{"id":"e1a6f38e-c483-46ed-898a-0e9238b609f6","arxiv_id":"2412.12768","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A single simulated Gaussian quantum trajectory of a multi-mode optical parametric oscillator network samples Ising spin configurations with Boltzmann statistics, with the effective temperature set by pump strength.","lead":"Researchers simulated an all-optical network of light modes whose quantum noise makes it visit low-energy spin patterns more often, following a Boltzmann law. If confirmed, such a device could sample hard optimization problems at optical speeds without electronic feedback.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Energy-histogram fits do not establish the Boltzmann claim: the paper never tests configuration-level probabilities, and for the degenerate K graph the stated PB(E) formula omits the multiplicity factor n(E).","rationale":"The reader's conditional verdict targeted the Gaussian truncation. That is a legitimate risk, and the paper itself defers non-Gaussian checks. But the more load-bearing gap is closer to the central claim: even granting the Gaussian model exactly, the evidence offered for 'Boltzmann sampling' is an energy histogram. A Boltzmann law is a statement about full spin configurations; it implies, among other things, that all configurations with the same Ising energy are equiprobable. The authors never demonstrate that. For the random binary K graph, degeneracies are severe, so an exponential P(E) can arise from non-Boltzmann distributions over configurations. The paper explicitly notes n(E) is non-uniform, which makes the stated formula PB(E)=e^{-βE}/Z suspect: if the red line is a pure exponential, it is not the Boltzmann energy distribution for a degenerate spectrum. This is not a question of external consensus or internal formal error in the SDEs; it is a mismatch between the measured observable and the claimed conclusion. It can be settled by a simple reanalysis of the same simulated data, which is why the appropriate disposition is conditional, not rejection. If the configuration-level test passes, the central claim would have real support; if it fails, the emergent equilibrium is an artifact of binning. The paper is otherwise careful: the Gaussian trajectory framework is standard, the derivations of the moment equations and threshold are clearly presented, and the figures are illustrative. My concern is strictly about the sufficiency of the evidence for the headline assertion.","tokens_in":13032,"tokens_out":11187,"duration_ms":112327,"concrete_test":"Using the same Gaussian SDEs (SM Eqs. S.4–S.6) and parameters as in Fig. 3, simulate or reuse the stored trajectory data. Compute the empirical frequency of each of the 2^N spin configurations. (1) For every pair σ,σ' with E(σ)=E(σ'), test whether P(σ) and P(σ') agree within sampling error; violation refutes Boltzmann. (2) Fit P(σ) ∝ exp(−β E(σ)) by maximum likelihood and compare the fitted β and goodness-of-fit with the Fig. 3 energy-histogram fit. (3) Estimate the autocorrelation time of E(σ(t)) and compare P(E) from the first and second halves of the trajectory; large discrepancies indicate non-ergodicity. If configuration-level probabilities match Boltzmann, the concern is resolved.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central assertion is that a single Gaussian quantum trajectory samples spin configurations as P(σ) ∝ exp(−β E_Ising(σ)). The only numerical evidence in Figs. 3, S.1, and S.2 is a fit to the aggregated energy histogram P(E). That evidence cannot certify the claim. For any distribution depending only on energy, P(E)=n(E) f(E); an exponential P(E) is compatible with many non-Boltzmann choices of P(σ) within degenerate energy shells. For the K graph, the paper itself states that n(E) is strongly non-uniform, yet the quoted fit PB(E)=e^{−βE}/Z omits n(E). Either the red curve is not the Boltzmann prediction for the K graph, or the text's formula is inaccurate. The paper never reports P(σ) for individual spin configurations, never checks equal probability of same-energy configurations, and never tests detailed balance, stationarity, or single-trajectory ergodicity (e.g., autocorrelation times or multiple seeds). Without a configuration-level Boltzmann test, the observed exponential energy histogram may simply reflect the sign statistics of correlated Gaussian quadratures under the Gaussian trajectory approximation, not thermal equilibrium of the Ising Hamiltonian.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript studies N degenerate optical modes driven by a two-photon pump and coupled through non-local dissipative jump operators that encode an Ising coupling matrix J. Using Gaussian quantum trajectories (SM Eqs. S.4-S.6), the authors define an instantaneous Ising spin from sign(Re[alpha_i(t)]), compute the Ising energy, and construct the time-marginal distribution P(E) from a single long trajectory (t_max = 20,000/gamma, 200,000 samples). They report that P(E) is exponential in E for both SK and K graphs at several pump strengths, extract an effective temperature by fitting, and find that T_eff decreases with pump above threshold. The paper interprets this as emergent thermal equilibrium and proposes the system as an ultrafast all-optical Boltzmann sampler.","tokens_in":13316,"tokens_out":10006,"duration_ms":96852,"significance":"If the configuration-level Boltzmann claim survives scrutiny, the result would be significant: a single quantum trajectory of a dissipative all-optical network would autonomously sample from an Ising distribution with the temperature set by the pump, without feedback. The paper is transparent about the model (jump operators, Gaussian-moment closure), and the SM provides explicit evolution equations, which is a strength. The SK energy histograms are also nontrivial evidence because the two-fold degeneracy is uniform, so an exponential P(E) carries genuine information about the configuration weights. However, the evidence is not yet at the level of the central claim: the K-graph prediction is misstated, and stationarity, ergodicity, fitting quality, and the validity of the Gaussian truncation are not established.","major_comments":[{"comment":"The central claim is a Boltzmann distribution over configurations, P(sigma) proportional to exp(-E(sigma)/(k_B T_eff)), but the only numerical evidence shown is the energy histogram P(E). For the K graph, the text states that the energy multiplicity n(E) is strongly non-uniform; therefore the correct Boltzmann prediction is P_B(E) = n(E) exp(-E/(k_B T_eff))/Z, not the formula P_B(E) = exp(-E/(k_B T_eff))/Z that is plotted. As written, the comparison in the right panel of Fig. 3 is not a test of the Boltzmann hypothesis. Please report configuration-level probabilities, or at minimum P(E)/n(E), and test whether configurations inside a degenerate energy shell are equally probable.","section":"Fig. 3, K graph, and accompanying text"},{"comment":"Equilibrium is asserted without stationarity or ergodicity diagnostics. The paper does not report whether the first and second halves of the trajectory give the same histogram, does not give autocorrelation or mixing times, and does not use multiple independent trajectories; with 200,000 samples from a single trajectory, the effective number of independent samples may be much smaller if the spins flip slowly. Please add these diagnostics and error bars based on trajectory-to-trajectory variability.","section":"Figs. 3-4 and text on emergent thermal equilibrium"},{"comment":"Because T_eff is fitted from the same P(E) that is used to claim equilibrium, the exponential fit has a slope and a normalization as free parameters and cannot by itself validate the Boltzmann form. Please report residuals or goodness-of-fit metrics, compare the fit against alternative functional forms (e.g., a quadratic in E), and provide an independent consistency check, such as T_eff extracted from a different observable or from a split-half analysis of the same trajectory.","section":"Fig. 4 and the effective-temperature fitting procedure"},{"comment":"All numerical results use the Gaussian moment closure for a system with two-photon loss, and the manuscript explicitly defers non-Gaussian noise to future work. Because the central claim concerns the statistics generated by quantum noise, this closure is load-bearing: without a benchmark against an exact master-equation solution for small N, or a quantitative validity criterion, the observed exponential energy statistics could be a truncation artifact. Please add such a benchmark or clearly restrict the claim to the validity regime of the Gaussian approximation.","section":"SM Eqs. (S.4)-(S.6) and Conclusions"}],"minor_comments":[{"comment":"The first line has a malformed bracket, 'langle[H, O rangle]dt'; it should be langle[H, O]rangle dt.","section":"Eq. (2)"},{"comment":"The heat maps lack a colorbar label and axis labels; please state explicitly that the color scale is Re(alpha_i) and indicate the time-axis units.","section":"Fig. 2"},{"comment":"The terms involving sum_j sqrt(eta) appear to have unbalanced parentheses; please re-check the equation.","section":"SM Eq. (S.5)"},{"comment":"The statement that for the SK graph 'all different spin configurations are associated to different energies' followed by 'n(E)=2' should be clarified: the factor 2 comes from the global sigma -> -sigma symmetry.","section":"Text after Fig. 3"},{"comment":"The sentence 'Similar results have however been checked for larger values of N' should be supported by a figure or at least a table in the SM.","section":"Main text, larger-N claim"},{"comment":"The bibliography entry contains the phrase 'published on 2024/12/02'; please use a standard citation format.","section":"Reference [33]"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this paper shows something potentially interesting—a single Gaussian quantum trajectory of an all-optical, feedback-free OPO network appears to visit low-energy Ising configurations with statistical weights that look thermal. But the evidence is a fit to energy histograms, not a test of the Boltzmann distribution over spin configurations. The claim is not established.\n\nWhat's new: the combination of the Gaussian trajectory formalism with the spatial all-optical Ising machine of Strinati et al., and the observation that the effective temperature is set by pump strength relative to threshold. The equations in the SM are a useful starting point for anyone modeling such systems.\n\nWhat's good: the explicit encoding of the Ising coupling in dissipators, the phase-sign mapping, and the fact that they show the SK graph energy histogram does behave exponentially. For SK, where the energy multiplicity is uniform, an exponential P(E) is at least consistent with a Boltzmann law in energy.\n\nThe soft spots are real and mostly rest on the absence of configuration-level evidence. P(σ) ∝ exp(-βE(σ)) is a statement about individual spin configurations, but the paper never checks it. For the K graph, the text states n(E) is strongly non-uniform, yet the quoted fit PB(E)=e^{-βE}/Z omits n(E). That's either a missed factor or a telling inconsistency. They also fit T_eff from the same histogram they use to claim equilibrium, which adds circularity. No error bars, no multiple seeds, no autocorrelation or mixing diagnostics, and no benchmark against exact master-equation or non-Gaussian corrections. The paper itself defers non-Gaussian effects to future work, which is honest but makes the claim provisional.\n\nThat said, I don't see a load-bearing internal contradiction. The exponential energy distribution, especially for SK, is a nontrivial observation even if it doesn't prove the full Boltzmann claim. The paper is not a fabrication; it's a possibly correct idea backed by too-weak evidence.\n\nWho's this for? People working on photonic Ising machines and open quantum systems will want to know about it. But I wouldn't cite it yet, and I wouldn't bring it to a reading group as a settled result.\n\nRecommendation: send it to peer review, but the referee should demand configuration-level probability tests, proper handling of degeneracies, multiple trajectories, and a discussion of Gaussian truncation errors. The paper has a testable core claim; making the evidence match the claim would turn it from plausible to convincing.","headline":"A plausible but under-evidenced claim of emergent Boltzmann sampling in an all-optical parametric network; worth a referee but not yet a citation.","tokens_in":13803,"tokens_out":2769,"would_cite":false,"duration_ms":25675,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["42.50.-p","42.65.Yj","05.50.+q"],"model":"deepseek-v4-flash","headline":"A single Gaussian quantum trajectory of a multimode two-photon-driven optical system samples spin configurations according to a Boltzmann distribution of the encoded Ising Hamiltonian, with effective temperature set by pump strength.","keywords":["Ising machine","quantum trajectory","Gaussian approximation","Boltzmann sampling","optical parametric oscillator","dissipative coupling","emergent thermal equilibrium","combinatorial optimization"],"falsifier":"Evolve the same few-mode system with an exact quantum master equation (or with non-Gaussian trajectory corrections) at the parameters of Figs. 3-4 and compare the long-time spin-sign distribution to P(E) ∝ $e^{{−E/k_B T_eff}}$; a measurable departure for N as small as 4-6 would break the claim. Alternatively, in an experiment, verify detailed balance in the inferred spin transitions.","tokens_in":12833,"feed_emoji":"🎲","tokens_out":7781,"duration_ms":63143,"temperature":0.7,"pith_summary":"The paper claims that a single Gaussian quantum trajectory of a multimode optical system can act as an ultra-fast Boltzmann sampler for Ising problems. In the proposed all-optical machine, N modes are driven by two-photon pumping and coupled only through non-local loss operators that encode the coupling matrix J_ij; at each instant the sign of Re α_i defines an Ising spin. Long-time statistics of these spin configurations follow the Boltzmann law P(E) ∝ $e^{{−E/k_B T_eff}}$, with effective temperature set by the pump strength relative to threshold. This means the ground state becomes the most probable outcome without any electronic feedback, and lower temperatures (and stronger ground-state bias) are reached simply by raising the pump. The result is supported numerically for both spin-glass and binary random graphs.","feed_headline":"Quantum noise alone turns one trajectory into a Boltzmann sampler","feed_subtitle":"All-optical Ising machine samples low-energy states fast, with pump setting the temperature.","key_machinery":"The Gaussian quantum trajectory formalism: the state is carried by first moments α_j=⟨a_j⟩ and second moments u_nm, v_nm, evolving under heterodyne unravelling. The Ising interaction enters through non-local jump operators Γ̂^{(3)}_{i,j} = \\sqrt{|J_{ij}|}(\\hat a_i − (J_{ij}/|J_{ij}|)\\hat a_j), combined with one-photon losses; these dissipators, not a Hamiltonian, encode the spin couplings. The threshold pump is obtained from linear stability of the mean-field equations, G_th=(γ−λ_max)/2, and the spin state is read out as sign(Re α_i).","core_discovery":"Emergent thermal equilibrium in a driven-dissipative all-optical Ising machine: for each single heterodyne trajectory solved at the Gaussian level, the time-marginal distribution of the spin configuration σ_i(t)=sign(Re α_i(t)) is Boltzmann-distributed with the encoded Ising energy, P(E) ∝ $e^{{−E/k_B T_eff}}$. The effective temperature decreases monotonically as the pump G rises above the threshold G_th=(γ−λ_max)/2, and the thermal law persists even in the limit G→0^+. The non-local jump operators Γ̂^{(3)}_{i,j}=\\sqrt{|J_{ij}|}(\\hat a_i − J_{ij}/|J_{ij}| \\hat a_j) together with one-photon losses implement the spin-spin interaction purely dissipatively, with no Hamiltonian coupling between modes.","pith_inferences":["If the thermal law persists under non-Gaussian corrections, the same dissipative encoding could be ported to other driven-dissipative platforms (e.g., microwave or exciton-polariton lattices) to obtain fast Boltzmann sampling.","The fitted T_eff(G) curve and the energy crossing point of the distributions suggest a calibration procedure: measure P(E) at two pump values and infer the machine's effective temperature, then use it to tune the sampling bias.","A direct consequence not tested in the paper: the spin dynamics should satisfy detailed balance; measuring transition rates between spin configurations in the trajectory would provide a stricter test of equilibration than the energy histogram.","For practical optimization, the relevant figure is the time-to-solution to reach the ground state with target probability; the paper's single-trajectory statistics imply this time scales with 1/P(E_gs), offering a concrete benchmark for hardware."],"forward_implications":["The most probable sampled configuration after long time is the Ising ground state, with an exponential advantage over higher-energy states.","Increasing the pump above threshold lowers the effective temperature, sharpening the Boltzmann distribution toward low-energy solutions without changing the coupling matrix.","Because encoding is purely dissipative and readout is instantaneous, sampling is limited by optical timescales rather than feedback electronics.","Boltzmann statistics survive at vanishing pump, so thermal sampling does not depend on operating above the oscillation threshold.","A single trajectory's time statistics are sufficient; no ensemble of devices or repeated fresh-state runs are needed."],"supporting_citations":[{"why":"supplies the Gaussian trajectory method by which the stochastic moment equations are derived and solved.","marker":"[22]"},{"why":"describes the all-optical spatial coherent Ising machine whose architecture realizes the dissipative non-local coupling.","marker":"[14]"},{"why":"earlier demonstration that a DOPO network samples from a thermal distribution, the baseline extended here to single all-optical trajectories.","marker":"[19]"},{"why":"shows noise-injected analog Ising machines can perform ultrafast statistical sampling, motivating the all-optical feedback-free design.","marker":"[20]"},{"why":"defines the Sherrington-Kirkpatrick spin-glass coupling matrix used as the main test graph.","marker":"[28]"},{"why":"provides the threshold condition G_th=(γ−λ_max)/2 used to normalize the pump and identify the oscillation threshold.","marker":"[30]"},{"why":"established emergent equilibrium in many-body optical bistability, the conceptual precedent for thermalization in driven-dissipative optical systems.","marker":"[32]"},{"why":"recent evidence of Boltzmann statistics in linear optical systems at vanishing drive, supporting the G→0^+ regime reported here.","marker":"[33]"},{"why":"gives the two-photon driving Hamiltonian used in the model.","marker":"[24]"}],"fun_headline_variants":["One trajectory, thermal Ising law: all-optical Boltzmann sampler","Pump strength sets temperature for ultra-fast Ising sampling","Gaussian noise alone gives thermal equilibrium in Ising machine","Dissipative coupling turns single light path into Boltzmann sampler"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"All numerical evidence comes from Gaussian (first- and second-moment) trajectory equations; if non-Gaussian quantum fluctuations change the sign statistics of Re[α_i], the Boltzmann result may not survive.","fun_headline_variants_meta":{"raw":{"variants":["One trajectory, thermal Ising law: all-optical Boltzmann sampler","Pump strength sets temperature for ultra-fast Ising sampling","Gaussian noise alone gives thermal equilibrium in Ising machine","Dissipative coupling turns single light path into Boltzmann sampler"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000784,"raw_usage":{"total_tokens":3400,"prompt_tokens":822,"completion_tokens":2578,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":438,"completion_tokens_details":{"reasoning_tokens":2509}},"tokens_in":438,"tokens_out":2578,"duration_ms":20155,"temperature":1.0,"reasoning_tokens":2509,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T13:46:05.875158+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Evolve the same few-mode system with an exact quantum master equation (or with non-Gaussian trajectory corrections) at the parameters of Figs. 3-4 and compare the long-time spin-sign distribution to P(E) ∝ $e^{{−E/k_B T_eff}}$; a measurable departure for N as small as 4-6 would break the claim. Alternatively, in an experiment, verify detailed balance in the inferred spin transitions.","supporting_citations":[{"cited_title":"Calvanese Strinati, D","cited_arxiv_id":null,"evidence_quote":"describes the all-optical spatial coherent Ising machine whose architecture realizes the dissipative non-local coupling."},{"cited_title":"B¨ ohm, D","cited_arxiv_id":null,"evidence_quote":"shows noise-injected analog Ising machines can perform ultrafast statistical sampling, motivating the all-optical feedback-free design."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"recent evidence of Boltzmann statistics in linear optical systems at vanishing drive, supporting the G→0^+ regime reported here."},{"cited_title":"Kinsler and P","cited_arxiv_id":null,"evidence_quote":"gives the two-photon driving Hamiltonian used in the model."}],"review_version":1}