{"id":"308cbbfe-074a-4a87-995b-12bcf2463454","arxiv_id":"2412.00536","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"In a noisy cyclic quantum walk, the eigenstate participation ratio correlates with spreading: below a numerically located noise strength near pi/3 the walker spreads, above it the walker localizes.","lead":"This paper studies how random static phase errors affect a quantum walker moving around a cyclic graph, and finds that a simple spectral measure called the participation ratio can signal where the walker will localize. It identifies a numerical crossover at a noise strength of about pi/3, below which the walker spreads and above which it tends to stay put.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The participation ratio is computed from the same full diagonalization of the noisy step operator that would already yield the dynamics; the claimed efficient predictive diagnostic is therefore not established.","rationale":"The paper has a clean formal skeleton: the noiseless spectrum of S_N is derived in closed form, the circuit implementation is a useful addition, and the crossover at phi_s = pi/3 is a plausible numerical observation. My concern is not that the authors are dishonest; it is that the central claim, that the eigenstate participation ratio is an efficient predictive diagnostic, is not established by the evidence presented. Because PR is computed from the eigenstates of the same noisy unitary that generates the dynamics, the PR-beta correlation is an in-sample relation between two outputs of one diagonalization. To support the efficiency claim, the paper would need either a way to estimate PR without full diagonalization, or a demonstration that PR predicts beta for unseen realizations and noise levels better than simply knowing phi_s. The ambiguous definition of Delta x^2 in Eq. (13) compounds the problem, making the reported beta values irreproducible as written. These issues are fixable with a reproducibility package and a clarified observable, so the conditional verdict remains appropriate rather than a rejection.","tokens_in":10682,"tokens_out":9441,"duration_ms":115739,"concrete_test":"Re-implement the walk-on-the-line analysis using the standard circular mean squared displacement Delta x_s^2(m) = sum_s min(|s - s_0|, N - |s - s_0|)^2 P(s,m), with a pre-registered fit window (e.g., m = 1..N/2), for 100 independent disorder realizations at each noise level phi_s, N = 128, and the Hadamard coin. Then compare (a) PR from full diagonalization, (b) beta from the power-law fit, and (c) a cheap predictor such as the noiseless PR or the noise amplitude alone. If beta's crossover shifts away from phi_s = pi/3, or if the PR-beta correlation disappears on held-out realizations, then the efficient spectral-diagnostic claim fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that the eigenstate participation ratio (PR) is an efficient spectral diagnostic that anticipates localization. But PR is evaluated on the noisy step operator S_noise = D_noise S_N (Sec. IV), so computing it requires diagonalizing the full 2N-by-2N noisy unitary. That same diagonalization suffices to compute the complete time evolution U^m|psi_0> for any m. Thus PR is not an alternative to full dynamical simulations: it is a post-processing of the same spectral data, and the reported correlation between PR and the spreading exponent beta is an in-sample consistency between two functionals of the same eigenvectors. The paper provides no independent way to estimate PR from the noise amplitude and the noiseless spectrum, no runtime comparison, and no held-out realization validation. The predictive content is further under-specified: beta is extracted from an unspecified 'initial steps' window, and Eq. (13) as written sums a site-independent Delta x(m), which would make Delta x^2(m) vanish for symmetric delocalized states and contradict Fig. 5. Without code or data, the central claim rests on correlational numerics whose defining observable is ambiguous.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies discrete-time quantum walks on a cyclic graph with a three-parameter unitary coin, adding static diagonal phase noise at the sites. It derives the noiseless spectral structure of the step operator, finding two spectral bands controlled by one coin parameter and degeneracies controlled by the half-sum of the two phases. The paper then divides the noisy dynamics into a walk-on-the-line regime and a walk-on-the-cycle regime, extracts a spreading exponent from the mean squared displacement in the former, and uses a coefficient-of-variation convergence criterion in the latter. The central claim is that the eigenstate participation ratio of the step operator is a computationally efficient static diagnostic that anticipates localization in both regimes, with a crossover near noise strength phi_s = pi/3.","tokens_in":10886,"tokens_out":4228,"duration_ms":44820,"significance":"If established, the participation-ratio diagnostic could be practically useful for predicting localization without long-time simulations, and the noiseless spectral analysis is a clean contribution: the paper explicitly shows how the coin parameters control spectral bands, gaps, and degeneracies, and it provides explicit eigenstates and eigenvalues of the noiseless walk. The main numerical claims, however, rest on limited simulations: two coin families, essentially one graph size in the walk-on-the-line regime, unspecified fit windows, and no reported disorder-realization statistics. The advertised computational-efficiency argument is also not supported as written because the participation ratio is computed from the same noisy step-operator diagonalization that would already provide the full dynamics. The crossover at phi_s = pi/3 is an interesting falsifiable numerical finding, but it needs a much more detailed statistical and finite-size analysis before it can be regarded as a general result.","major_comments":[{"comment":"Equation (13) is not a mean squared displacement as written. The quantity Dx(m) defined in Eq. (14) is the distance between the mean position x(m) and the initial site, independent of the summation index s, so summing [Dx(m)]^2 P(s,m) over s merely gives [Dx(m)]^2. For the symmetric coin starting at a localized site, the mean position remains at the initial site and this quantity is identically zero, which contradicts the nonzero curves in Fig. 5. Since the spreading exponent beta in Eq. (15) is the central dynamical observable, please replace Eq. (13) with a proper cycle-distance mean squared displacement, such as sum_s d(s,s0)^2 P(s,m), or clearly define the estimator being used.","section":"Sec. IV, Eqs. (13)-(15)"},{"comment":"The claim that the participation ratio is a computationally efficient alternative to full dynamical simulations is not supported by the computations described. The participation ratio in Sec. V is evaluated on eigenstates of the noisy step operator S_noise = D_noise S_N from Eq. (12), which requires diagonalizing the full 2N-by-2N unitary; the same diagonalization provides arbitrary powers U^m and therefore the complete dynamics. No runtime comparison, no independent estimator of the participation ratio from the noise amplitude and noiseless spectrum, and no held-out or out-of-sample validation are provided. Please either benchmark the diagnostic against direct simulation, provide a way to obtain it without full diagonalization, or restate the conclusion as a correlational static indicator rather than an alternative to dynamical simulation.","section":"Secs. IV-V and Sec. VIII"},{"comment":"The numerical extraction of beta is under-specified. The text says the power-law fit is restricted to 'the initial steps where the system resembles a quantum walk on a line,' but no concrete fit window is given. The figures show mean squared displacement curves and participation-ratio boxplots without stating the number of independent disorder realizations, the graph sizes used in the line-regime analysis, or any error bars or confidence intervals on the reported beta values. Because the crossover at phi_s = pi/3 is identified numerically from beta, this missing statistical information is load-bearing for the paper's central claim.","section":"Sec. V, Figs. 4-6"},{"comment":"The claimed robustness of the crossover and of the participation-ratio correlation is not demonstrated. In the walk-on-the-line regime, results are shown for N=128 and two coin parameter sets, while the text states the trend 'appears robust for graphs with diverse size N' without presenting that data. The walk-on-the-cycle regime likewise shows saturation levels for several N, but the connection between the participation ratio and the saturation level is discussed qualitatively rather than quantified. Please provide the finite-size scaling and parameter scans, or explicitly limit the claim to the reported cases.","section":"Secs. V-VI"}],"minor_comments":[{"comment":"The phase gate definition contains a typo: it should be P_j = |0><0| + e^{i pi/2^j}|1><1|, not |0><0|+|0><1|+|1><0|+e^{i pi/2^j}|1><1|.","section":"Sec. VII, Eq. (17) and Fig. 11"},{"comment":"The eigenvalue expression for c_j is garbled in the typesetting, with a missing radicand; please re-typeset this equation so the square root structure is clear.","section":"Sec. II, Eq. (2)"},{"comment":"There are minor language issues: 'finite-size effects become dominate' should be 'become dominant,' and 'the effects of the cyclic topology are null' should be 'are zero' or 'vanish.'","section":"Secs. I and IV"},{"comment":"The convergence criterion CV <= 0.01 and the moving window size n = max{10, N/4} are introduced without a sensitivity analysis; a short discussion of how the saturation levels depend on these choices would strengthen the walk-on-the-cycle analysis.","section":"Sec. VI"},{"comment":"The paper states that data may be obtained upon reasonable request, but no code or data are provided. Given the centrality of the numerical crossover and the PR-dynamics correlation, a supplementary package with simulation scripts and fit details would substantially aid reproducibility.","section":"Data availability"}],"recommendation":"major_revision","confidential_remarks":"The noiseless spectral derivation in Sec. II is solid and self-contained, and the paper contains an interesting numerical observation about a crossover near phi_s = pi/3. My main concern is that the load-bearing claims are currently supported by correlational numerics with unspecified fit windows and no statistical detail, and the 'efficient alternative to dynamics' framing conflicts with the fact that the participation ratio is computed from the same diagonalization that yields the dynamics. These issues are fixable within the manuscript's scope, so I recommend major revision rather than rejection. I would also encourage the authors to make their simulation and fitting code available, since several of my concerns could be resolved by direct inspection of the scripts."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, the noiseless spectral section is the real payload: closed-form eigenvalues of the three-parameter coin on a cycle, with band structure set by gamma and degeneracy by the half-sum of phases, and it checks out. Second, the numerical case for the participation ratio as a cheap localization diagnostic is not established yet, and there is a concrete error in the MSD definition that needs fixing.\n\nWhat is good: The derivation in Sec. II is clean and self-contained. The distinction between degenerate (sinusoidal) and non-degenerate (flat) eigenstate probability distributions is nice. Using the participation ratio to anticipate localization is a sensible idea, and the observed crossover near phi_s = pi/3 across two coins and several N is the kind of thing worth reporting if it survives scrutiny. The circuit implementation is standard but not harmful.\n\nWhere it softens: Eq. (13) defines MSD as a sum over sites of [Delta x(m)]^2 P(s,m), but Delta x(m) in Eq. (14) is a single number depending only on m, the distance of the mean position. That makes the sum equal to [Delta x(m)]^2, which for a symmetric walk is always zero. The figures show non-zero MSD, so either the notation is wrong or the observable is misdefined. This needs to be corrected before the numerics can be interpreted. The fit window for beta is described only as 'initial steps' with no quantitative criterion; there are no error bars or realization counts; and the code/data are not available. That matters because the crossover is found numerically.\n\nThe efficiency claim also needs a softer framing. The PR is computed from the full diagonalization of the noisy step operator, and that same diagonalization gives you the full time evolution. So PR is not an alternative to simulating dynamics; it is a functional of the same spectral data. If the authors want to claim efficiency, they need to show how to estimate PR without full diagonalization, or compare runtimes. As it stands, the correlation between PR and beta is in-sample consistency between two functionals of the same eigenvectors, not a predictive shortcut.\n\nThe paper is worth a serious referee—the spectral part alone justifies it—but the numerical claims need major revision before acceptance. I would send it to review with a request for a corrected MSD definition, a reproducibility package, and a more honest statement about what PR adds over direct simulation. Not desk reject, but not close to accept as is.","headline":"Solid noiseless spectral analysis; the numerical diagnostic claim is undercut by an MSD definition error and by the fact that PR and dynamics come from the same diagonalization.","tokens_in":11426,"tokens_out":2576,"would_cite":false,"duration_ms":59757,"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":"For a cyclic quantum walk with static site noise, the participation ratio of the step operator's eigenstates predicts localization before and after a full traversal, and a sharp crossover in both spectral and dynamical indicators occurs…","keywords":["quantum random walk","cyclic graph","Anderson localization","static phase noise","participation ratio","spreading exponent","mean squared displacement","quantum Fourier transform"],"falsifier":"Compute the spreading exponent on the same cycle before and after adding buffer sites so the walk cannot feel the boundary within the fit window; if the exponents diverge for different buffer sizes near $\\phi_s=\\pi/3$, the line-cycle equivalence that underpins the crossover is broken. Independently, search for a noisy realization with a high participation ratio whose dynamics are sub-diffusive; one such case would refute the proposed diagnostic.","tokens_in":10450,"feed_emoji":"⚛️","tokens_out":11701,"duration_ms":107054,"temperature":0.7,"pith_summary":"This paper asks whether a single spectral quantity can predict where a noisy quantum walk will localize without running the walk. The system is a discrete-time quantum walk on a ring of $N$ sites, with a three-parameter unitary coin and independent random phase noise on each site. The authors show that the participation ratio of the step-operator eigenstates, a number between $1$ and $N$ measuring how spread out each eigenstate is, tracks the dynamics: low values coincide with sub-diffusive spreading before the walker rounds the ring, and with early saturation of the mean squared displacement after it has gone around. Across both regimes a sharp crossover appears near static noise strength $\\phi_s = \\pi/3$, marked by a drop in the participation ratio. If the claim holds, localization in these walks can be anticipated from the spectrum alone, avoiding long-time dynamical simulations.","feed_headline":"Participation ratio predicts localization in noisy quantum walks","feed_subtitle":"One number from the step-operator spectrum flags when static site noise stalls the walker, in both propagation regimes.","key_machinery":"The central object is the noisy step operator $\\hat{S}_{\\mathrm{noise}}=(\\hat{D}_{\\mathrm{noise}}\\otimes\\hat{1})\\hat{S}_N$, where $\\hat{D}_{\\mathrm{noise}}$ adds independent random site phases in $[-\\phi_s,\\phi_s]$. The noiseless step operator $\\hat{S}_N$ is diagonalized by the quantum Fourier transform, giving eigenstates $|\\zeta_k\\rangle$ whose eigenvalues sit on two arcs of the unit circle, separated by a gap set by $\\gamma$ and rotated by $(\\theta+\\phi)/2$. The diagnostic that carries the argument is the participation ratio $\\mathrm{PR}(\\zeta_j)=(\\sum_s P(s))^2/\\sum_s P(s)^2$ of the marginal site distribution of each step eigenstate, a number between $1$ and $N$ that measures how many sites an eigenstate effectively occupies. Dynamically, the paper uses the step-resolved mean squared displacement $\\Delta x^2(m)$, a power-law fit $\\Delta x^2(m)\\approx\\alpha m^\\beta$ to extract the spreading exponent in the walk-on-the-line regime, and a bias-corrected coefficient of variation of $\\Delta x^2$ to define saturation in the walk-on-the-cycle regime.","core_discovery":"The paper establishes that, for a discrete-time quantum walk on a cyclic graph with static site phase noise, the eigenstate participation ratio of the step operator is a reliable spectral proxy for localization in both dynamical regimes. Before the walker has traversed the whole ring, low participation ratios predict sub-diffusive spread ($\\beta<1$) and high participation ratios predict ballistic or super-diffusive spread ($\\beta>1$); after a full traversal, low participation ratios predict early saturation of the mean squared displacement at a low level, while high ratios correspond to persistent oscillations. A sharp crossover between these behaviours occurs near $\\phi_s=\\pi/3$, where the participation ratio drops. The three-parameter coin controls the noiseless spectrum: $\\gamma$ opens two spectral bands, the half-sum $(\\theta+\\phi)/2$ rotates the spectrum and produces twofold degeneracy when $(\\theta+\\phi)/2=m\\pi/N$, and degenerate spectra give sinusoidal marginal distributions while non-degenerate spectra give flat ones. The authors conclude that the participation ratio is a fast static diagnostic, complementary to full dynamical simulations, for analysing noisy quantum devices.","pith_inferences":["The paper tests two coin settings; whether the crossover at $\\phi_s=\\pi/3$ persists for other values of $\\gamma$ or for other phase combinations is left untested and is a natural next check.","The same participation-ratio diagnostic could plausibly extend to dynamic disorder or to two-dimensional lattice walks, but those extensions are outside the paper's scope.","A practical implementation would likely need a measurable proxy for the participation ratio, since reconstructing step-eigenstate probabilities may be as expensive as running the dynamics."],"forward_implications":["For a fixed coin and graph size, one can scan noise levels by computing participation ratios of the noisy step operator and identify the localization crossover without propagating the walk.","The crossover at $\\phi_s=\\pi/3$ appears for both the Hadamard and symmetric coins and across different graph sizes, indicating a coin-independent feature of this static phase-noise model.","In the walk-on-the-cycle regime, the coefficient-of-variation convergence criterion supplies a quantitative saturation level that can be used as a localization measure on finite graphs, where the spreading exponent is no longer defined.","The proposed gate-based circuits for the step operator and noise mean the model, and the spectral diagnostic, can be tested on small quantum processors."],"supporting_citations":[{"why":"Supplies the three-parameter unitary coin and the generalized discrete-time quantum walk setting in which Anderson localization is studied.","marker":"[4]"},{"why":"Cited as one source for fitting the mean squared displacement to a power law to quantify how fast the walker spreads.","marker":"[13]"},{"why":"Cited as the source for extracting the spreading exponent beta from the noisy walk's mean squared displacement.","marker":"[15]"},{"why":"Supplies the quantum Fourier transform implementation used to diagonalize the cyclic step operator.","marker":"[31]"},{"why":"Provides the eigenstates and eigenvalues of the step operator on the cyclic graph used throughout the spectral analysis.","marker":"[32]"},{"why":"Introduces quantum walks with random phase shifts, the static site-noise model on which the noisy step operator is built.","marker":"[34]"},{"why":"Provides the classification of spreading regimes via the exponent beta used to characterize the walk-on-the-line dynamics.","marker":"[35]"},{"why":"Supplies the bias-corrected coefficient of variation used to define convergence and saturation of the mean squared displacement.","marker":"[36]"}],"fun_headline_variants":["Participation ratio predicts localization in noisy quantum walks","Spectral metric forecasts quantum walk stalling","One number flags localization in noisy quantum walks","Static noise crossover via participation ratio"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the early-time portion of the walk on the cycle is indistinguishable from a walk on an unbounded line, and that the 'initial steps' window used to fit $\\beta$ lies entirely inside that portion for every noise level and graph size.","fun_headline_variants_meta":{"raw":{"variants":["Participation ratio predicts localization in noisy quantum walks","Spectral metric forecasts quantum walk stalling","One number flags localization in noisy quantum walks","Static noise crossover via participation ratio"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00024,"raw_usage":{"total_tokens":1566,"prompt_tokens":1040,"completion_tokens":526,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":656,"completion_tokens_details":{"reasoning_tokens":472}},"tokens_in":656,"tokens_out":526,"duration_ms":6077,"temperature":1.0,"reasoning_tokens":472,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T05:16:16.056479+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the spreading exponent on the same cycle before and after adding buffer sites so the walk cannot feel the boundary within the fit window; if the exponents diverge for different buffer sizes near $\\phi_s=\\pi/3$, the line-cycle equivalence that underpins the crossover is broken. Independently, search for a noisy realization with a high participation ratio whose dynamics are sub-diffusive; one such case would refute the proposed diagnostic.","supporting_citations":[{"cited_title":"Anderson localization in generalized discrete time quantum walks","cited_arxiv_id":"1708.08194","evidence_quote":"Supplies the three-parameter unitary coin and the generalized discrete-time quantum walk setting in which Anderson localization is studied."},{"cited_title":"Quantum information spreading in a disordered quantum walk","cited_arxiv_id":"2010.10592","evidence_quote":"Cited as one source for fitting the mean squared displacement to a power law to quantify how fast the walker spreads."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Cited as the source for extracting the spreading exponent beta from the noisy walk's mean squared displacement."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the quantum Fourier transform implementation used to diagonalize the cyclic step operator."},{"cited_title":"Necklaces of $\\mathcal{PT}$-symmetric dimers","cited_arxiv_id":"1709.00498","evidence_quote":"Provides the eigenstates and eigenvalues of the step operator on the cyclic graph used throughout the spectral analysis."},{"cited_title":"Sassetti, H","cited_arxiv_id":null,"evidence_quote":"Provides the classification of spreading regimes via the exponent beta used to characterize the walk-on-the-line dynamics."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the bias-corrected coefficient of variation used to define convergence and saturation of the mean squared displacement."}],"review_version":1}