{"id":"29f77e75-6c9c-4aab-8317-851f083f84a3","arxiv_id":"2507.22132","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Cold cesium atoms with measurement-based feedback reproduce the dynamics of the Lipkin-Meshkov-Glick and kicked-top spin models, including symmetry breaking, chaotic orbits, and a time-crystal-like phase.","lead":"This dissertation reports an experiment that uses careful laser measurements and rapid electronic feedback on a cloud of cold cesium atoms to mimic the behavior of theoretical spinning magnet models. The work is a step toward using tabletop atomic systems to study quantum chaos, phase transitions, and time crystals without a full quantum computer.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Latency of ατ_delay≈0.23 violates the paper's own smallness criterion, so the observed LMG decay and order-parameter deficits may be feedback artifacts rather than faithful emulation.","rationale":"The reader's weakest assumption overlaps with this concern but is broader, including mean-field linearization and QND fidelity. I focus on latency because the paper itself identifies it as the primary driver of dynamical decay and provides a numerical model rather than a controlled experimental test. The abstract's LMG emulation claim is most directly threatened by a 23% control error: the derivative condition Eq. 6.5 is violated, so the feedback signal is systematically stale. Fig. 6.8 makes the mechanism plausible, but plausibility is not calibration. The QKT results are less affected by latency, yet the time-crystal rigidity window is already narrower than simulation, and the Lyapunov estimates are explicitly lower bounds from five points; both reinforce the conditional verdict. I do not see an internal inconsistency or a clear refutation, and the fixed-point/s=2/3 agreement provides positive evidence. The absence of raw data and error bars makes quantitative verification impossible from the manuscript, but that is a verification concern rather than by itself a flaw in the central argument. A reduced-α experiment is the cleanest discriminator between faithful emulation and delayed-feedback artifact.","tokens_in":46745,"tokens_out":5725,"duration_ms":74698,"concrete_test":"Run the LMG emulation at a reduced linear rate, for example α = 2π×1.25 kHz with k increased to keep s fixed, so that ατ_delay drops from 0.23 to ≈0.05. If latency is the cause, the anomalous dynamical decay in Fig. 6.5 should largely disappear and the Czz∞ deficit in Fig. 6.6 should shrink; if decay persists, the latency model is falsified and the emulation claim needs re-evaluation. As a complementary check, fit the semi-classical simulator used in Fig. 6.8 to the shot-resolved tDD distribution of Fig. 6.4 with τ_delay as a free parameter and compare the best fit to the independently measured 6 µs latency.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 6.1.3 establishes the criterion ατ_delay ≪ 1 for the LMG control loop (Eq. 6.5), then reports ατ_delay ≈ 0.23, which is not small. The paper attributes the otherwise unexpected dissipative decay to this latency, and Fig. 6.8 shows that simulated latency reproduces decay qualitatively. However, no quantitative fit of the latency/noise model to the experimental time series in Figs. 6.1, 6.5, or 6.7 is presented, and the predicted dependence on α was not tested experimentally. The same chapter admits that bias/probe misalignment corrupts the Jz measurement and therefore the feedback strength, and that the time-crystal rigidity window is roughly three times narrower than theory. Because the central claim is that feedback-controlled dynamics emulate the target Hamiltonians, a regime in which the control error is a first-order (23%) effect leaves open that the reported symmetry breaking and phase-transition signatures are properties of a delayed dissipative feedback loop rather than of the LMG Hamiltonian. This is a load-bearing concern, not a refutation: the fixed-point locations and the s=2/3 crossing are consistent with LMG, so the method may still work, but the current data do not quantitatively establish it.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper reports an experimental implementation of measurement-based feedback emulation of nonlinear collective-spin Hamiltonians in a cold atomic ensemble of cesium atoms. The protocol uses Faraday QND measurements of Jz and an FPGA feedback loop whose control voltage is proportional to the measured Jz, replacing the quadratic Jz^2 terms of the target Hamiltonians by their mean-field form. The reported results are: LMG symmetry breaking at an unstable fixed point for s > 0.5 with final trajectories near predicted fixed points; LMG order parameters Z_infinity and C_zz^infinity showing a transition near s = 2/3; kicked-top chaos characterized by Lyapunov-exponent estimates from only the first five stroboscopic steps; and a Floquet time-crystal phase whose rigidity window is roughly one third of the simulated width. The manuscript is candid that the energy-conserving LMG evolution is dissipative in the data, that this decay is attributed to control-loop latency with alpha*tau_delay ≈ 0.23, and that a later-discovered probe/bias misalignment corrupts the Jz measurement.","tokens_in":46980,"tokens_out":9250,"duration_ms":115187,"significance":"Demonstrating programmable emulation of nonlinear spin Hamiltonians in a mesoscopic atomic spin would be a useful addition to quantum-simulation toolkits, especially because the atom number can in principle be varied into the quantum regime. The manuscript's strengths are its new experimental data, its use of externally well-studied target models (LMG, kicked top), the consistency of fixed-point locations and the s=2/3 crossing with mean-field theory, and its unusually frank treatment of failure modes such as latency, measurement transients, and probe/bias misalignment. The semiclassical simulations and detailed apparatus documentation are also valuable. The claim is not yet quantitatively established: the operating point violates the paper's own small-latency criterion by an order-one margin, no quantitative fit of the latency/noise model to the data is presented, and the Lyapunov and time-crystal analyses rest on very short or indirectly related observables. As it stands, the paper is a promising experimental demonstration with qualitative agreement, not a closed verification that the emulated dynamics are those of the target Hamiltonians.","major_comments":[{"comment":"The paper's own latency criterion is violated by the operating point. Equation (6.5) requires alpha*tau_delay ≪ 1, but the reported alpha = 2π × 6.25 kHz and tau_delay ≈ 6 µs give alpha*tau_delay ≈ 0.23. Section 6.1.3 identifies this same latency as the primary driver of the dissipative decay toward fixed points seen in Figs. 6.1 and 6.5, and Fig. 6.8 provides only a qualitative semiclassical simulation. Because the control error is a first-order effect, the symmetry-breaking and phase-transition signatures in Figs. 6.1, 6.5, and 6.6–6.7 could, for all the paper currently shows, be properties of a delayed, dissipative feedback loop rather than of the LMG Hamiltonian. The manuscript should fit the latency/noise model to the experimental time series, report residuals and uncertainties, and either test the predicted dependence on alpha or explicitly show that the observed decay is quantitatively reproduced for multiple operating points.","section":"§6.1.3, Eq. (6.5)"},{"comment":"The Lyapunov-exponent estimator is not validated and is too indirect to support the chaos claim as it stands. Equation (6.9) equates the standard deviation of the measured elevation angle across experimental shots with an exponentially growing separation of neighboring trajectories, then fits this quantity using only the first five stroboscopic steps. Since only Jz is measured and the azimuthal degree of freedom is unobserved, the estimator is at best a heuristic lower bound. No synthetic-data test is given for the same step count, measurement axis, and noise level, so the reported values cannot be distinguished from projection effects or finite-ensemble fluctuations. The text itself notes that standard time-series methods require several hundred steps, which further indicates that this estimator needs separate numerical validation before the agreement shown in Fig. 6.11(b) can be assessed.","section":"§6.2.1, Eq. (6.9)"},{"comment":"The time-crystal rigidity result is a factor of three narrower than the simulation and is not modeled quantitatively. The experiment shows subharmonic rigidity only out to approximately (1 ± 0.07)π in alpha, whereas Fig. 6.13(a) shows the simulated phase extending to roughly (1 ± 0.2)π; Chapter 7 attributes the discrepancy to latency combined with signal transients and also admits that a tenth-of-a-degree bias/probe misalignment corrupts Jz and therefore the feedback strength. Because the time-crystal phase is one of the headline results of the abstract, the manuscript should either include a quantitative error model that reproduces the observed narrowing, or explicitly present this as a qualitative observation and weaken the corresponding claim. A quantitative statement of how the misalignment biases Jz and the feedback gain is also needed, since it affects all of the LMG and QKT results.","section":"§6.2.2 and Chapter 7"}],"minor_comments":[{"comment":"The x-axis is labeled 'Time [s]' although the data span about 1.5 ms; the units appear to be microseconds, not seconds.","section":"Fig. 6.3"},{"comment":"The summary says the Floquet time-crystal state was observed with 'α = π/2', but Section 6.2.2 states that the FTC phase occurs for α nominally equal to π; one of these statements is a typo.","section":"Chapter 7"},{"comment":"The collective spin is denoted Fz in Eqs. (2.4)–(2.5) and Jz elsewhere; the relationship between these notations should be defined once.","section":"§2.1, §2.4"},{"comment":"There are several typographical errors (e.g., 'feedforw ard', 'Pad` e', 'inital') that should be corrected in a journal version.","section":"Throughout"},{"comment":"A reproducibility statement with raw data and analysis code would strengthen the experimental claims substantially, since the LabVIEW block diagrams alone do not allow quantitative reanalysis.","section":"Chapter 6"}],"recommendation":"major_revision","confidential_remarks":"To the editor: this is an arXiv preprint of a PhD dissertation, and the level of apparatus detail is appropriate for a thesis but excessive for a research article; the authors should be encouraged to restructure it for a journal. The theoretical comparison curves in Figs. 6.6 and 6.11–6.13 come substantially from the group's own earlier papers (Refs. [13] and [59]), so an independent numerical implementation of the target models would strengthen the verification. The absence of a data/code release is another reason the quantitative comparisons should be judged carefully."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First thing to know: this thesis reports the first experimental implementation of the QMF protocol on a collective atomic spin, and the headline phenomena—the LMG symmetry-breaking bifurcation and the s=2/3 dynamical phase transition—show up in the data. The Lyapunov-exponent estimates for the kicked top are rough but in the right range. That is a genuine new capability for AMO quantum simulation.\n\nWhat the paper does well: it gives a full description of the apparatus, including state purification, light-shift cancellation, the FPGA feedback controller, and the control-law synthesis. I also give credit for honesty: the author states plainly that the dynamics are dissipative, that the latency figure ατ_delay ≈ 0.23 violates his own smallness criterion, and that a bias/probe misalignment corrupts the Jz measurement. The numerical study in Sec. 6.1.3 shows that latency, not shot noise, can produce the observed decay.\n\nThe soft spot is exactly that latency. The central claim is that the feedback-controlled spin emulates a conservative Hamiltonian, but the measured control error is a first-order effect (23%). The dissipation is attributed to latency, but no quantitative fit of the latency/noise model to the experimental time series is presented, and the predicted dependence on α was not tested. So one cannot yet tell whether the symmetry-breaking and order-parameter signatures are properties of the LMG Hamiltonian or of a delayed dissipative feedback loop. The fixed-point locations and the s=2/3 crossing are consistent with LMG, so this is a load-bearing but not fatal concern. The time-crystal rigidity window is about three times narrower than theory, which the author also leaves unexplained. No data or code are provided, and most figures lack error bars.\n\nIn short: the paper is a promising first demonstration, not a finished quantitative one. It is worth a serious referee, and I would send it to peer review with the request that the latency model be fitted to the data, the α-dependence be measured, and error bars and data be included. It is a bit long for a normal article, but the thesis format is fine for an arXiv preprint; a journal version would need to be condensed.\n\nI'd bring it to reading group: there's a good discussion to be had about what counts as emulation when the control loop is imperfect.","headline":"First experimental QMF emulation of LMG and kicked-top dynamics, with a real latency problem that leaves the demonstration qualitatively right but quantitatively unproven.","tokens_in":47493,"tokens_out":2374,"would_cite":true,"duration_ms":29258,"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":"A cold atomic spin, measured continuously and steered by real-time feedback, can be made to follow the dynamics of target nonlinear spin Hamiltonians such as the Lipkin–Meshkov–Glick model and the kicked top.","keywords":["quantum measurement and feedback","collective atomic spin","Lipkin-Meshkov-Glick model","kicked top","Faraday rotation QND measurement","quantum chaos","Floquet time crystal","mean-field emulation"],"falsifier":"Repeat the LMG emulation at a tenfold smaller linear rotation rate (so that $\\alpha\\tau_{\\mathrm{delay}} \\approx 0.02$) with $s$ fixed; if the spin still decays toward the fixed points with an unchanged settling-time scaling, the claim that loop latency is the primary driver of the dissipative decay is falsified. As a positive control, add artificial delays of roughly 3, 6, and 12 $\\mu$s to the loop and check that the signed midpoint decay time $t_{DD}$ shortens monotonically with delay, as the paper's own latency simulations predict.","tokens_in":46537,"feed_emoji":"⚛️","tokens_out":14327,"duration_ms":156019,"temperature":0.7,"pith_summary":"This dissertation reports an experimental method for making an ensemble of roughly a million cold cesium atoms evolve as if it were governed by a chosen nonlinear spin Hamiltonian. The method repeatedly measures the collective spin projection $J_z$ without destroying the state and feeds each measurement outcome forward to set the strength of a magnetic rotation, replacing the nonlinear $J_z^2$ term with the product of the measured value and $J_z$. The author demonstrates the approach on two benchmark models: the Lipkin–Meshkov–Glick (LMG) Hamiltonian, where the data show spontaneous symmetry breaking and a dynamical phase transition at the predicted parameter values, and the Quantum Kicked Top, where the data show chaotic trajectories with Lyapunov exponents of the expected order of magnitude and a Floquet time-crystal phase with partial rigidity. If the method is sound, a single apparatus can be reprogrammed to study phase transitions, chaos, and out-of-equilibrium phases in collective spin systems, with atom number acting as a dial between the quantum and classical regimes. The paper is candid about its main shortcoming: roughly $6\\,\\mu\\mathrm{s}$ of control-loop latency spoils the continuous-time LMG emulation, quantified by $\\alpha\\tau_{\\mathrm{delay}} \\approx 0.23$, producing a dissipative decay toward the fixed points that a Hamiltonian evolution would not have.","feed_headline":"An atomic spin emulates symmetry breaking, chaos, time crystals","feed_subtitle":"Measurement-based feedback makes a million cold cesium atoms follow two nonlinear spin Hamiltonians.","key_machinery":"The load-bearing object is the mean-field linearization of the nonlinear term, $J_z^2 \\to \\langle J_z\\rangle J_z$, which turns a two-body interaction into a single-body rotation whose strength is updated by each measurement. The physical realization is a three-part loop: the Faraday interaction, in which the collective spin projection rotates the polarization of a probe laser by an angle proportional to $\\langle J_z\\rangle$ (a quantum non-demolition measurement); a field-programmable gate array (FPGA) controller sampling at 500 kHz that digitizes the polarimetry signal and computes the control voltage $V_{\\mathrm{ctl}} = g_{\\mathrm{ctl}}^{-1}\\left(s\\Lambda \\langle J_z\\rangle/\\langle J\\rangle\\right)$; and magnetic coils that execute the resulting rotation about $\\hat{z}$ while a constant rotation about $\\hat{x}$ runs simultaneously. Supporting machinery keeps the loop faithful within fixed-point hardware: an exponential-decay model of the mean spin length, a Padé approximant for the decay factor, and, for the kicked top, a binary bit-mask operation that implements the kick-angle modulo $2\\pi$ so that arbitrarily large nonlinear strengths can be reached.","core_discovery":"The central claim is that quantum measurement and feedback (QMF) with a mean-field control law makes a collective atomic spin follow the classical (mean-field) dynamics of a target Hamiltonian. Concretely, a QND Faraday measurement yields an outcome $m \\propto \\langle J_z\\rangle$, and the feedback law sets the $z$-rotation rate proportional to $m$, realizing the linearized Hamiltonian $H_{\\mathrm{MF}} = -\\gamma(1-s)J_x - \\gamma \\frac{s}{2}\\frac{m}{\\langle J\\rangle}J_z$. For the LMG model this produces the expected bifurcation of the stable fixed point for $s>0.5$: initialized at the unstable point $+\\hat{x}$, the spin falls randomly into one of the two symmetry-broken wells, with the choice correlated with the first measurement outcome, and the time-averaged magnetization places the dynamical phase transition at the expected $s=2/3$. For the kicked top, the same protocol run stroboscopically yields trajectory spreads consistent with positive Lyapunov exponents in the chaotic regime and, near $\\alpha=\\pi$, subharmonic oscillations with a rigidity window that is real but about three times narrower than simulation predicts. The paper argues that the deviations from ideal Hamiltonian evolution—chiefly a non-Hamiltonian decay of the spin toward the model's fixed points in the LMG case and a narrowed rigidity window in the kicked top—are dominated by control-loop latency and measurement transients rather than by quantum projection noise or classical control noise.","pith_inferences":["The latency-induced decay the paper identifies could be turned into a feature: by deliberately choosing the loop delay, the same apparatus could emulate controlled dissipative or open-system dynamics (for instance Lindblad-type evolution toward fixed points), which the paper does not pursue.","Because the symmetry-breaking direction is a function of the first measurement outcome, the setup offers a quantitative probe of the quantum-to-classical transition: at smaller atom number the initial spread in $Z(0)$ grows, so the distribution of decay times $t_{DD}$ should broaden and the correlation between $Z(0)$ and the final well should weaken—a directly testable prediction.","The single-projection readout caps the Lyapunov estimate; adding a second probe direction or a stroboscopic tomography step after each kick would remove the lower-bound ambiguity and should bring the experimental Lyapunov exponents closer to the classical spectrum.","The discrepancy between the simulated and observed time-crystal rigidity windows (about $0.8\\pi$ versus $0.93\\pi$) provides a single-number benchmark for the combined effect of latency, transients, and the small bias–probe misalignment, so future implementations can use the measured rigidity width to diagnose control fidelity."],"forward_implications":["A single apparatus can be reprogrammed to emulate any Hamiltonian whose nonlinear terms are polynomial in $J_z$, simply by changing the feedback control law, without hardware changes.","The LMG experiments give an experimental handle on spontaneous symmetry breaking and dynamical phase transitions: the order parameter $Z_\\infty$ crosses at the expected $s=2/3$, and the shot-to-shot symmetry-breaking choice is correlated with the first measurement outcome.","The kicked-top emulation provides a route to studying quantum chaos in a controlled mesoscopic system; the measured Lyapunov exponents, estimated from the growth of the spread in the elevation angle, sit below the classical values as expected for a lower bound from a single measured projection.","The Floquet time-crystal data show subharmonic (period-$2T$) oscillations with finite rigidity around $\\alpha=\\pi$, indicating the emulator can realize out-of-equilibrium phases of matter on demand.","Scaling the atom number from $10^6$ downward should slide the same experiment from the classical to the quantum regime, since the signal scales as $N$ while quantum projection noise scales as $\\sqrt{N}$."],"supporting_citations":[{"why":"introduces quantum measurement and feedback with weak measurements, the general protocol that this work implements on a collective spin.","marker":"[7]"},{"why":"applies QMF to collective-spin systems and provides the numerical demonstration that simulated quantum trajectories track positive-Lyapunov regions, the theoretical basis for this experiment.","marker":"[8]"},{"why":"defines the kicked-top model and its classical and quantum chaos, the discrete-time target of the emulation.","marker":"[9]"},{"why":"defines the LMG Hamiltonian whose phase transition and order parameters are the continuous-time target of the emulation.","marker":"[10]"},{"why":"derives the kicked-top Floquet time-crystal phase diagram and rigidity criterion that the time-crystal experiment is compared against.","marker":"[13]"},{"why":"supplies the three-dimensional probe–atom coupling model and the effective atom-number weighting used to calibrate the Faraday measurement and the quantum projection noise.","marker":"[32]"},{"why":"supplies the master-equation and mean-field predictions for the LMG order parameters $Z_\\infty$ and $C_{zz}$ against which the phase-transition data are compared.","marker":"[59]"}],"fun_headline_variants":["Feedback-controlled atomic spin acts like nonlinear quantum models","Atomic ensemble mimics spin Hamiltonians via measurement feedback","Closed-loop atomic spin emulates chaos, time crystals","Measurement feedback drives atomic spin to mimic LMG and kicked top","Cold atoms controlled to emulate symmetry breaking and chaos"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that at roughly $10^6$ atoms the measured spin projection can stand in for the operator in the nonlinear term ($J_z^2 \\to \\langle J_z\\rangle J_z$) and that the feedback loop applies the corrected rotation before the state changes appreciably—a condition the paper's own estimate $\\alpha\\tau_{\\mathrm{delay}} \\approx 0.23$ shows is only marginally met.","fun_headline_variants_meta":{"raw":{"variants":["Feedback-controlled atomic spin acts like nonlinear quantum models","Atomic ensemble mimics spin Hamiltonians via measurement feedback","Closed-loop atomic spin emulates chaos, time crystals","Measurement feedback drives atomic spin to mimic LMG and kicked top","Cold atoms controlled to emulate symmetry breaking and chaos"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000566,"raw_usage":{"total_tokens":2766,"prompt_tokens":1112,"completion_tokens":1654,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":728,"completion_tokens_details":{"reasoning_tokens":1579}},"tokens_in":728,"tokens_out":1654,"duration_ms":14475,"temperature":1.0,"reasoning_tokens":1579,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T12:01:35.407930+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Repeat the LMG emulation at a tenfold smaller linear rotation rate (so that $\\alpha\\tau_{\\mathrm{delay}} \\approx 0.02$) with $s$ fixed; if the spin still decays toward the fixed points with an unchanged settling-time scaling, the claim that loop latency is the primary driver of the dissipative decay is falsified. As a positive control, add artificial delays of roughly 3, 6, and 12 $\\mu$s to the loop and check that the signed midpoint decay time $t_{DD}$ shortens monotonically with delay, as the paper's own latency simulations predict.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"introduces quantum measurement and feedback with weak measurements, the general protocol that this work implements on a collective spin."},{"cited_title":"Mu˜ noz Arias, Pablo M","cited_arxiv_id":null,"evidence_quote":"applies QMF to collective-spin systems and provides the numerical demonstration that simulated quantum trajectories track positive-Lyapunov regions, the theoretical basis for this experiment."},{"cited_title":"Scharf F","cited_arxiv_id":null,"evidence_quote":"defines the kicked-top model and its classical and quantum chaos, the discrete-time target of the emulation."},{"cited_title":"Lipkin, N","cited_arxiv_id":null,"evidence_quote":"defines the LMG Hamiltonian whose phase transition and order parameters are the continuous-time target of the emulation."},{"cited_title":"Mu˜ noz Arias, Karthik Chinni, and Pablo M","cited_arxiv_id":null,"evidence_quote":"derives the kicked-top Floquet time-crystal phase diagram and rigidity criterion that the time-crystal experiment is compared against."},{"cited_title":"PhD thesis, University of Arizona, 2015","cited_arxiv_id":null,"evidence_quote":"supplies the three-dimensional probe–atom coupling model and the effective atom-number weighting used to calibrate the Faraday measurement and the quantum projection noise."},{"cited_title":"Mu˜ noz Arias, Ivan H","cited_arxiv_id":null,"evidence_quote":"supplies the master-equation and mean-field predictions for the LMG order parameters $Z_\\infty$ and $C_{zz}$ against which the phase-transition data are compared."}],"review_version":1}