{"id":"03c63211-283f-4a68-8b36-5cb9809e93c4","arxiv_id":"2411.14304","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Tuning the correlation exponent of long-range disordered cavity frequencies switches atomic spontaneous emission from non-Markovian to near-Markovian decay, tracking the environment's localization-delocalization transition.","lead":"An atom coupled to a chain of cavities with correlated disorder changes its decay from non-Markovian to nearly Markovian as the disorder correlation exponent crosses about 2. The result suggests cavity arrays with tunable correlated disorder can act as controllable reservoirs and that atomic decay can serve as a probe of the environment's phase.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Markovian regime claim for α>2 rests on ensemble-averaged N; single-realization non-Markovianity is not characterized, so the probe interpretation is not yet supported.","rationale":"The reader's weakest assumption already identifies the ensemble-averaged, finite-time nature of N as the main vulnerability, and I agree that this is the most load-bearing concern. The central claim—that α>2 marks a Markovian regime and that the atom's evolution reflects the localization-delocalization transition—requires that a typical single realization behave in a Markovian way. The paper demonstrates this only for one representative sample (Fig. 3(b)) and for the ensemble average (Fig. 4). Without a distribution of single-realization N, the small average could be an artifact of mixing a few strongly non-Markovian realizations with many Markovian ones. This matters practically because a single atom in a single disordered sample would then not reliably indicate the phase. I considered other potential issues: the formulas in Eqs. (12) and (14) appear to have transcription errors (they fail the r→0 limit as printed, likely missing factors of 1/r and r/Δ), but the underlying NV expressions from Ref. [49] are probably correct and do not affect the full-Hamiltonian results. The disorder-generation variable L in Eq. (4) is undefined, but this is likely a typo for N and does not change the numerics. The finite-time bound-state contribution is a genuine secondary concern, but the more fundamental question is whether the ensemble average hides per-realization non-Markovianity. The proposed concrete test—computing the histogram of N over realizations—would settle this directly. Because the reader's verdict is already CONDITIONAL, and my concern reinforces exactly the condition that should be added, the verdict remains unchanged.","tokens_in":13486,"tokens_out":22544,"duration_ms":194805,"concrete_test":"For α=3 (and, if feasible, α=2.5 and α=4), compute the non-Markovianity N_i for each of the 10^3 individual disorder realizations using the same parameters as Fig. 4 (g=0.1J, ωa=0, N=6201, tJ=600). Plot the histogram of N_i and report the median, the 90th percentile, and the fraction of realizations with N_i > 0.1. Additionally, for a subset of realizations, extend the time window to tJ=2400 and check whether N_i changes significantly. If the median N_i is near zero and the tail is negligible, the Markovian-regime claim holds for typical single realizations; if a substantial fraction of realizations have N_i > 0.1, the averaged N is not representative of a single atom, and the central claim must be weakened or qualified.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim is that for α>2 the atomic dynamics becomes Markovian, as evidenced by the ensemble-averaged excitation and the averaged non-Markovianity N (Fig. 4). However, N is computed per realization and then averaged over 10^3 disorder realizations. Non-Markovianity is a property of a single realization's reduced dynamics, not of a disorder average. The authors themselves note that for any finite realization the band is offset from ω=0, so ωa=0 may lie near localized states at the mobility edge for a subset of realizations. Such realizations will exhibit strong non-Markovian behavior even when α>2. The single-realization example in Fig. 3(b) is only one typical sample; the paper does not provide the distribution of N over realizations. Consequently, the observed small average N could arise from a mixture of mostly Markovian realizations and a minority of strongly non-Markovian ones, rather than from a uniform Markovian regime. This weakens the conclusion that 'the evolution of the atom reflects the localization-delocalization transition' and the proposed use of the atom as a probe of the environment phase: a single atom in a given realization would not reliably see a Markovian reservoir unless the per-realization N distribution is concentrated near zero. The finite-time window (tJ≤600) is a secondary issue: for α≥2 bound states produce long-time oscillations that may further increase N at longer times, so the identification of a Markovian regime requires demonstrating that N has converged.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript studies spontaneous emission of a two-level atom placed at the center of a one-dimensional coupled-cavity array with on-site disorder generated by a fractional Brownian motion with power-law spectrum k^{-alpha}. The authors compute the atomic excitation probability and the non-Markovianity measure N of Eq. (6) for ensembles of 10^3 disorder realizations and show that increasing alpha from 0 to larger values turns the decay from population trapping toward an approximately exponential decay. They associate the transition between alpha=1 and alpha=2 with the localization-delocalization transition of the 1D Anderson model with long-range correlated disorder, propose that the atom can be used as a probe of the environment phase, and present two effective models (Markovian bath plus auxiliary mode, and Lorentzian bath) that reproduce the dependence of N on alpha using the participation ratio and the spectral density at the band center.","tokens_in":13810,"tokens_out":9778,"duration_ms":86454,"significance":"If the central claim is correct, the paper would provide a simple, tunable platform in which a localization-delocalization transition in the environment is mirrored by the Markovian/non-Markovian character of a single emitter, with potential applications in reservoir engineering and quantum probing. The numerical methods are solid: large arrays (N=6201), long time windows (tJ<=600), small time step, norm conservation, and averaging over 10^3 realizations are appropriate for the ensemble-averaged quantities. The effective models are analytically transparent and give a useful phenomenological benchmark. However, the significance is moderated by the fact that the central 'Markovian regime' claim is made for ensemble averages rather than for the single-realization dynamics that a probe atom would actually experience, and by the lack of a quantitative convergence analysis in time and system size.","major_comments":[{"comment":"The central claim that a Markovian regime is reached for alpha>2 is based on the ensemble-averaged non-Markovianity N and the ensemble-averaged excitation pe. As the paper notes in the discussion of Fig. 3(a), each realization has a band that is offset from omega=0, so localized modes can appear near omega_a=0 for some realizations even when alpha>2. Because N is defined for a single realization's reduced dynamics, a small average over 10^3 realizations could be produced by a mixture of mostly Markovian realizations and a minority of strongly non-Markovian ones. The text reports that N is 'numerically evaluated for each realization and averaged afterwards,' but no distribution, histogram, or typical/outlier statistics are shown. Please provide the per-realization distribution of N (e.g., percentiles or the fraction of realizations with N below a threshold) and, if the probe interpretation in the introduction is retained, demonstrate that a single realization reliably yields near-Markovian dynamics for alpha>2.","section":"III A, Fig. 4"},{"comment":"The finite simulation window tJ<=600 and the finite chain size N=6201 do not by themselves establish an asymptotic Markovian regime. The paper acknowledges that for alpha>=2 bound-state contributions persist at long times, that more than (but not all) 90% of the excitation is released, and that N=0 only for an infinite homogeneous CCA. The observed small N could therefore be a finite-time/finite-size effect rather than a genuine Markovian limit. To support the phrase 'Markovian regime is reached for alpha>2', the manuscript should either define an operational criterion (e.g., N below a specified threshold that is stable under increasing N and t_max) or study the long-time limit more directly, including an estimate of pe(infinity) and its dependence on N and alpha.","section":"III A, Figs. 2 and 4"},{"comment":"The effective-model curves are not fitted to the direct N data, but they depend on two choices that are not derived from the microscopic model: the unit proportionality constant in g_l = g/sqrt(xi) and the spectral bin width 0.1J used to define gamma(omega_a=0). Since both choices enter r=gamma/g_l, and r fully determines the model predictions N(r), the reported 'remarkable agreement' in Fig. 4 could be partly by construction. Please show the sensitivity of the model curves to these choices (e.g., varying the bin width and the proportionality constant over a reasonable range), or derive the constant and bin width from the Hamiltonian. In addition, Eqs. (12) and (14) are derived under exact resonance, while the actual mode frequency omega_l is detuned from omega_a=0; the text asserts this does not matter 'on average' but does not quantify the effect.","section":"III B, Fig. 4"}],"minor_comments":[{"comment":"The quantity eNV is never explicitly defined; please give its definition as eNV = -∫_{∂t pe<0} (d pe^2/dt) dt. Also, the simplification eNV = NV + 1 assumes pe(infinity)=0, which is not strictly valid for a finite disordered CCA; please state this assumption or use the exact expression eNV = NV + 1 - pe(infinity)^2.","section":"II C, Eq. (6)"},{"comment":"The symbol Δ is carried over from Eq. (12), where Δ = sqrt(16-r^2), but the Lorentzian model's NV formula is stated to be valid for r∈[0,2), which suggests a different Δ (likely sqrt(4-r^2)). Please define Δ separately for each model or use distinct notation.","section":"III B, Eq. (14)"},{"comment":"'Symbols are fittings originated from the effective models' is misleading; the symbols are evaluations of the analytic formulas with r(alpha) as input, not fits to the direct N data. Please rephrase.","section":"Fig. 4 caption"},{"comment":"The sentence 'We refer to the reader the reviews in Refs.' should read 'We refer the reader to the reviews in Refs.'; similarly, 'our goal here resumes to find' should be 'our goal here reduces to finding'.","section":"II C, p. 3"}],"recommendation":"major_revision","confidential_remarks":"The numerical work is standard and the qualitative trend is clear, but the central conceptual claim overreaches the ensemble-averaged finite-time data. The requested per-realization distribution of N and a time/size convergence study are feasible within the scope of the manuscript and should be obtainable without changing the core methodology. No concerns about citation practices or the journal fit beyond the need to temper the 'Markovian regime' language."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a competent numerical study of a known model, and the central trend—increasing α at fixed disorder strength suppresses non-Markovianity—is credible. The paper is honest about the finite-N non-Markovianity and the bound-state leftovers. What's actually new is applying the de Moura–Lyra correlated-disorder model to the CCA spontaneous-emission setup and showing that the correlation exponent, not the disorder strength, controls the transition. The two effective models (Markovian bath plus auxiliary mode, and Lorentzian bath) reproduce the averaged N quite well, and the numerics are careful: large arrays, norm preservation, ensemble sizes.\n\nSoft spots are where the conclusions outrun the evidence. The Markovian-regime claim for α>2 is based on the ensemble-averaged N. Non-Markovianity is a property of a single realization's dynamics; a minority of realizations with band offsets near the mobility edge could still be strongly non-Markovian while the average looks Markovian. The paper acknowledges this indirectly (band offset, finite N) but does not provide the distribution of N or per-realization examples for α>2. So the 'atom as a probe of the environment phase' statement is not yet supported—a single atom in a given realization would not reliably see a Markovian reservoir. The finite time window (tJ ≤ 600) is a genuine secondary issue: bound states produce long-time oscillations that could increase N at later times. The effective-model formulas assume exact resonance and pe(∞)=0, which the authors note, but the match to the averaged N is then partly by construction.\n\nNone of this is a load-bearing error for the qualitative trend. It does mean the sharp transition at α=2 and the probe interpretation need extra support. The paper would benefit from a histogram of N across realizations, a convergence check for N as t_max grows, and a softer wording in the Introduction/Conclusions.\n\nOverall, I'd send this to a serious referee. The reservation is about overclaiming, not about invalid work. Read it if you work on structured reservoirs or correlated disorder; the two effective models are the most reusable part.","headline":"Correlated disorder exponent α plausibly tunes spontaneous emission toward Markovian decay in a coupled-cavity array, but the ensemble-averaged non-Markovianity measure does not yet support the single-shot probe claim.","tokens_in":14342,"tokens_out":2357,"would_cite":false,"duration_ms":22044,"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":"By tuning the correlation exponent of on-site disorder in a coupled-cavity array, spontaneous emission switches from non-Markovian population trapping to nearly exponential decay, with the crossover between $\\alpha=1$ and $\\alpha=2$.","keywords":["open quantum systems","non-Markovian dynamics","spontaneous emission","coupled-cavity array","correlated disorder","Anderson localization","localization-delocalization transition","quantum probing"],"falsifier":"Compute the ensemble-averaged $N$ and $p_e(t)$ for $\\alpha=2.5$ using a longer time window (for example $tJ>2000$) and larger arrays; if the distribution of $p_e(t)$ develops revivals or if $N$ does not stay near its short-time value as the window grows, the Markovian classification would fail. Likewise, a single typical realization with $\\omega_a$ at the band center could be monitored: if revivals recur after the apparent exponential decay, the flat-spectral-density argument is incomplete.","tokens_in":13334,"feed_emoji":"⚛️","tokens_out":6610,"duration_ms":58321,"temperature":0.7,"pith_summary":"The paper studies a two-level atom coupled to an array of coupled cavities whose on-site frequencies carry long-range correlated disorder, and asks whether the atom's spontaneous emission can be steered between memory-laden and memoryless decay by changing only the correlation exponent $\\alpha$. It shows that raising $\\alpha$ at fixed disorder strength progressively removes the population trapping caused by Anderson localization, and that the decay becomes nearly exponential once $\\alpha$ exceeds 2. That crossover coincides with the known localization-delocalization transition of the cavity modes, where delocalized states create a flat spectral density near the band center. A sympathetic reader should care because it turns a disordered photonic environment into a tunable reservoir and makes the qubit's decay curve a witness of the environment's phase.","feed_headline":"Raising disorder correlations flips atomic decay to Markovian","feed_subtitle":"A qubit's decay switches from trapped oscillations to near-exponential as the disorder correlation exponent passes 2.","key_machinery":"The load-bearing object is the localization-delocalization transition of the one-dimensional Anderson model with long-range correlated disorder, generated by the fractional-Brownian-motion series $\\epsilon_n \\propto \\sum_k k^{-\\alpha/2}\\cos(2\\pi nk/L+\\varphi_k)$. In this model, $\\alpha=2$ marks the appearance of delocalized states with mobility edges, and the spectral density $G(\\omega)=g^2(\\omega)\\rho(\\omega)$ becomes flat in the center of the band; a flat spectral density is what produces Markovian exponential decay in the weak-coupling regime. The non-Markovianity is quantified by $N=N_V/|\\tilde N_V|$, built from the positive and negative slopes of $p_e^2(t)$, and two effective models - an emitter coupled to a Markovian bath plus one auxiliary mode, and an emitter in a Lorentzian bath - reproduce $N(\\alpha)$ from a single ratio $r=\\gamma/g_\\ell$ that combines the local decay rate with the localization length via the participation ratio.","core_discovery":"The central claim is that the degree of memory in the atomic decay is controlled by $\\alpha$, the exponent of the power-law spectrum $k^{-\\alpha}$ of the on-site disorder. For uncorrelated disorder ($\\alpha=0$) the field modes are localized and the atomic population is trapped in oscillatory non-Markovian dynamics; as $\\alpha$ increases toward 2 and beyond, the modes delocalize around the band center, the spectral density there flattens, and the atom releases more than 90% of its excitation in a nearly exponential curve close to the homogeneous-chain benchmark $p_e(t)=e^{-g^2t/J}$. The authors establish the Markovian regime for $\\alpha>2$, while noting that $N=0$ strictly holds only for an infinite homogeneous chain and that small bound-state remnants survive.","pith_inferences":["Because the Markovian label refers to an ensemble-averaged, finite-time measure, a longer-time or single-realization experiment could show residual recurrences from bound states; the sharp $\\alpha=2$ transition is therefore an operational boundary, not an exact equality.","The same probe logic could be tested in other settings with tunable localization transitions, such as quasiperiodic or flat-band lattices, where the spectral density near the probe frequency can be flattened by delocalization.","A direct superconducting-circuit realization - one transmon coupled to a chain of resonators with engineered on-site disorder - could map $N$ versus $\\alpha$ in real time and test whether the crossover region between 1 and 2 is as narrow as the simulations suggest."],"forward_implications":["For $\\alpha>2$, the disordered coupled-cavity array acts as a memoryless reservoir: an initially excited atom decays nearly exponentially, releasing more than 90% of its excitation, despite unaltered disorder strength.","The same decay curve can be read as a probe of the environment's phase: the onset of Markovianity between $\\alpha=1$ and $\\alpha=2$ marks the localization-delocalization transition of the field modes.","The non-Markovianity $N(\\alpha)$ is captured by simple effective models whose only material input is the ratio $r=\\gamma/g_\\ell$ computed from the free-field spectrum, so the environment's phase is encoded in one scalar parameter.","Keeping the atomic frequency at the band center is the condition for Markovian behavior; away from the center, localized modes and band-edge states reintroduce memory effects and trapping."],"supporting_citations":[{"why":"Supplies the correlated-disorder model and the localization-delocalization transition at $\\alpha=2$ that the paper ties to Markovian decay.","marker":"[57]"},{"why":"Provides the prior result that Anderson localization induces quantum non-Markovianity in the same coupled-cavity setup, plus the effective dissipative model used here.","marker":"[49]"},{"why":"Gives the homogeneous-CCA Markovian benchmark $p_e(t)=e^{-g^2t/J}$ that the paper compares against for large $\\alpha$.","marker":"[19]"},{"why":"Supplies the structured-reservoir spectral-density framework and the Lorentzian-bath solution used in the second effective model.","marker":"[46]"},{"why":"Provides the geometric non-Markovianity measure underlying $N=N_V/|\\tilde N_V|$.","marker":"[61]"},{"why":"Offers the Markovian-bath-plus-auxiliary-mode model in a related context, used here as one phenomenological description.","marker":"[62]"},{"why":"Supports the sublinear participation ratio for $1<\\alpha<2$, used to locate the onset of delocalization.","marker":"[67]"}],"fun_headline_variants":["Tune disorder correlations to flip atomic decay from memory-rich to Markovian","Alpha > 2 turns atomic decay exponential, erasing memory","Qubit decay switches from trapped to exponential via disorder correlations","Correlated disorder controls Markovianity of spontaneous emission","Alpha tunes qubit decay: from non-Markovian oscillations to pure exponential"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The Markovian-regime claim for $\\alpha>2$ rests on the assumption that the ensemble-averaged non-Markovianity computed over a finite time window ($tJ\\le 600$, $10^3$ realizations) represents the asymptotic dynamics, even though the cavity array has a discrete spectrum and bound-state contributions keep the excitation from ever fully decaying.","fun_headline_variants_meta":{"raw":{"variants":["Tune disorder correlations to flip atomic decay from memory-rich to Markovian","Alpha > 2 turns atomic decay exponential, erasing memory","Qubit decay switches from trapped to exponential via disorder correlations","Correlated disorder controls Markovianity of spontaneous emission","Alpha tunes qubit decay: from non-Markovian oscillations to pure exponential"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000679,"raw_usage":{"total_tokens":3075,"prompt_tokens":926,"completion_tokens":2149,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":542,"completion_tokens_details":{"reasoning_tokens":2041}},"tokens_in":542,"tokens_out":2149,"duration_ms":13172,"temperature":1.0,"reasoning_tokens":2041,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T15:19:44.426558+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the ensemble-averaged $N$ and $p_e(t)$ for $\\alpha=2.5$ using a longer time window (for example $tJ>2000$) and larger arrays; if the distribution of $p_e(t)$ develops revivals or if $N$ does not stay near its short-time value as the window grows, the Markovian classification would fail. Likewise, a single typical realization with $\\omega_a$ at the band center could be monitored: if revivals recur after the apparent exponential decay, the flat-spectral-density argument is incomplete.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the correlated-disorder model and the localization-delocalization transition at $\\alpha=2$ that the paper ties to Markovian decay."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the prior result that Anderson localization induces quantum non-Markovianity in the same coupled-cavity setup, plus the effective dissipative model used here."},{"cited_title":"Lombardo, F","cited_arxiv_id":null,"evidence_quote":"Gives the homogeneous-CCA Markovian benchmark $p_e(t)=e^{-g^2t/J}$ that the paper compares against for large $\\alpha$."},{"cited_title":"Lambropoulos, G","cited_arxiv_id":null,"evidence_quote":"Supplies the structured-reservoir spectral-density framework and the Lorentzian-bath solution used in the second effective model."},{"cited_title":"Mouloudakis and P","cited_arxiv_id":null,"evidence_quote":"Offers the Markovian-bath-plus-auxiliary-mode model in a related context, used here as one phenomenological description."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supports the sublinear participation ratio for $1<\\alpha<2$, used to locate the onset of delocalization."}],"review_version":1}