{"id":"c2ae127b-211b-4c52-9e98-32126998abd5","arxiv_id":"2501.16995","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A quantum-gas-microscope experiment shows that the subsystem Loschmidt echo, a local return probability, detects dynamical quantum phase transitions and quantitatively maps the accessible Hilbert-space dimension in ergodic and fragmented regimes of a Bose-Hubbard chain.","lead":"Experiments in a quantum gas microscope measure the subsystem Loschmidt echo, the probability that a small block of lattice sites returns to its initial density pattern after a quantum quench. The echo reveals a dynamical quantum phase transition and, at later times, measures how much of the many-body Hilbert space the system can actually access, including signatures of fragmented, non-thermalizing dynamics.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Long-time dimension extraction assumes quasi-equilibration by t≈230–280 ħ/J, which is not established for the ergodic domain-wall state; a later-time slope check would settle whether Eq. (5) measures Hilbert-space dimension or transient dynamics.","rationale":"I read the paper as making two connected claims: the SLE captures DQPT non-analyticities through higher-order correlations, and the long-time SLE slope measures the accessible Hilbert-space dimension. The first claim is well supported: the rate-function sharpening, the exponential decrease of τ_c^N for N=1–5, and the connected-correlation expansion (Eq. 3, Fig. 2d) are consistent and backed by TEBD numerics. The DQPT conclusion does not depend on the excluded N=6,7 points in a load-bearing way, since the trend is already visible. The second claim is the one that carries the quantitative headline result. Its validity requires that by t≈230–280 ħ/J the local string probabilities have thermalized within the accessible space, so that -ln L_N equals the thermodynamic entropy up to the small β²δE² correction. The available evidence is suggestive but incomplete: Krylov numerics in Fig. S4 show SLE relaxation for the CDW ergodic and DW fragmented cases, and the fragmented-regime slopes agree with TDVP and with the combinatorial fragment dimensions (√2 and 1), which independently supports that part. The ergodic DW case, however, still shows 6–15% residual imbalance (Table S2), and the extracted slope varies from 2.29 to 2.19 as the ROI grows, indicating that the observable has not fully forgotten the initial state. This is precisely the regime where Eq. (4) could fail. The β²δE² correction is never estimated experimentally, so even a small finite-temperature offset could bias the slope. I do not see an internal contradiction or a fatal error; the concern is an unverified premise, not a demonstrated false one. The proposed later-time and energy-variance checks would either confirm the quasi-equilibrium assumption or show that the quoted dimension is a transient, and the verdict should stay conditional until then.","tokens_in":29065,"tokens_out":5891,"duration_ms":59625,"concrete_test":"For the same ergodic quench (U/J≈2.7, Δ=0, domain-wall state), measure or simulate -ln L_N in additional time windows, e.g., t≈400–500 and 600–700 ħ/J, using TDVP for L≈60 with the same boson cutoff; if the fitted slope d(-ln L_N)/dN changes by more than the quoted statistical uncertainty (≈0.01) relative to the 230–280 ħ/J window, the extracted (dim H_eff)^{1/L} is transient. Independently, compute ½β²δE² from the initial-state energy variance in the post-quench Hamiltonian and verify it is small compared with the slope error.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim—that the slope of -ln L_N vs N equals the thermodynamic entropy density and yields (dim H_eff)^{1/L} via Eq. (5)—requires the SLE to have reached its quasi-equilibrium value by the measurement window t≈230–280 ħ/J. This premise is least secure for the ergodic domain-wall state: Table S2 reports a residual density imbalance of ≈6–15% depending on ROI, and the numerical relaxation check in Fig. S4 is shown for the CDW ergodic and DW fragmented cases, not for the DW ergodic case. If the DW ergodic subsystem strings are still relaxing, the slope d(-ln L_N)/dN is a transient rate rather than an entropy. A second, unquantified contribution is the ½β²δE² correction in Eq. (4): the energy variance δE² after the quench is not measured, so the reported slopes may include a finite-temperature offset. The first-order effective Hamiltonian Eq. (6) is a further assumption, but its effect is bounded by the good agreement of the fragmented-regime slopes with TDVP and combinatorial values, so the equilibration premise is the load-bearing one.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper experimentally introduces the subsystem Loschmidt echo (SLE) as a quasi-local observable that can be measured with a quantum gas microscope and demonstrates its utility in two regimes of the one-dimensional Bose-Hubbard model. In the short-time regime, after a quench from a charge-density-wave state in the hard-core limit, the SLE shows a sharpening kink-like feature around t ≈ 0.9 ħ/J that becomes more pronounced with increasing subsystem size N; the rate function −ln L_N/N develops a collapse for larger N, and the authors decompose L_N into connected density correlation functions, showing that genuine 3- and 4-point correlations dominate near the dynamical critical point. In the long-time regime, the time-averaged SLE is measured for subsystem sizes N = 1–5 in both an ergodic regime (U/J ≈ 2.7, Δ = 0) and a fragmented regime (U ≈ Δ, Δ/J ≈ 16.7). From the slope of −ln L_N versus N, the authors extract (dim H_eff)^{1/L} = 2.29(1) and 2.34(1) for the ergodic domain-wall and CDW states, and 1.54(1) and 1.12(1) for the fragmented domain-wall and CDW states, interpreting the reduction as direct evidence of Hilbert-space fragmentation. The results are compared with Krylov subspace, TEBD, and TDVP simulations, and the fragmented-regime slopes are also compared with a simple combinatorial counting based on the first-order effective Hamiltonian.","tokens_in":29327,"tokens_out":7127,"duration_ms":70347,"significance":"The SLE is a valuable new observable: it is exponentially less sensitive to noise than the full Loschmidt echo, it can be measured with existing quantum gas microscope techniques, and it gives access to both dynamical critical behavior and, through its long-time average, the effective Hilbert-space dimension. The short-time DQPT observation is convincing and is strengthened by the explicit decomposition into higher-order connected correlations and by the agreement with TEBD numerics. The long-time dimension extraction is supported by independent Krylov and TDVP simulations of the same Hamiltonian, which is a notable strength. If the equilibration premise is validated, the method provides a broadly applicable probe of ergodicity breaking and thermalization in quantum simulators. The paper is therefore of high significance for the quantum simulation and many-body physics communities.","major_comments":[{"comment":"The central quantitative claim that the slope of −ln L_N versus N equals the thermodynamic entropy density and yields (dim H_eff)^{1/L} via Eq. (5) requires that the SLE has reached its quasi-equilibrium value by the measurement window t ≈ 230–280 ħ/J. This premise is not established for the ergodic domain-wall state: Table S2 shows that the extracted (dim H_eff)^{1/L} varies from 2.29(1) to 2.19(1) as the ROI increases from 20 to 40 sites, with a residual imbalance of 6–15%, and the numerical relaxation check in Fig. S4 is shown for the CDW ergodic and DW fragmented cases, not for the DW ergodic case. If the DW ergodic subsystem strings are still relaxing at the measurement time, the slope measures a transient rate rather than an entropy, and the agreement with Krylov numerics could be coincidental if the numerics are averaged over a similar non-equilibrated time window. Please provide a direct relaxation check (numerical or experimental) for the DW ergodic case and quantify the systematic error due to the ROI dependence.","section":"Long-time dynamics, Eq. (5), Table S2, Fig. S4"},{"comment":"The correction term ½ β² δE² in Eq. (4) is never quantified. The authors argue that β is small for high-energy initial states, but the domain-wall initial state in the ergodic regime may have a finite energy density, and the observed residual imbalance suggests non-equilibrated dynamics on the probed timescale. Please estimate β and δE² from the quench parameters or from the independent numerical simulations and show that ½ β² δE² is below the reported statistical error of the slopes; otherwise the reported values of (dim H_eff)^{1/L} may contain a finite-temperature offset that is not included in the error budget.","section":"Eq. (4)"},{"comment":"The linear fits used to extract the slopes in Fig. 3b are not documented in sufficient detail. In particular, the SI fits the numerical data in Fig. S13 with a line forced through the origin, but if the intercept is nonzero for the small subsystem sizes N = 1–5 probed experimentally, forcing through the origin biases the slope. Please report free linear fits with intercepts, their uncertainties, and the reduced chi-square for both the experimental and numerical data, and show the sensitivity of (dim H_eff)^{1/L} to excluding the N = 1 point.","section":"Fig. 3b and SI Sec. V.A"}],"minor_comments":[{"comment":"The word 'echos' in the title should be 'echoes'.","section":"Title and abstract"},{"comment":"The formal definition of the SLE in the paragraph after Eq. (1) contains garbled notation: 'L(i)N = DQi+N −1 i ˆPi E' should be rewritten as an expectation value of a product of local projectors, for example ⟨∏_{j=i}^{i+N-1} P̂_j⟩.","section":"Main text, definition of SLE"},{"comment":"The exponential fit to τ_c^N is shown in Fig. 2c, but the fit function and the extracted parameters are not given in the main text; please provide them or refer explicitly to the SI equation (S19) and the fit results in Fig. S8.","section":"Main text, Fig. 2c"},{"comment":"The notation S_eff^N for the effective entropy is introduced but not defined; please define it explicitly and connect it to −ln L_N.","section":"Main text, page 5"},{"comment":"The post-selection on chain filling between 0.42 and 0.55 is described, but the effect of this post-selection on the SLE and its potential bias is not discussed; a sentence justifying the chosen window and its impact on the results would be helpful.","section":"SI Sec. III.A.2"}],"recommendation":"major_revision","confidential_remarks":"The paper is strong and the numerical benchmarks are a real asset. The main risk to the central long-time claim is the unverified quasi-equilibration for the ergodic domain-wall state, made concrete by the ROI dependence in Table S2. I would ask the authors for the additional relaxation check and for a quantified bound on the β² δE² correction before publication; the short-time DQPT part does not require changes beyond presentation. The fit documentation issue should also be fixed, since the extraction of the slope is the quantitative heart of the long-time claim."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, the short-time DQPT measurement is the strongest part of the paper: the kink sharpens systematically with subsystem size, the rate function collapses, and the decomposition of L4 into connected correlations shows that 3- and 4-point correlations carry the cusp. That is a real, clean result. Second, the long-time Hilbert-space dimension extraction is plausible but not airtight. It depends on the SLE having reached quasi-equilibrium by t ≈ 230–280 ħ/J, and for the ergodic domain-wall state the residual imbalance of 6–15% (Table S2) says that premise is shaky. The paper would be stronger with a later-time check for that state.\n\nWhat is actually new: the measured SLE dynamics in the 1D Bose-Hubbard model, the higher-order-correlation explanation of the short-time DQPT, and the quantitative fragment dimensions 1.54(1) and 1.12(1) in the fragmented regime. The SLE concept and its DQPT connection come from prior theory by the same groups (refs 30,31), but the experimental realization and the entropy-to-dimension extraction are new. The numerical benchmarks—Krylov and TDVP—are independent simulations of the same Hamiltonian, and the imperfection models (atom loss, initial-state errors) are a good-faith attempt to close the theory-data gap. No code is shipped, but the SI gives enough detail for a competent group to reproduce the main numbers.\n\nSoft spots, in proportion. The equilibration worry is the main one: the domain-wall state in the ergodic regime retains 6–15% imbalance depending on ROI, and the extracted dimension drifts from 2.29(1) to 2.19(1) as the ROI grows (Table S2). That drift is small but in the direction you would expect from transient dynamics, so a later-time measurement or a numerical relaxation plot for that specific state would settle it. The ½β²δE² correction in Eq. (4) is never quantified experimentally; for a quench from a deep CDW at half-filling the temperature is plausibly high enough that this is minor, but it should be stated. The DQPT sharpness fit drops N=6,7 because χr>3; that is disclosed honestly, but it weakens the exponential scaling claim from measured to consistent with. Post-selecting up to 40% of chains on filling is standard, but could bias the entropy if loss is dynamics-dependent; the imperfection simulations partially cover this. The citation pattern is fine: the SLE framework is from refs 30,31, which are cited, and the new results stand on their own.\n\nBottom line: the central narrative is sound. The short-time result is solid, and the ergodic-versus-fragmented contrast is large and robust. This paper is for people working on DQPTs, Hilbert-space fragmentation, and quantum simulation with gas microscopes, and they will cite it. It deserves a serious referee: send it to review, asking for the equilibration check for the DW ergodic state and a sentence on the β²δE² term.","headline":"Solid experimental demonstration of the SLE as a probe of DQPTs and Hilbert-space fragmentation; the long-time dimension extraction needs an equilibration check for the ergodic domain-wall state.","tokens_in":29941,"tokens_out":3729,"would_cite":true,"duration_ms":32334,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper claims that the subsystem Loschmidt echo, a locally measurable string-overlap probability, captures dynamical quantum phase transitions and, through its late-time average, quantitatively reveals the dimension of the accessible…","keywords":["subsystem Loschmidt echo","dynamical quantum phase transition","Hilbert space fragmentation","accessible Hilbert space dimension","quantum gas microscope","Bose-Hubbard model","higher-order correlations","ergodicity breaking"],"falsifier":"Re-measure $-\\ln L_N$ in later time windows, for example at 400, 600, and 1000 $\\hbar/J$, for the same initial states; if the extracted slope changes or drifts monotonically, the quasi-equilibrium premise fails. Alternatively, fix $\\beta$ from the measured density distribution and check the long-time identity with an independent estimate of the energy variance $\\delta E^2$.","tokens_in":28811,"feed_emoji":"⚛️","tokens_out":9001,"duration_ms":73687,"temperature":0.7,"pith_summary":"The paper tries to establish that a locally measurable quantity, the subsystem Loschmidt echo (the probability that a block of N adjacent sites still shows its initial occupation string at time t), carries the same deep information as the full return probability while remaining practical to measure. If true, a single snapshot-based observable would serve both sides of many-body physics: its short-time rate function locates a dynamical quantum phase transition, and its long-time average gives the effective thermodynamic entropy and hence the dimension of the Hilbert space a system can actually reach. The authors support this with ultracold-atom experiments in a one-dimensional Bose-Hubbard chain, showing a sharpening DQPT feature at about $t\\approx 0.9\\,\\hbar/J$ and a linear growth of $-\\ln L_N$ with $N$ that yields $(\\dim\\mathcal{H}_{\\mathrm{eff}})^{1/L}\\approx 2.3$ per site in the ergodic regime and about 1.5 versus 1.1 for two states in the fragmented regime. This matters because the full Loschmidt echo decays exponentially in system size, while the subsystem echo is robust and already informative for small blocks.","feed_headline":"A local echo reveals the size of the accessible Hilbert space","feed_subtitle":"In cold-atom snapshots, the subsystem Loschmidt echo measures entropy and Hilbert-space fragmentation directly.","key_machinery":"The central object is the subsystem Loschmidt echo, defined as the spatial average of local projectors onto the initial occupation string, $L_N(t)=\\frac{1}{L-N+1}\\sum_{i=1}^{L-N+1}\\langle\\prod_{j=i}^{i+N-1}\\hat P_j\\rangle_t$. The short-time mechanism is the expansion of $L_N$ into connected density correlation functions, with each $n$-point term carrying irreducible information; the sharpening of the rate function as $N$ grows comes from higher-order terms acting at the critical time. The long-time mechanism is the identity $-\\ln L_N = S_N(\\beta,\\mu) - \\tfrac{1}{2}\\beta^2\\delta E^2 \\le S_N(\\beta,\\mu)$, which converts a string probability into a thermodynamic entropy, plus the thermodynamic-limit relation $(\\dim\\mathcal{H}_{\\mathrm{eff}})^{1/L}=\\exp\\!\\left(\\frac{d(-\\ln L_N)}{dN}\\right)$, which turns the measured slope into a Hilbert-space dimension per site.","core_discovery":"On the paper's own terms, the subsystem Loschmidt echo is defined as $L_N(t)=\\frac{1}{L-N+1}\\sum_{i=1}^{L-N+1}\\langle\\prod_{j=i}^{i+N-1}\\hat P_j\\rangle_t$, the spatial average of local projectors onto the initial occupation string. The central discovery is that this observable has two complementary readings. After a quench from a charge-density-wave state in the hard-core Bose-Hubbard regime ($U/J\\approx 25$, $\\Delta=0$), the rate function $-\\ln L_N/N$ develops a progressively sharper kink near $t_c\\approx 0.9\\,\\hbar/J$; expanding $L_N$ in connected density correlations shows that genuine three- and four-point correlations are what shape the kink. In the long-time regime ($t\\approx 230\\!-\\!280\\,\\hbar/J$), the time-averaged value obeys $-\\ln L_N = S_N(\\beta,\\mu) - \\tfrac{1}{2}\\beta^2\\delta E^2 \\le S_N(\\beta,\\mu)$, so its slope with $N$ becomes $(\\dim\\mathcal{H}_{\\mathrm{eff}})^{1/L}=\\exp\\!\\left(\\frac{d(-\\ln L_N)}{dN}\\right)$. Measured values are $2.29(1)$ and $2.34(1)$ for domain-wall and CDW initial states in the ergodic regime, versus $1.54(1)$ and $1.12(1)$ in the fragmented regime, with the difference directly attributed to different Hilbert-space fragments.","pith_inferences":["Editorial inference: because the extraction needs only site-resolved snapshots, the same protocol could be used to produce space- and time-resolved entropy maps, for example to watch a thermalization front move through a system or to detect localized regions in a many-body localized phase.","Editorial inference: the correction term in the long-time identity is a built-in thermometer; with an independent estimate of the energy variance, the difference between $-\\ln L_N$ and the fitted thermodynamic entropy would calibrate an effective temperature for states that do not reach the infinite-temperature limit.","Editorial inference: the fact that high-order connected correlations dominate the kink suggests that string observables of length comparable to the correlation length could serve as sensitive detectors of rare many-body correlations in quenches beyond the charge-density-wave case studied here."],"forward_implications":["If the central claim is right, any quantum simulator with site-resolved readout can measure dynamical quantum phase transitions through the subsystem Loschmidt echo without facing the exponentially small signal of the full Loschmidt echo.","In the same run, the slope of $-\\ln L_N$ versus $N$ gives a quantitative, model-free estimate of the accessible Hilbert-space dimension per site in the thermodynamic limit.","The differing slopes for domain-wall and CDW states in the fragmented regime are direct evidence that those states live in different uncoupled fragments, i.e. ergodicity breaking by kinetic constraints.","The robustness of the subsystem Loschmidt echo against particle loss makes entropy and Hilbert-space-dimension measurements feasible in systems where the full return probability would vanish from any lost atom."],"supporting_citations":[{"why":"introduces the subsystem Loschmidt echo and shows its rate function develops the DQPT kink; this is the theoretical anchor for the short-time measurement.","marker":"[30]"},{"why":"provides the criterion that the subsystem Loschmidt echo sharpens with increasing subsystem size; used to interpret the measured kink sharpening.","marker":"[31]"},{"why":"defines dynamical quantum phase transitions through return-probability non-analyticities, the target phenomenon of the short-time study.","marker":"[36]"},{"why":"supplies the rate-function formalism used to define the dynamical free energy density and extract the critical time.","marker":"[37]"},{"why":"furnishes the Hilbert-space fragmentation framework in which the long-time dimension measurement is interpreted.","marker":"[35]"},{"why":"provides a foundational model of Hilbert-space fragmentation, the phenomenon the long-time measurement claims to observe.","marker":"[58]"}],"fun_headline_variants":["Subsystem Loschmidt echoes expose Hilbert-space fragmentation","Local echo measures quantum many-body dynamics in cold atoms","A quasi-local echo probes quantum many-body physics","Measuring ergodicity breaking with subsystem Loschmidt echoes"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that by about 230 to 280 $\\hbar/J$ the system has reached quasi-equilibrium inside its accessible fragment, so the measured $-\\ln L_N$ equals the equilibrium thermodynamic entropy; if it has not, the slope measures transient relaxation instead.","fun_headline_variants_meta":{"raw":{"variants":["Subsystem Loschmidt echoes expose Hilbert-space fragmentation","Local echo measures quantum many-body dynamics in cold atoms","A quasi-local echo probes quantum many-body physics","Measuring ergodicity breaking with subsystem Loschmidt echoes"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000236,"raw_usage":{"total_tokens":1583,"prompt_tokens":1102,"completion_tokens":481,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":718,"completion_tokens_details":{"reasoning_tokens":427}},"tokens_in":718,"tokens_out":481,"duration_ms":4428,"temperature":1.0,"reasoning_tokens":427,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T05:15:35.850788+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-measure $-\\ln L_N$ in later time windows, for example at 400, 600, and 1000 $\\hbar/J$, for the same initial states; if the extracted slope changes or drifts monotonically, the quasi-equilibrium premise fails. Alternatively, fix $\\beta$ from the measured density distribution and check the long-time identity with an independent estimate of the energy variance $\\delta E^2$.","supporting_citations":[{"cited_title":"Bandyopadhyay, A","cited_arxiv_id":null,"evidence_quote":"introduces the subsystem Loschmidt echo and shows its rate function develops the DQPT kink; this is the theoretical anchor for the short-time measurement."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"provides the criterion that the subsystem Loschmidt echo sharpens with increasing subsystem size; used to interpret the measured kink sharpening."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"defines dynamical quantum phase transitions through return-probability non-analyticities, the target phenomenon of the short-time study."}],"review_version":1}