{"id":"b260e26e-019a-472c-8593-28b59e6f3b99","arxiv_id":"2606.25155","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":6.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":0,"one_line_summary":"A framework detects multiscale recurrences in irregular time series by applying 1-Wasserstein distance with thresholds derived from Brownian motion scale invariance.","lead":"The paper introduces a recurrence analysis framework using the 1-Wasserstein distance to detect repeating patterns in multiscale time series with irregular resolution and abrupt changes, such as paleoclimate records. A smart generalist might read it for new tools to analyze complex temporal patterns in climate and other irregular data systems.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.3","headline":"Reliance on 1D Brownian motion null model for recurrence threshold may fail to match correlation structure or non-stationarity of paleoclimate series","rationale":"The reader's weakest_assumption correctly isolates the single point where the argument is least secure: the transfer of a BM-derived threshold to data whose statistical properties are not shown to match. No other internal inconsistency is visible from the abstract and the stated claim; the numerical scale-invariance result for BM itself is not contested here.","tokens_in":1816,"tokens_out":360,"duration_ms":11712,"concrete_test":"Generate phase-randomized or fractional-Gaussian-noise surrogates that preserve the power spectrum and variance of the ice-core and debris records; recompute the empirical CDF of 1-Wasserstein distances between non-overlapping segments at the same set of scales used in the paper; if the 95th-percentile threshold differs by more than 15% from the BM-derived value at any scale, the BM null model is not transferable.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central construction defines recurrences via local minima of 1-Wasserstein distances that fall below a threshold taken from the tail probability under 1D Brownian motion. This threshold is justified by a numerical demonstration of scale invariance in the W1 distance distribution for BM patterns. For the claim to hold in the target application, the distance statistics of the paleoclimate records (ice-rafted debris, ice cores) must be sufficiently close to those of BM; otherwise the chosen cutoff has no calibrated meaning. Paleoclimate series typically contain trends, regime shifts, and long-memory correlations that deviate from standard BM, so the null distribution used for thresholding is not guaranteed to be appropriate.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The manuscript introduces a recurrence analysis framework for multiscale time series with irregular resolution and abrupt changes, using a mathematical transform of the 1-Wasserstein distance. Recurrences are defined as local minima of this distance that fall below a threshold derived from the tail probability under one-dimensional Brownian motion, justified by a numerical demonstration of scale invariance in the W1 distance distributions for BM patterns. Recurrences are then combined under a non-overlapping condition to identify multiscale events. Applications are shown on ice-rafted debris and ice core records, with detected recurrences spanning over two orders of magnitude in duration.","tokens_in":1970,"tokens_out":451,"duration_ms":20148,"significance":"If the BM-derived threshold is shown to be appropriate, the approach supplies a scale-invariant, numerically grounded method for recurrence detection in non-stationary paleoclimate series, extending recurrence analysis beyond standard distance functions and regular sampling. The explicit numerical demonstration of scale invariance under BM is a positive feature that supports the parameter-light threshold choice.","major_comments":[{"comment":"Abstract and threshold construction: the claim that the threshold is 'principled' depends on 1D Brownian motion supplying a suitable null model whose distance statistics match those of the target paleoclimate series. Paleoclimate records typically contain trends, regime shifts, and long-memory correlations absent from standard BM; without explicit checks (e.g., comparison of empirical W1 distributions or surrogate tests preserving the series' autocorrelation structure), the threshold lacks calibrated meaning for the claimed applications.","section":"Abstract (threshold definition)"}],"minor_comments":[{"comment":"The 'simple mathematical transform' that enables use of the 1-Wasserstein distance on time series is referenced but not shown; including the explicit mapping or pseudocode would aid reproducibility.","section":"Abstract"},{"comment":"The title refers to 'recurrence networks' yet the abstract and central construction emphasize recurrence detection and combination; a short clarification of how the detected events form or are analyzed as a network would improve consistency.","section":null}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":{"model":"grok-4.3","summary":"We thank the referee for the constructive feedback. We address the single major comment below.","responses":[{"response":"We agree that standard Brownian motion does not reproduce trends, regime shifts, or long-memory correlations typical of paleoclimate records, and that the manuscript does not include surrogate tests that preserve autocorrelation structure or direct comparisons of empirical W1 distributions. The BM reference is used solely to exploit the numerically demonstrated scale invariance of the W1 distance tail probabilities, thereby supplying a single, scale-consistent threshold without additional parameters. This is a modeling choice for threshold construction rather than a claim that BM statistics match the data. In revision we will (i) rephrase the abstract and methods to avoid the word 'principled' and instead describe the threshold as 'scale-invariant under a Brownian-motion reference,' (ii) add an explicit limitations paragraph noting the mismatch with paleoclimate autocorrelation, and (iii) outline how surrogate-based calibration could be performed in future applications. No new numerical checks will be added at this stage.","revision_made":"partial","referee_comment":"[Abstract (threshold definition)] Abstract and threshold construction: the claim that the threshold is 'principled' depends on 1D Brownian motion supplying a suitable null model whose distance statistics match those of the target paleoclimate series. Paleoclimate records typically contain trends, regime shifts, and long-memory correlations absent from standard BM; without explicit checks (e.g., comparison of empirical W1 distributions or surrogate tests preserving the series' autocorrelation structure), the threshold lacks calibrated meaning for the claimed applications."}],"tokens_in":1405,"tokens_out":340,"duration_ms":18240,"standing_objections":[]},"desk_editor":{"model":"grok-4.3","letter":"The main point is a recurrence detection scheme that turns time series segments into 1-Wasserstein distances, flags local minima below a cutoff taken from the tail probability under 1D Brownian motion, and then merges non-overlapping events across scales. They demonstrate it on ice-rafted debris and ice core records, recovering patterns that last from decades to millennia.\n\nWhat is actually new is the concrete package: the distance transform itself, the numerical check that the W1 distribution stays scale-invariant for Brownian motion, and the non-overlapping aggregation step. Standard recurrence plots struggle with uneven sampling and scale mixing; this setup tries to address both at once. The examples show the method can run on real archives without heavy preprocessing, which is a practical plus.\n\nThe soft spot is the threshold choice. The cutoff rests on the assumption that 1D Brownian motion supplies a reasonable null for what counts as similar in paleoclimate data. Those series usually carry trends, long-range correlations, and regime shifts that plain Brownian motion lacks. If the distance statistics differ, the probability threshold stops being calibrated. The abstract does not report robustness checks against other null models or synthetic series with known structure, so that part stays thin.\n\nThis is for people who already work on pattern detection in uneven geophysical time series. A reader in paleoclimatology or nonlinear time series methods could test the code on their own records and see whether the detected events line up with known events.\n\nIt is worth sending to peer review. The construction is distinct enough and the application concrete enough that referees can usefully press on the null-model justification and ask for more validation metrics.","headline":"The paper gives a new recurrence rule using 1-Wasserstein local minima below a Brownian-motion probability threshold, aimed at irregular multiscale series, but the null model fit to paleoclimate statistics is the open issue.","tokens_in":2451,"tokens_out":421,"would_cite":false,"duration_ms":28078,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.3","headline":"The 1-Wasserstein distance between time series patterns is scale-invariant under Brownian motion, supplying a threshold for identifying recurrences at any scale.","keywords":["recurrence analysis","Wasserstein distance","paleoclimate time series","multiscale patterns","Brownian motion","threshold selection","ice core records"],"falsifier":"Empirical 1-Wasserstein distance distributions computed on paleoclimate records that deviate markedly from the scale-invariant form predicted by Brownian motion would falsify the threshold rule.","tokens_in":2735,"feed_emoji":"","tokens_out":658,"duration_ms":23178,"temperature":0.7,"pith_summary":"The paper develops a recurrence detection method for multiscale time series that have irregular sampling and abrupt shifts, such as paleoclimate archives. It applies a transform of the 1-Wasserstein distance so that repeating patterns appear as local minima below a threshold taken from the probability of equally similar patterns in one-dimensional Brownian motion. Numerical checks show that the distance distributions remain scale-invariant in Brownian motion, removing the need for arbitrary cutoffs that vary with resolution. A reader would care because the approach can flag climatic events whose durations range over two orders of magnitude within a single record.","feed_headline":"Brownian motion sets recurrence thresholds for multiscale climate data","feed_subtitle":"1-Wasserstein distances supply scale-invariant local-minima thresholds that flag events spanning two orders of magnitude in irregular record","key_machinery":"The 1-Wasserstein distance applied to time-series segments via a mathematical transform, with recurrence thresholds taken from the scale-invariant distribution of distances in one-dimensional Brownian motion.","core_discovery":"By demonstrating numerically the scale invariance of 1-Wasserstein distance distributions between patterns drawn from one-dimensional Brownian motion, the framework defines recurrences at any time scale as local minima of the distance that fall below the probability threshold given by that null model; non-overlapping recurrences can then be aggregated to produce a set of distinct multiscale recurring events, as illustrated on ice-rafted debris and ice-core records.","pith_inferences":["The same Brownian-derived threshold rule could be tested on synthetic series that embed known repeating patterns at multiple scales.","If the identified recurrences align with independently dated climate events, the framework would supply an objective way to compare recurrence statistics across different archives.","Extension to multivariate or higher-dimensional series would require checking whether the scale-invariance property survives the change in distance definition."],"forward_implications":["Recurrences can be identified consistently across scales from seasonal to multi-millennial within the same record.","Non-overlapping aggregation produces a distinct collection of multiscale recurring events.","The method applies directly to records with abrupt changes and irregular temporal resolution.","Detected events in ice-core and debris records exhibit durations spanning more than two orders of magnitude."],"fun_headline_variants":["Wasserstein networks detect multiscale recurrences in climate records","1-Wasserstein distance thresholds from Brownian motion for time series","Recurrence analysis uses Wasserstein distances for multiscale patterns","Brownian motion informs Wasserstein recurrence networks in paleoclimate","Scale-invariant 1-Wasserstein recurrences in irregular time series data"],"cache_read_input_tokens":2112,"weakest_assumption_plain":"That the statistical properties of one-dimensional Brownian motion supply an appropriate null model and threshold for determining what counts as a recurrence in real paleoclimate time series.","fun_headline_variants_meta":{"raw":{"variants":["Wasserstein networks detect multiscale recurrences in climate records","1-Wasserstein distance thresholds from Brownian motion for time series","Recurrence analysis uses Wasserstein distances for multiscale patterns","Brownian motion informs Wasserstein recurrence networks in paleoclimate","Scale-invariant 1-Wasserstein recurrences in irregular time series data"]},"model":"grok-4.3","cost_usd":0.005304,"raw_usage":{"total_tokens":2600,"prompt_tokens":742,"num_sources_used":0,"completion_tokens":85,"cost_in_usd_ticks":53037000,"prompt_tokens_details":{"text_tokens":742,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":1773,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":742,"tokens_out":85,"duration_ms":8463,"temperature":1.0,"reasoning_tokens":1773,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-06-25T21:48:35.454774+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"Empirical 1-Wasserstein distance distributions computed on paleoclimate records that deviate markedly from the scale-invariant form predicted by Brownian motion would falsify the threshold rule.","supporting_citations":[],"review_version":1}