{"id":"870c16fc-98d9-4ab7-a592-64db1ac0b2a6","arxiv_id":"1908.09683","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A grid-based Poincaré recurrence method reproduces Lyapunov stability charts and, via an approximate relation, promises global maps of chaotic diffusion timescales.","lead":"The authors present a numerical method for mapping chaotic dynamics by recording how long orbits take to return to a small neighborhood of their starting point on dense grids of initial conditions. It matches Lyapunov-based charts in two test systems and adds a route to estimating local diffusion timescales.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The diffusion-timescale claim rests on Eq. (5), which is unvalidated and conflicts with Kac's return-time law; PRM is supported only as a chaos/stability indicator.","rationale":"The reader's weakest_assumption correctly identifies the unvalidated relation Eq. (5) as the load-bearing link between PRM charts and diffusion timescales. My independent reading sharpens this concern: for bounded Hamiltonian systems, the mean first-return time to a small phase-space neighborhood is governed by Kac's lemma and depends on accessible phase-space volume and ε, not on the diffusion coefficient. Equation (5) can only hold under additional assumptions about a diffusive scale Δy that the paper never states. Thus the central advertised advantage over LE, FLI, MEGNO, and FA is not established by the reported computations. The method itself is clearly specified, the code is publicly archived, and the qualitative agreement between PRM and LE charts supports PRM as a practical chaos indicator. A conditional verdict is appropriate, requiring either direct validation of Eq. (5) against measured diffusion coefficients or a softened formulation of the diffusion-rate claim. Since the reader already recommends CONDITIONAL on the same assumption, no verdict change is needed.","tokens_in":13987,"tokens_out":4862,"duration_ms":62664,"concrete_test":"Re-run the existing PRM_HH and LE codes on the same Hénon–Heiles grid at E = 0.1, and additionally estimate local action-diffusion coefficients D_y from long trajectories (e.g., from the variance of p2, q2, or an action variable over sliding time windows). Then test two predictions for each chaotic grid point: Eq. (5) predicts T_r·D_y/(Δy)^2 ≈ const, while Kac's formula predicts ⟨T_r⟩ ≈ V/ε^n. If the observed T_r either does not track D_y or scales with ε as V/ε^n rather than as (Δy)^2/D_y, the diffusion-chart claim is falsified; if T_r does track D_y, the method is supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 3 converts first-recurrence times into local diffusion timescales through Eq. (5), τ ∼ (Δy)^2/D_y, citing [38]. This is the only bridge from the computed T_r field to the advertised \"charts of local diffusion timescales\", yet it is neither derived in this paper nor checked against any direct measurement of D_y. The concern is not merely that \"an approximation may be rough\": for the bounded Hénon–Heiles application, the mean first-return time to an ε-neighborhood on an ergodic chaotic component of measure V obeys Kac's lemma, ⟨T_r⟩ ≈ V/ε^n, which contains no diffusion coefficient at all. Eq. (5) would require the return statistics to be dominated by diffusion across a specified scale Δy, and Δy is never identified. For the restricted three-body problem, the situation is even less direct because orbits may escape, and a recurrence to the initial box can occur before any diffusive transport has taken place. The figures therefore establish that PRM agrees with LE as a chaos/stability indicator, but they do not establish the paper's central claim of mapping diffusion rates. The Discussion's stronger statement that inverse recurrence times \"provide massive measures of the local diffusion rates\" is unsupported.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces the Poincaré recurrence method (PRM), a numerical tool that computes first recurrence times to small neighborhoods of initial conditions on massive grids, and applies it to two Hamiltonian systems: the bounded Hénon–Heiles system and the non-bounded planar restricted three-body problem. The authors compare PRM charts with Lyapunov exponent (LE) charts and report close structural agreement. The central claimed novelty is that PRM, unlike LE, FLI, MEGNO, or FA, can construct charts of local diffusion timescales via the approximate relation τ ∼ (Δy)^2/D_y (Eq. 5). The paper also provides implementation details and links to Fortran codes.","tokens_in":14210,"tokens_out":2806,"duration_ms":32299,"significance":"If the diffusion-mapping claim were established, PRM would be a valuable addition to the toolbox of global chaos diagnostics, especially because the method is algorithmically simple, computationally parallelizable, and the authors provide openly accessible code. The visual agreement between PRM and LE charts in Figs. 1 and 3 is credible and supports PRM as a chaos/stability indicator at zero fitted-parameter cost. However, the paper's advertised advantage over existing methods rests entirely on Eq. (5), and the manuscript does not validate this relation against any direct diffusion measurement. The significance of the contribution therefore depends on whether that validation can be supplied or whether the claims are appropriately weakened; as it stands, the paper establishes a useful chaos indicator but not the claimed diffusion-timescale mapping.","major_comments":[{"comment":"The central claim that PRM charts diffusion rates rests on the relation τ ∼ (Δy)^2/D_y, which is quoted from Ref. [38] but neither derived nor validated in this paper. No independent computation of D_y is performed, and the quantity Δy is never identified for either application. For the bounded Hénon–Heiles case, Kac's lemma gives a mean first-return time to an ε-neighborhood that scales as V/ε^n for an ergodic chaotic component, with no dependence on a diffusion coefficient; the observed recurrence statistics could therefore reflect volumetric recurrence geometry rather than diffusive transport. To support the diffusion-timescale interpretation, the authors should either compare PRM-derived local diffusion times with directly measured diffusion rates (e.g., from ensembles of trajectories spreading in action space) or identify Δy explicitly and show that Eq. (5) reproduces known diffusion coefficients.","section":"Section 3, Eq. (5)"},{"comment":"The PRM charts for the Hénon–Heiles system assign the color red to trajectories with Tr > 10^5, which equals the integration time Tint; for the most chaotic regions, no recurrence is observed within the integration interval, so the chart does not provide a numerical recurrence time for those regions. This means the claimed diffusion timescale is undefined precisely in the strongly chaotic domains where diffusion is most relevant, and the red regions merely indicate 'no recurrence within Tint', analogous to a truncated-time chaos indicator. The paper should state this limitation and, if diffusion timescales are the goal, demonstrate that the integration time is long enough to resolve recurrences in the chaotic regions or discuss how the upper-bound values should be interpreted.","section":"Section 2.2, Fig. 1"},{"comment":"For the restricted three-body problem, the phase space is non-bounded and particles can escape before experiencing diffusive transport; a recurrence to the initial box can also occur early, before the trajectory has spread appreciably. Equation (5) assumes normal diffusion across a characteristic scale Δy, and this assumption is not examined for the escape-dominated dynamics of the circumbinary problem. The paper should either justify the applicability of Eq. (5) to the non-bounded case or restrict the diffusion-timescale claim to bounded systems, with a separate indicator (e.g., escape time) for unbounded ones.","section":"Section 2.4 and Section 3"},{"comment":"The Discussion states that 'global charts of the massively computed Poincaré recurrence times provide direct global representations of spatial distributions of the local diffusion times' and that inverse recurrence times provide 'massive measures of the local diffusion rates.' This is stronger than what the preceding sections establish, because the connection between recurrence times and diffusion rates is only the approximate Eq. (5) and is not tested on either system. The overstatement should be corrected unless direct diffusion validation is added.","section":"Section 4, Discussion"}],"minor_comments":[{"comment":"The fitting of the integral recurrence distribution is reported with a correlation coefficient R = 0.99, but no uncertainty on the fitted exponents α and β is given; since the fits are used to characterize the distributions, confidence intervals would be helpful.","section":"Section 2.2, Fig. 2"},{"comment":"The recurrence condition (4) uses the same box half-width ΔX = ΔY = ΔPX = ΔPY = 10^-3, but the text does not explain how this choice relates to the natural scales of position and momentum in the restricted three-body problem; a sentence justifying the common half-width would improve reproducibility.","section":"Section 2.4, Eq. (4)"},{"comment":"The caption says 'relationships between the Lyapunov and Poincaré recurrence times', but the text states that the plots show no correlation; the wording should be consistent, for example 'correlation plots' or 'scatter plots'.","section":"Figure 4"},{"comment":"The phrase 'massive measures of the local diffusion rates' in Section 4 is informal; 'massively computed' is used elsewhere, and the paper would read better if the terminology were unified.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The paper's advertised advance over existing chaos indicators is the diffusion-timescale capability. I would ask the authors to validate Eq. (5) against direct diffusion measurements or to substantially soften the diffusion claims; without that, the central novelty reduces to a new chaos/stability indicator that agrees with LE, which is useful but not the claimed breakthrough."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper's headline—that PRM can chart local diffusion timescales—is not established. What is established is that the Poincaré recurrence method is a simple, cheap way to produce global stability charts that agree visually with Lyapunov exponent charts. The two worked examples (Hénon–Heiles and the restricted three-body problem) are reasonable, the algorithm is clearly specified, and the code is available on Zenodo. That part is solid.\n\nThe trouble is Section 3. The bridge from recurrence time to diffusion time is Eq. (5), τ ~ (Δy)^2/D_y, taken from Chirikov and Shepelyansky. It is never derived or validated here. That matters for two reasons. First, in a bounded system like Hénon–Heiles, Kac's lemma says the mean first-return time to an ε-ball on a chaotic component is roughly V/ε^n, which has no diffusion coefficient in it. Eq. (5) requires the return time to be dominated by diffusion across a scale Δy, and Δy is never identified. Second, in the three-body case, orbits can escape; a return to the initial box can happen before any diffusive transport has occurred. So the charts show signs of regular versus chaotic regions, but they do not show local diffusion rates. The authors write \"in some approximation,\" but an approximation still needs at least one numerical check against a known diffusion rate. There is none.\n\nThe comparison with LE charts is also only visual; no quantitative agreement metric is given. That is a minor point, given how obvious the structural agreement looks in the figures.\n\nIf the diffusion-timescale claim were removed or softened, the paper would be a decent methods note: PRM as a chaos/stability indicator is simple, cheap, and works. As it stands, the main advertised advantage rests on an unvalidated formula. A referee should ask for either a direct test of Eq. (5) on a system with a known diffusion coefficient (the standard map would be the obvious candidate) or a revised framing that does not claim to map diffusion rates.\n\nI'd send this to a serious referee, because the method is worth having and the claim is fixable. But I would not cite the diffusion-timescale result until it is tested.","headline":"A cheap, simple chaos/stability indicator that matches Lyapunov charts, but the advertised diffusion-timescale capability is unsupported and needs either validation or a softer claim.","tokens_in":14743,"tokens_out":2897,"would_cite":false,"duration_ms":26977,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["02.60.-x","05.45.-a","05.45.Pq","45.50.Pk","95.10.Ce","95.10.Fh"],"model":"deepseek-v4-flash","headline":"A grid of first-return times, not Lyapunov exponents, can chart where chaotic diffusion is fast and slow across a Hamiltonian phase space.","keywords":["Poincaré recurrences","chaotic diffusion","Hamiltonian systems","stability charts","Lyapunov exponents","restricted three-body problem","numerical methods","diffusion timescales"],"falsifier":"Compute, for a Hamiltonian with a known stochastic layer, both the PRM-derived local diffusion coefficient $D_y=(\\Delta y)^2/\\tau$ at a grid of points and an independent direct estimate of diffusion from the growth of the variance of the slow action with time; if the two disagree systematically across the chaotic layer, the diffusion-chart interpretation of PRM fails while its chaos-detection value remains.","tokens_in":13780,"feed_emoji":"🌀","tokens_out":8290,"duration_ms":80837,"temperature":0.7,"pith_summary":"The paper proposes the Poincaré recurrence method (PRM): on a fine grid of initial conditions, integrate each orbit until it first returns to a small $\\varepsilon$-neighbourhood of its starting point and record the return time. Across two very different Hamiltonian regimes, bounded (a two-degree-of-freedom potential) and unbounded (a planar restricted three-body problem), the resulting log-recurrence-time charts reproduce the same regular-versus-chaotic structure as finite-time Lyapunov charts. The added payoff is that, through the approximate relation $\\tau\\sim(\\Delta y)^2/D_y$, the same charts can be read as maps of local chaotic diffusion timescales, something standard chaos indicators do not provide. A sympathetic reader would take this as a practical new tool for global stability analysis that is simpler than and complementary to Lyapunov-based methods.","feed_headline":"Return times map chaotic diffusion across phase space","feed_subtitle":"Massive grids of first Poincaré recurrences match Lyapunov stability charts and add diffusion timescales.","key_machinery":"The engine is a per-node return-time measurement: for each initial condition, fix a neighbourhood of radius $\\varepsilon$ (a sphere or box), integrate the equations of motion, and record the first time $T_r$ the orbit re-enters that neighbourhood; painting $T_r$ over the grid produces the chart. The theoretical bridge from return times to diffusion is the approximate identity $\\tau\\sim(\\Delta y)^2/D_y$ between mean recurrence time, a characteristic distance in the slow variable, and the local diffusion coefficient $D_y$. The paper also uses the integral distribution $F(T_r)\\propto T_r^{-\\alpha}$ with $\\alpha\\approx 3/2$–$1.6$ to connect recurrence statistics to the long-time algebraically decaying component caused by sticky chaotic motion; the numerical integrations are carried out with an adaptive high-order Runge–Kutta integrator, supplemented by a regularization scheme for close encounters in the three-body case.","core_discovery":"The paper's central claim is that the global chart of massively computed first-return times is itself a global chart of local diffusion times: where the return time is short, chaotic diffusion is fast, and where it is long, diffusion is slow. The claim is backed by two demonstrations. In the bounded case, a $500\\times500$ grid of return times in the section plane reproduces the known island-and-stochastic-layer structure of the two-degree-of-freedom Hamiltonian at $E=0.1$, in close agreement with a parallel Lyapunov-time chart. In the unbounded case, a $201\\times201$ grid of first-return times in the pericentric-distance–eccentricity plane of the restricted three-body problem reproduces the Lyapunov chart including the fractal resonant 'teeth' at the order/chaos boundary. The paper also shows that the integral distribution of return times splits into an exponential part and an algebraic tail with exponent near $3/2$, connecting the method to established recurrence-statistics theory.","pith_inferences":["Editorial inference: the relation $\\tau\\sim(\\Delta y)^2/D_y$ suggests PRM charts could be made anisotropic by measuring returns along separate coordinate directions, which would recover directional diffusion coefficients instead of a scalar; the paper does not do this.","Editorial inference: because the method needs no variational equations, it may transfer directly to stochastic or dissipative systems where recurrence can be defined without a conserved Hamiltonian; the paper explicitly leaves dissipative systems to future work.","Editorial inference: the empirical recipe for choosing $\\varepsilon$ (lower it until the chart is noiseless, or match resolution) could be automated into an adaptive-resolution scheme that scans phase space at multiple scales, something the paper does not attempt.","Editorial inference: a cheap PRM pre-filter could be used to decide where to spend expensive Lyapunov or FLI computations in high-dimensional systems, concentrating variational integration on the chaotic regions that PRM flags."],"forward_implications":["If the method holds, any Hamiltonian system whose equations of motion can be integrated can receive a global diffusion map at roughly the same computational effort as a Lyapunov chart.","The recurrence-time chart reproduces the regular/chaotic boundaries that Lyapunov charts show, so PRM can serve as a direct replacement for LE-based stability charts in bounded and non-bounded systems.","Because the recurrence distribution has an algebraic tail with nearly universal exponent, recurrence charts of modest integration time remain stable indicators of chaos even when sticky regions contribute.","Running LE and PRM in parallel on long timescales opens a route to massive statistical studies of Lyapunov-time versus recurrence-time correlations, including escape-related power laws in systems with unbounded phase space."],"supporting_citations":[{"why":"Supplies the approximate relation $\\tau\\sim(\\Delta y)^2/D_y$ linking mean recurrence time to diffusion, the property that lets PRM charts be read as diffusion maps.","marker":"[38]"},{"why":"Provides the canonical bounded-phase-space Hamiltonian used to benchmark PRM and its chaotic domains.","marker":"[45]"},{"why":"Supplies the adaptive high-order Runge–Kutta integrator used to compute trajectories and return times in both applications.","marker":"[46]"},{"why":"Provides the reference Lyapunov-time computation and phase-space chaotic fraction for the bounded case at the chosen energy.","marker":"[4]"},{"why":"Documents the nearly universal algebraic decay exponent in recurrence statistics that the paper's long-time interpretation draws on.","marker":"[43]"},{"why":"Supplies the prior restricted-three-body integration code that the unbounded-phase-space PRM implementation extends.","marker":"[50]"},{"why":"Introduces the notion of dynamical temperature that the paper uses to frame diffusion-rate charts.","marker":"[52]"}],"fun_headline_variants":["Recurrence grids map diffusion, matching Lyapunov charts","First-return time maps reveal chaotic diffusion rates","Poincaré recurrences on grids chart local diffusion","Return-time statistics rival Lyapunov methods for chaos"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument's load-bearing premise is that the approximate relation $\\tau\\sim(\\Delta y)^2/D_y$ between mean recurrence time and diffusion coefficient, borrowed from the study of the standard map, holds locally for the Hamiltonian system being charted; the paper does not verify this relation independently.","fun_headline_variants_meta":{"raw":{"variants":["Recurrence grids map diffusion, matching Lyapunov charts","First-return time maps reveal chaotic diffusion rates","Poincaré recurrences on grids chart local diffusion","Return-time statistics rival Lyapunov methods for chaos"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00049,"raw_usage":{"total_tokens":2395,"prompt_tokens":915,"completion_tokens":1480,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":531,"completion_tokens_details":{"reasoning_tokens":1416}},"tokens_in":531,"tokens_out":1480,"duration_ms":11835,"temperature":1.0,"reasoning_tokens":1416,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:04:01.775867+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute, for a Hamiltonian with a known stochastic layer, both the PRM-derived local diffusion coefficient $D_y=(\\Delta y)^2/\\tau$ at a grid of points and an independent direct estimate of diffusion from the growth of the variance of the slow action with time; if the two disagree systematically across the chaotic layer, the diffusion-chart interpretation of PRM fails while its chaos-detection value remains.","supporting_citations":[{"cited_title":"Chirikov, D.L","cited_arxiv_id":null,"evidence_quote":"Supplies the approximate relation $\\tau\\sim(\\Delta y)^2/D_y$ linking mean recurrence time to diffusion, the property that lets PRM charts be read as diffusion maps."},{"cited_title":"H´ enon, C","cited_arxiv_id":null,"evidence_quote":"Provides the canonical bounded-phase-space Hamiltonian used to benchmark PRM and its chaotic domains."},{"cited_title":"Hairer, S.P","cited_arxiv_id":null,"evidence_quote":"Supplies the adaptive high-order Runge–Kutta integrator used to compute trajectories and return times in both applications."},{"cited_title":"Shevchenko, A.V","cited_arxiv_id":null,"evidence_quote":"Provides the reference Lyapunov-time computation and phase-space chaotic fraction for the bounded case at the chosen energy."},{"cited_title":"Cristadoro, R","cited_arxiv_id":null,"evidence_quote":"Documents the nearly universal algebraic decay exponent in recurrence statistics that the paper's long-time interpretation draws on."},{"cited_title":"Rollin, J","cited_arxiv_id":null,"evidence_quote":"Supplies the prior restricted-three-body integration code that the unbounded-phase-space PRM implementation extends."},{"cited_title":"Shevchenko, Physica A 386 (2007) 85–91","cited_arxiv_id":null,"evidence_quote":"Introduces the notion of dynamical temperature that the paper uses to frame diffusion-rate charts."}],"review_version":1}