{"id":"ff96b185-add2-4856-92c4-ce765598994e","arxiv_id":"2412.14280","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"The variable-range SYK2 spectral form factor is essentially unchanged for alpha < 1/2, then develops a higher dip and a secondary plateau that trace the single-particle ergodic-to-non-ergodic transition.","lead":"This paper computes the many-body spectral form factor of a quadratic SYK model with power-law decaying interactions, showing the chaotic signature is robust until a critical decay exponent and then transitions to new spectral regimes. It offers a way to detect single-particle localization transitions through a many-body observable, relevant for realistic experiments with long-range interacting quantum simulators.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The uniform-saddle ansatz for α<1/2 is load-bearing: Eq. (S14) leaves O(1) Fourier modes for fixed k, so a competing non-uniform saddle is not ruled out; if such a saddle dominates, the robustness claim fails.","rationale":"The paper's strongest positive evidence is the clean analytic derivation of Eq. (15) and the α=0.3 numerical agreement in Fig. 1b. However, that agreement is at a single representative α and does not test the boundary of the claimed regime. The reader's weakest assumption correctly identifies uniform-saddle dominance as the load-bearing point, and I agree with that assessment. I would not change the CONDITIONAL verdict: the one-loop analytic machinery is credible, but the robustness claim needs the saddle-dominance check before it can be accepted as quantitative. The transition markers for α>1/2 (dip fit at α=0.49, concavity at α=1.02) are secondary; even if those fits were perfect, the α<1/2 robustness claim stands or falls with the saddle assumption. No code or data are provided, which further supports keeping the verdict conditional rather than accepting the quantitative boundary claims as settled.","tokens_in":21846,"tokens_out":14787,"duration_ms":126998,"concrete_test":"Numerically minimize the effective action Eq. (S23) over time-translation-invariant but spatially non-uniform configurations, parameterized by the first few Fourier modes Σ_k(ω) for k=0,1,2,3, at α=0.1, 0.3, 0.45, 0.49, 0.5 and N=200–1000, and compare the action with the uniform-saddle value Eq. (7). If a stationary non-uniform configuration has lower action for any α<1/2, the robustness claim fails; if none exists, the ansatz is supported in the regime where it matters.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central robustness claim (Eq. (15) valid for 0≤α<1/2) rests on the assumption, stated after Eq. (6), that the dominant saddle is site-independent. The support offered in footnote [74] is that 'most' eigenvalues of A vanish in the thermodynamic limit, 'forcing' Σ to be uniform. This does not follow: Eq. (S14) gives Σ̂_{-k} = λ_k · (saddle equation), and Eq. (14) shows that for every fixed k (k=O(1)) the eigenvalue λ_k is O(1) for any 0<α<1/2, not O(1/N); only modes with k∝N vanish (Eq. (13)). A finite number of O(1) Fourier modes is enough to build a non-uniform saddle with action of order N, so the uniform saddle is not forced by the spectrum. Section G only tests time-translation-invariant perturbations around the uniform saddle and establishes breakdown for α>1/2; it does not rule out a competing non-uniform saddle for α<1/2. Thus the sharp boundary at α=1/2, and the entire 'robustness' conclusion, depend on an unproven dominance assumption.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the spectral form factor (SFF) of a quadratic SYK model with power-law distance-dependent couplings, the variable-range SYK2 model of Eq. (1). The authors derive a path-integral representation for the SFF, identify the saddle-point solutions, and compute the one-loop determinant including zero modes associated with the eigenvalues of the coupling-envelope matrix A. Their central analytic result, Eq. (15), states that for 0 ≤ α < 1/2 the SFF is essentially unchanged from the α = 0 case, with subleading eigenvalue-dependent corrections. For α > 1/2 they argue that perturbation theory around the uniform saddle breaks down, and they numerically observe a higher dip and a secondary plateau, which they connect to the single-particle PRBM ergodic-to-non-ergodic transition at α ≈ 1/2 and to Anderson criticality near α ≈ 1. The supplemental material contains the detailed derivations of the saddle-point structure, the eigenvalue spectrum of A, the one-loop and soft-mode contributions, and a perturbative breakdown argument in Section G.","tokens_in":22032,"tokens_out":6036,"duration_ms":56408,"significance":"If the central claim holds, the paper provides a useful bridge between single-particle PRBM physics and many-body spectral statistics: it gives an analytic SFF formula with explicit finite-N eigenvalue corrections, and it identifies observable SFF markers—dip height and secondary plateau—that track known PRBM transitions. The derivation is largely self-contained and not fitted to data: Eq. (15) follows from the path integral with eigenvalues of A computed from the model, and the PRBM phase diagram is used as an external benchmark rather than as an input. The numerical comparison at α = 0.3, N = 200 is excellent, and the paper is notably honest about statistical limitations, including the large sample-to-sample fluctuations of the quadratic SYK SFF. The significance is, however, conditional on the uniform-saddle assumption, which is not fully justified and underpins the sharp α = 1/2 boundary.","major_comments":[{"comment":"The central robustness claim—that Eq. (15) is valid for all 0 ≤ α < 1/2—rests on the assumption that the dominant saddle is site-independent. This assumption is not established. In the Fourier-transformed saddle equation (S14), the non-uniform component Σ̂_{−k} is proportional to λ_k, and Eq. (14) shows that for every fixed k the eigenvalue λ_k is O(1) for 0 < α < 1/2, not O(1/N); only modes with k ∝ N vanish, per Eq. (13). A finite number of O(1) Fourier modes is sufficient to build a non-uniform saddle with action of order N, so the statement in footnote [74] that 'most eigenvalues vanish' does not force the solution to be uniform. Section G tests only time-translation-invariant perturbations around the uniform saddle and establishes breakdown for α > 1/2; it does not rule out a competing non-uniform saddle for α < 1/2. The numerical agreement at α = 0.3 (Fig. 1b) is encouraging but is a single point. The sharp boundary at α = 1/2 and the 'robustness' conclusion therefore require either a stability analysis of the uniform saddle against non-uniform perturbations or a substantially softened claim.","section":"Saddle point equation, after Eq. (6); footnote [74]; Supplemental Eq. (S14)"},{"comment":"The numerical markers of the transitions are based on fitting procedures whose robustness is not quantified. The estimate α ≈ 0.49 is obtained from a linear fit to D(α) for α ≥ 0.6 and its intersection with D = 0, and the estimate α ≈ 1.02 comes from a fourth-order polynomial fit above α = 0.8 with a concavity change. No error bars, fit-range dependence, or finite-size scaling analysis is provided. Since these values are used to associate SFF features with PRBM transitions, the authors should either provide such robustness checks or explicitly label these estimates as heuristic diagnostics rather than precise transition points.","section":"SFF and localization, Fig. 3"},{"comment":"The statement after Eq. (11) that for α > 1/2 'an infinite number of zero modes should appear in the thermodynamic limit, each of them bringing a factor proportional to the coset volume' is not derived in detail. The t.t.i. one-loop expression in Eq. (S34) contains factors (1 − λ_k)^{-1} that become singular as λ_k → 1, but the paper does not analyze whether the resulting contribution is really a product of coset volumes, or whether a resummation is required. The separate perturbative argument in Section G gives a cleaner breakdown signal, but the intermediate 'infinite zero modes' claim should be clarified or removed.","section":"Quadratic fluctuations and Eq. (11)"}],"minor_comments":[{"comment":"There are small typos: 'J1 is a is a Bessel function' should read 'J1 is a Bessel function', and 'forα = 0.3' in the Fig. 1b caption needs a space.","section":"Eq. (15) and Fig. 1b caption"},{"comment":"The sentence 'since most of the vanishes in the thermodynamic limit' is grammatically incomplete; it should read 'since most of the eigenvalues vanish in the thermodynamic limit'. More importantly, the logical argument should be revised in light of Major Comment 1.","section":"Footnote [74]"},{"comment":"The definition of h in Eq. (16) averages over the time interval JT ∈ [10, 15], and the collapse in Fig. 3b is obtained by applying a constant shift to h. The choice of interval and the shift procedure should be stated explicitly in the main text, together with an indication of the sensitivity of the collapsed curves to these choices.","section":"Fig. 3b and Eq. (16)"},{"comment":"In Eq. (S32), the zero-mode volume is quoted as 4πN(1 − x_n^2/4). The derivation of the radius of the vacuum manifold would benefit from a few more intermediate steps, especially the factors of sqrt(N), so that the reader can verify the normalization against the α = 0 limit.","section":"Supplemental Section D"},{"comment":"The discussion of insufficient sampling producing a systematic downward shift of the SFF is valuable and should be mentioned briefly in the main text, since it affects the interpretation of the plateau-height data in Fig. 3b.","section":"Supplemental Section H and Fig. S1"}],"recommendation":"major_revision","confidential_remarks":"The paper is honest and technically substantial, and the analytic SFF derivation for the uniform saddle is a solid contribution. My main concern is the uniform-saddle dominance assumption, which is load-bearing for the central α < 1/2 robustness claim. If the authors can supply a stability analysis, a numerical test of non-uniform saddles, or a carefully restricted claim, the paper would be suitable for publication. The fitting-based estimates of transition points should be presented as diagnostics rather than precise results."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should read this paper if you care about spectral statistics beyond the all-to-all SYK limit. The core new result is an honest analytic formula for the SFF (Eq. 15) of variable-range SYK2, including O(1) eigenvalue corrections, and a clear perturbative argument in Section G that the uniform-saddle expansion breaks down for alpha > 1/2. The alpha -> 0 limit correctly recovers Winer–Jian–Swingle, and the comparison with N=200 numerics at alpha = 0.3 is genuinely good, not just qualitative. That is real progress, not a repackaging.\n\nThe weak spot is exactly where the stress-test note lands. The claim that the site-independent saddle dominates for all alpha < 1/2 rests on footnote 74, which says most eigenvalues of A vanish and therefore force a uniform solution. But Eq. (14) shows eigenvalues for any fixed k are O(1) for 0 < alpha < 1/2; only modes with k proportional to N vanish. So the footnote's logic is incomplete, and the paper does not rule out a competing non-uniform saddle with O(N) action. I would not call the paper wrong on this basis, because the explicit numerical agreement at alpha = 0.3 is real evidence that the uniform saddle is at least the right one in that regime, and the breakdown at alpha > 1/2 is consistent with PRBM expectations. But the sharp boundary at alpha = 1/2 is not proven, and a referee should ask the authors to either prove saddle uniqueness in a limited regime or soften the claim.\n\nThe secondary soft spots are in the transition-point identification: the alpha = 0.49 and alpha = 1.02 markers come from post-hoc fits, shift-collapsed data, and a fourth-order polynomial concavity analysis, all without error bars. The alpha = 3/2 integrable transition is explicitly not resolved. No code or data is shipped, which is a practical problem for a numerics-heavy paper. None of this is fatal, but it means the paper's strongest claims should be framed as evidence, not proof.\n\nBottom line: this deserves a serious referee and, after revision, publication. I would cite it. The authors have done real analytical work, and the gap in the saddle-point justification is a legitimate invitation for follow-up, not a sign of carelessness.","headline":"A genuine analytic extension of the SYK2 spectral form factor to variable-range couplings, with a caveat about the unproven uniform-saddle assumption that needs referee scrutiny.","tokens_in":22686,"tokens_out":1864,"would_cite":true,"duration_ms":17906,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["82B44","81Q50","15B52"],"pacs":["05.30.-d","05.45.Mt","72.15.Rn"],"model":"deepseek-v4-flash","headline":"The many-body spectral form factor of the variable-range SYK2 model is essentially unchanged for interaction-range powers $\\alpha<1/2$, then sharply reorganises near $\\alpha\\simeq 1/2$ into a higher dip and a secondary plateau that track…","keywords":["SYK2 model","variable-range interactions","spectral form factor","power-law random banded matrix","ergodic-non-ergodic transition","many-body level statistics","Anderson transition","path-integral saddle point"],"falsifier":"A direct check would be to fix a system size around $N\\approx 200\\text{--}400$, compute the first perturbative correction $C$ in Eq. (S57) at $\\alpha=0.45$ and $\\alpha=0.55$, and see whether $C$ jumps from $O(1/N)$ to $O(N)$ exactly at $1/2$; a jump already at $\\alpha=0.45$ would falsify the claim. Alternatively, measure the dip shift $D=(1/N)\\log[g(T_{\\rm dip},\\alpha)/g(T_{\\rm dip},0)]$ at $\\alpha=0.45$: Eq. (15) predicts it stays near zero for all $\\alpha<1/2$, so a measurable rise below $\\alpha=0.5$ would break the uniform-saddle assumption.","tokens_in":21545,"feed_emoji":"📉","tokens_out":8100,"duration_ms":67631,"temperature":0.7,"pith_summary":"This paper argues that the many-body spectral statistics of the quadratic SYK model remain remarkably robust when the random all-to-all hopping is replaced by a power-law decay $r^{-\\alpha}$: for $0 \\le \\alpha < 1/2$, the spectral form factor is essentially the same as for $\\alpha=0$, and the paper derives an analytic formula, Eq. (15), that includes the subleading corrections. At $\\alpha = 1/2$ the perturbative expansion around the spatially uniform saddle point breaks down, and for larger $\\alpha$ the SFF develops a higher dip and a secondary plateau. These features track the known single-particle transitions of the underlying power-law random banded matrix: the ergodic-to-non-ergodic transition near $\\alpha=0.5$, Anderson criticality near $\\alpha=1$, and the onset of integrability beyond $\\alpha=1.5$. A sympathetic reader would care because the result identifies which spectral markers survive in realistic, finite-range implementations of SYK-type quantum matter and gives a many-body diagnostic of single-particle localization transitions.","feed_headline":"Chaos in SYK2 survives until power-law decay hits α=1/2","feed_subtitle":"Then the spectral form factor develops a higher dip and a secondary plateau, marking non-ergodicity.","key_machinery":"The central object is the path-integral representation of the spectral form factor in terms of site-resolved collective fields $\\Sigma^{ab}_i(t,t')$, introduced through local two-point functions $G^{ab}_i=\\psi^a_i\\psi^b_i$, rather than the spatially averaged field used in the standard SYK treatment. Because the coupling matrix $A_{ij}=a(i-j)^2/N^2$ is circulant, the calculation diagonalises in Fourier space, and its eigenvalues $\\lambda_k$, given by Eqs. (12)-(14), control the fluctuation structure. Zero modes of the quadratic fluctuation kernel occur exactly when $\\lambda_k=1$; the uniform mode always has $\\lambda_0=1$, and the exponential ramp arises from the resulting SU(2)/U(1) coset volume. The remaining eigenvalues vanish in the thermodynamic limit for $\\alpha<1/2$ and become order one for $\\alpha>1/2$, which is the mechanism that keeps the uniform saddle stable below the threshold and breaks perturbation theory above it. The analytic SFF Eq. (15) combines the classical action, the time-symmetric zero-mode contribution, and the soft-mode corrections.","core_discovery":"The paper's central claim is that the variable-range SYK2 Hamiltonian with hopping amplitude $a(i-j)=\\min(|i-j|,N-|i-j|)^{-\\alpha}$ has a many-body spectral form factor that, for $\\alpha<1/2$, is given by Eq. (15): the classical slope is unchanged, the exponential ramp keeps its $\\alpha=0$ coefficient, and the spatial profile enters only through a subleading sum over eigenvalues $\\lambda_k$ of the circulant matrix $A$. The same eigenvalue spectrum changes character at $\\alpha=1/2$: for $\\alpha>1/2$ infinitely many $\\lambda_k$ become order one, producing an infinite set of would-be zero modes of the fluctuation kernel, and the first perturbative correction to the saddle action then scales as $O(N)$ instead of $O(1/N)$, so perturbation theory around the uniform saddle fails. The numerical SFF shows the predicted robustness below $\\alpha=1/2$ and, above it, a rising dip and a secondary plateau whose height grows with $\\alpha$, merges with the late-time plateau near $\\alpha\\simeq 3/2$, and shows a concavity change at $\\alpha\\simeq 1.02$. The paper reads these as the many-body imprints of the single-particle ergodic-to-non-ergodic transition at $\\alpha\\simeq 1/2$, Anderson criticality at $\\alpha\\simeq 1$, and integrability for $\\alpha\\gtrsim 3/2$.","pith_inferences":["If the uniform saddle actually gives way to a non-uniform one for $\\alpha>1/2$, the same $\\lambda_k=1$ zero-mode criterion could help locate that new saddle explicitly; finding its form would produce an analytic SFF for the non-ergodic regime, which the paper leaves open.","The structural similarity between the matrix $A$ here and the adjacency matrix used for sparse SYK suggests that the eigenvalue criterion $\\lambda_k=1$ may serve as a common diagnostics for connectivity-driven transitions; testing whether sparsification triggers the same dip-height and secondary-plateau markers would extend this logic to random-graph ensembles.","The interpretation of the secondary plateau as prethermalization tied to nearly conserved charges could be tested dynamically by constructing approximate local integrals of motion from the hopping matrix in the localized regime; if they exist, their support sizes should control the plateau height.","The dip height and plateau-height probes could be transferred to out-of-time-order correlators or two-time correlation functions to ask whether the $\\alpha\\simeq 1/2$ and $\\alpha\\simeq 1$ thresholds appear in the dynamics itself, not only in spectral statistics."],"forward_implications":["For every $0\\le \\alpha<1/2$, the spectral form factor follows the single analytic formula Eq. (15), so chaos diagnostics such as the slope and the exponential ramp keep their $\\alpha=0$ values with only subleading corrections.","The dip height $D(\\alpha)=(1/N)\\log[g(T_{\\rm dip},\\alpha)/g(T_{\\rm dip},0)]$ rises sharply near $\\alpha\\simeq 0.5$, with a finite-size extrapolation giving $\\alpha=0.49$, providing a practical marker of the ergodic-to-non-ergodic transition.","A secondary plateau develops for $1/2\\lesssim\\alpha\\lesssim 3/2$; its height grows with $\\alpha$ and merges with the late-time plateau near $\\alpha\\simeq 3/2$, signalling a prethermalization regime in which the Hilbert space is not fully explored.","The plateau-height curve changes concavity at $\\alpha\\simeq 1.02$, matching the Anderson critical point of the single-particle power-law random banded matrix.","For $\\alpha\\gtrsim 3/2$ the exponential ramp disappears, consistent with the onset of integrability, although the numerics cannot resolve a sharp transition there because the SFF fluctuations grow with $N T$ and limit the attainable statistics."],"supporting_citations":[{"why":"Supplies the quadratic-SYK spectral form factor calculation, the exponential-ramp zero-mode mechanism, and the $\\alpha=0$ baseline that Eq. (15) generalises.","marker":"[63]"},{"why":"Defines the power-law random banded matrix phases ($\\alpha<1/2$ ergodic, $\\alpha=1$ Anderson critical, $\\alpha>3/2$ integrable) whose transitions the paper's many-body markers are compared with.","marker":"[40]"},{"why":"Introduces the spectral form factor as the level-statistics diagnostic whose universal dip-ramp-plateau structure organises the analysis.","marker":"[45]"},{"why":"Provides the path-integral treatment of the spectral form factor that the paper adapts to a site-resolved collective-field formulation.","marker":"[46]"},{"why":"Introduces the local two-point collective-field construction and the adjacency matrix with uniform eigenvalue 1, which the present variable-range calculation follows.","marker":"[72]"},{"why":"Documents the secondary plateau in spectral form factors of non-chaotic systems, used here as the reference for the prethermalization regime.","marker":"[48]"},{"why":"Links pre-ramp spectral features to nearly conserved charges, the interpretation the paper adopts for the secondary plateau.","marker":"[59]"},{"why":"Quantifies the $g(T)$ fluctuations that scale with $N T$ and limit the accessible system sizes and statistics, shaping the numerical evidence.","marker":"[76]"}],"fun_headline_variants":["SYK2 chaos survives until α=1/2, then plateau splits","α=1/2: sharp transition in SYK2 spectral form factor","Variable-range SYK2: transition at α=1/2, new spectra regime","SYK2 with power-law decay: chaos ends at α=1/2, new plateau","Power-law SYK2: robustness below α=1/2, non-ergodicity above"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The derivation assumes that the spatially uniform saddle point of Eq. (7) is the dominant saddle for every $\\alpha<1/2$; if a non-uniform, site-dependent saddle already wins inside that window, the robust-formula conclusion would fail.","fun_headline_variants_meta":{"raw":{"variants":["SYK2 chaos survives until α=1/2, then plateau splits","α=1/2: sharp transition in SYK2 spectral form factor","Variable-range SYK2: transition at α=1/2, new spectra regime","SYK2 with power-law decay: chaos ends at α=1/2, new plateau","Power-law SYK2: robustness below α=1/2, non-ergodicity above"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000752,"raw_usage":{"total_tokens":3393,"prompt_tokens":1036,"completion_tokens":2357,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":652,"completion_tokens_details":{"reasoning_tokens":2245}},"tokens_in":652,"tokens_out":2357,"duration_ms":15006,"temperature":1.0,"reasoning_tokens":2245,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T12:23:30.004445+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct check would be to fix a system size around $N\\approx 200\\text{--}400$, compute the first perturbative correction $C$ in Eq. (S57) at $\\alpha=0.45$ and $\\alpha=0.55$, and see whether $C$ jumps from $O(1/N)$ to $O(N)$ exactly at $1/2$; a jump already at $\\alpha=0.45$ would falsify the claim. Alternatively, measure the dip shift $D=(1/N)\\log[g(T_{\\rm dip},\\alpha)/g(T_{\\rm dip},0)]$ at $\\alpha=0.45$: Eq. (15) predicts it stays near zero for all $\\alpha<1/2$, so a measurable rise below $\\alpha=0.5$ would break the uniform-saddle assumption.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the spectral form factor as the level-statistics diagnostic whose universal dip-ramp-plateau structure organises the analysis."},{"cited_title":"Br´ ezin and S","cited_arxiv_id":null,"evidence_quote":"Provides the path-integral treatment of the spectral form factor that the paper adapts to a site-resolved collective-field formulation."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the local two-point collective-field construction and the adjacency matrix with uniform eigenvalue 1, which the present variable-range calculation follows."},{"cited_title":"Hopjan and L","cited_arxiv_id":null,"evidence_quote":"Links pre-ramp spectral features to nearly conserved charges, the interpretation the paper adopts for the secondary plateau."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Quantifies the $g(T)$ fluctuations that scale with $N T$ and limit the accessible system sizes and statistics, shaping the numerical evidence."}],"review_version":1}