bond dimensions (or a different trotterization) at $N_t=50$, $P_x=0.06$ on $L_0=48$ and check whether the area-law saturated entropy and the data collapse of Eq. (2) are stable; if the entropy at fixed $P_x>P_{x,c}$ decreases systematically with bond dimension, or if the collapse fails when $L_0=48$ data are included at the quoted exponents, the central claim is not supported.","tokens_in":10402,"feed_emoji":"⚛️","tokens_out":6526,"duration_ms":57700,"temperature":0.7,"pith_summary":"This paper tries to establish that a measurement-induced entanglement phase transition is not a special feature of abstract random circuits, but occurs in a microscopic, non-integrable lattice model: a one-dimensional Bose-Hubbard chain. Simulating quantum quench dynamics with random local projective measurements, it finds a stable volume-law entangling phase at low measurement rates that gives way to an area-law disentangling phase once the spatial measurement rate exceeds a critical value. For fixed measurement frequency in time, the critical rate is $P_{x,c}\\approx 0.060\\pm 0.004$ and the correlation-length exponent is $\\nu\\approx 2.00\\pm 0.15$, both extracted from a data collapse of the von Neumann entropy. If correct, the result generalizes the measurement-induced transition from Clifford and random unitary circuits to a generic interacting lattice Hamiltonian, and supports a single, possibly conformal, description of the transition.","feed_headline":"Random measurements switch entanglement from volume to area law","feed_subtitle":"Bose-Hubbard quench data put measurement-induced criticality on a non-integrable lattice model, with ν≈2.","key_machinery":"The carrying mechanism is the matrix-product-state time-evolving block decimation simulation of the Bose-Hubbard Hamiltonian $H=-J_0\\sum_i(b_i^\\dagger b_{i+1}+\\mathrm{h.c.})+\\frac{U}{2}\\sum_i n_i(n_i-1)$, with projective measurements inserted at randomly chosen time layers and spatial sites. Two rates control the dynamics: $N_t$, the number of measured layers per unit time (which sets the scrambling depth between measurements), and $P_x$, the probability a given site is projected on a measured layer. The finite-size scaling ansatz $S(P_x)-S(P_{x,c})=F((P_x-P_{x,c})L^{1/\\nu})$ is the instrument that turns the entropy curves into a transition: data collapse yields the quoted critical rate and exponent.","core_discovery":"The central claim is that repeated local projective measurements $|1_x\\rangle\\langle 1_x|$ compete with unitary evolution to produce a genuine dynamical phase transition in the entanglement structure of the post-quench steady state. In the entangling phase, the von Neumann entropy of a subsystem saturates near its volume-law value; in the disentangling phase it saturates at small area-law values. At the transition the entropy grows logarithmically both in time ($S\\simeq a\\ln t+b$ with $a\\approx 0.22$) and in subsystem size ($a\\approx 0.21$), and the mutual information between distant sites decays as a power law. A finite-size scaling collapse using $S(P_x)-S(P_{x,c})=F((P_x-P_{x,c})L^{1/\\nu})$ gives $P_{x,c}\\approx 0.060\\pm0.004$ and $\\nu\\approx 2.00\\pm 0.15$. The authors read these results as evidence that the transition belongs to a single universality class shared with random unitary circuit models and, through the boundary percolation value $\\nu=2$, may admit a conformal field theory description.","pith_inferences":["If the mutual-information exponent keeps moving toward the CFT value $\\Delta=2$ with system size as the paper's $L_0=36$ to $48$ trend suggests, the apparent non-universality is finite-size, and larger simulations should find $\\Delta$ approaching 2.","The role of randomness in the unitary layers could be tested by replacing the Bose-Hubbard Trotter evolution with a periodic (non-random) drive at fixed $N_t$; if $\\nu$ changes, randomness of the unitary gates is essential, and if not, the transition is driven by measurement randomness alone.","A direct cold-atom realization would use a quantum gas microscope to perform projective number measurements on individual sites; the predicted markers are the crossover of the subsystem entropy from volume to area scaling and the critical logarithmic growth, measurable in principle."],"forward_implications":["The volume-law phase survives small but finite measurement rates in a non-integrable lattice model, so the transition is not an artifact of Clifford or random-unitary circuit structure.","The extracted exponent $\\nu\\approx 2$ places the transition in the same universality class as random unitary circuit numerics and the boundary percolation prediction, making a unified CFT description plausible.","At criticality, scale invariance shows up as logarithmic entropy growth in time and space and as power-law mutual information decay, giving concrete signatures for other models.","The critical spatial rate grows faster than linearly as $N_t$ decreases, showing that the density of measurements alone ($N_t P_x$) does not set the phase boundary; scrambling between measurements matters.","The single-site entanglement entropy distribution changes character across the transition, offering a local probe that connects the measurement transition to the thermal-to-MBL transition."],"supporting_citations":[{"why":"Introduces the random-unitary-circuit measurement transition and percolation picture; supplies the logarithmic-growth and scaling expectations the Bose-Hubbard results are compared against.","marker":"[36]"},{"why":"Demonstrates a stable volume-law phase at finite measurement rate in Clifford circuits, the baseline for a robust entangling phase.","marker":"[37]"},{"why":"Provides Clifford-circuit numerics of the transition and the mutual-information diagnostics used here.","marker":"[38]"},{"why":"Shows how scrambling depth and measurement fraction control the transition, motivating the $N_t$ dependence mapped in the phase diagram.","marker":"[39]"},{"why":"Introduces continuous measurement strength via POVMs and finds a transition at finite strength, a comparison point for finite-rate behavior.","marker":"[40]"},{"why":"Reports a non-interacting Bell-pair model where no transition occurs, the contrast case that motivates testing a non-integrable lattice model.","marker":"[41]"},{"why":"Uses the single-site entanglement entropy distribution as a diagnostic of the MBL transition, which this paper adapts to the measurement transition.","marker":"[47]"},{"why":"Provides the random-tensor-network boundary-percolation argument that the physical exponent is $\\nu=2$, the theoretical value the numerical $\\nu\\approx2$ is matched against.","marker":"[60]"},{"why":"Proves logarithmic entanglement scaling at critical percolation, supporting the scale-invariant interpretation of the critical entropy growth.","marker":"[61]"}],"fun_headline_variants":["DMRG maps measurement-induced entanglement transition in Bose-Hubbard","Volume-to-area law transition driven by measurements in non-integrable model","Bose-Hubbard measurements: volume-to-area entanglement transition","Measurement-induced criticality: log growth at transition in Bose-Hubbard","Non-integrable Bose-Hubbard shows measurement-driven volume-to-area transition"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The identification of the phase transition rests on the numerical premise that matrix-product-state simulations with bond dimension up to 2048 over time $T=30$ faithfully capture the post-measurement dynamics on chains of length 36 and 48, and on the assumed one-parameter finite-size scaling form; if truncation error grows under repeated projections or the scaling form is invalid, the extracted $P_{x,c}$ and $\\nu$ could be numerical artifacts.","fun_headline_variants_meta":{"raw":{"variants":["DMRG maps measurement-induced entanglement transition in Bose-Hubbard","Volume-to-area law transition driven by measurements in non-integrable model","Bose-Hubbard measurements: volume-to-area entanglement transition","Measurement-induced criticality: log growth at transition in Bose-Hubbard","Non-integrable Bose-Hubbard shows measurement-driven volume-to-area transition"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00118,"raw_usage":{"total_tokens":4875,"prompt_tokens":943,"completion_tokens":3932,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":559,"completion_tokens_details":{"reasoning_tokens":3836}},"tokens_in":559,"tokens_out":3932,"duration_ms":27632,"temperature":1.0,"reasoning_tokens":3836,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T10:20:54.742671+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the same quench with successively larger bond dimensions (or a different trotterization) at $N_t=50$, $P_x=0.06$ on $L_0=48$ and check whether the area-law saturated entropy and the data collapse of Eq. (2) are stable; if the entropy at fixed $P_x>P_{x,c}$ decreases systematically with bond dimension, or if the collapse fails when $L_0=48$ data are included at the quoted exponents, the central claim is not supported.","supporting_citations":[{"cited_title":"Quantum Error Correction in Scrambling Dynamics and Measurement-Induced Phase Transition","cited_arxiv_id":"1903.05124","evidence_quote":"Shows how scrambling depth and measurement fraction control the transition, motivating the $N_t$ dependence mapped in the phase diagram."},{"cited_title":"Vasseur, A","cited_arxiv_id":null,"evidence_quote":"Provides the random-tensor-network boundary-percolation argument that the physical exponent is $\\nu=2$, the theoretical value the numerical $\\nu\\approx2$ is matched against."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Proves logarithmic entanglement scaling at critical percolation, supporting the scale-invariant interpretation of the critical entropy growth."}],"review_version":1}