{"id":"8679451a-8cb5-41f8-ae9b-711cf00ef535","arxiv_id":"2509.01540","paper_version":9,"verdict":"REJECT","confidence":"MODERATE","novelty_score":3.0,"correctness_risk":"high","formal_verification":"none","parameter_count":5,"one_line_summary":"The manuscript as provided is a methods comparison on synthetic data; the abstract's El Niño/solar-forcing discovery is absent from the body.","lead":"This preprint's abstract claims its Discrete Chi-square Method (DCM) detects solar forcing in El Niño records and outperforms official forecasts. The provided full text, however, only presents simulated comparisons of DCM against a discrete Fourier transform and contains no El Niño analysis, no solar forcing test, and no forecast comparison.","discovery_kind":"unclear","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The El Niño/solar-forcing claim is absent from the paper: the body contains only synthetic DCM-vs-DFT comparisons, so the abstract's central causal assertion is unsupported.","rationale":"The reader's strongest_claim correctly identifies the abstract's causal assertion as the central claim, and their rationale notes the body does not support it. My concern is the same: the claimed El Niño/solar discovery is entirely missing from the manuscript. However, the reader's weakest_assumption focuses on the Gauss-Markov/grid-search justification in Section 4.4, which is a secondary methodological issue; the missing evidence is more fundamental. I credit the paper's reproducible synthetic-data experiments and code availability, but those do not speak to the El Niño claim. If the manuscript were reframed as a methods comparison, the verdict could be different, but as submitted its headline claim is unsupported. Therefore the existing REJECT verdict stands.","tokens_in":29586,"tokens_out":4577,"duration_ms":58688,"concrete_test":"Run the published DCM code on a standard Niño-3.4 monthly SST anomaly record (e.g., NOAA ERSSTv5, 1950–2025), using the control-file conventions described in the paper, and test for the claimed multi-periodic 'Big wave' signal. Compare the detected minima against sunspot-cycle minima and evaluate one-step-ahead forecast error against an AR(1)+trend baseline. If no solar-aligned periodicity is found or forecasts do not beat the baseline, the abstract's causal claim is unsupported.","verdict_should_be":"REJECT","load_bearing_attack":"The paper's most load-bearing claim is the abstract's causal statement: 'Only solar forcing, not chaotic ocean-atmosphere coupling, can cause the Big wave cooling the Pacific Ocean at sunspot minima.' The full text does not contain this analysis. It is a methods paper comparing DCM and DFT on seven simulated data sets (Sections 3.1–3.7), with Tables 1–7 and Figures 1–7. No El Niño index, sea-surface temperature time series, sunspot record, fitted 'Big wave' model, forecast comparison, or causal analysis appears anywhere in the results. The only links to real solar/ENSO data are references to 'Jetsu (2025)' and a footnote to a submitted manuscript ('Do planets cause the sunspot cycle?'), neither of which is included or summarized. The manuscript's own page-1 abstract is about DCM versus DFT, not about El Niño or solar forcing. Thus, even granting every synthetic-data result, the discovery claim in the title/abstract is unverifiable from the submitted text. This is prior to the Gauss-Markov/grid-search issue: no amount of methodological soundness can establish a causal claim for data that are never analyzed. The claim also requires excluding stochastic ENSO mechanisms, which is not attempted here.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript, as represented by the full text, is a methods paper. It proposes the Discrete Chi-Square Method (DCM): a two-stage grid search over frequency combinations, followed by linear least-squares fits at each grid point, bootstrap error estimates, Fisher-test model selection, and a Prediction-test on a held-out slice. The paper compares DCM with the Horne and Baliunas DFT on seven simulated datasets that combine pure-sine or double-wave signals, polynomial trends, close frequencies, and periods longer than the data window; it reports that DCM recovers the simulation parameters (with some exceptions) while DFT fails. The supplied arXiv abstract, however, makes much stronger claims: application to El Niño time series, detection of a 'Big wave', deterministic solar forcing, and forecasts outperforming official agencies. The full text contains no such real-data analysis.","tokens_in":29972,"tokens_out":5892,"duration_ms":66344,"significance":"Strengths first: the paper makes its synthetic data and control files available on Zenodo, gives explicit model equations, and demonstrates a plausible computational strategy for period searches with trends. If the simulation results were replicated, the method could be a useful addition to periodogram-based tools. The significance for a general statistical audience is, however, undermined by three problems. First, the headline discovery claim is unsupported by any real-data analysis in the submitted text. Second, the central 'cannot fail' argument misuses the Gauss-Markov theorem and is contradicted by the paper's own Table 5. Third, the model-selection protocol includes a post hoc rejection of 'unstable' models, which needs formalization before one can interpret the reported high success rates. As submitted, the manuscript cannot justify its advertised conclusions.","major_comments":[{"comment":"The abstract asserts that 'Only solar forcing, not chaotic ocean-atmosphere coupling, can cause the Big wave cooling the Pacific Ocean at sunspot minima' and that DCM forecasts outperform official agencies. The full text does not contain an El Niño index, sea-surface temperature series, sunspot series, fitted 'Big wave' model, or forecast comparison. Sections 3.1-3.7 analyze only simulated data; Section 3.10's Prediction-test illustration uses Model 3 synthetic data. The only links to real data are a footnote to a submitted manuscript and references to Jetsu (2025), neither of which is summarized or included. The causal claim is therefore not verifiable from this submission.","section":"Abstract; Sections 1-4"},{"comment":"The statement 'The Gauß-Markov theorem ensures that the model having the lowest R or chi2 is eventually always found' and the later 'DCM can fail if, and only if, the Gauß-Markov theorem is not valid. That is impossible' are not supported. Gauss-Markov concerns best linear unbiased estimation for a fixed linear model; it does not guarantee global minimization over the nonconvex, finite grid of frequency combinations defined in Eq. (18). Moreover, Table 5, column 2 (n=10,000, SN=1000) shows DCM failing to recover P2=1.9 (estimated 5.24 +/- 0.83), with wildly wrong trend coefficients; Section 3.5 acknowledges that 'DCM can fail'. The 'cannot fail' claim needs to be removed or replaced with a concrete convergence guarantee.","section":"Section 4.4; Table 5"},{"comment":"The Fisher-test is used to select model orders from the same data that are then used for the Prediction-test; this can induce selection bias, and the paper does not report how often the selected model is the true model under repeated simulation. More importantly, the rejection of 'unstable models' (UM) relies on signatures (intersecting frequencies, dispersing amplitudes, leaking periods) that are not defined by numerical thresholds. Because UM models are identified after examining fits, the claim that all non-UM models are stable is circular unless the criteria are fixed in advance and validated on independent replicates.","section":"Sections 3.8-3.9"},{"comment":"The WD-effect claim that performance depends only on n and SN, not on Delta T, is not established. For periods longer than the window, identifiability depends on the assumed trigonometric shape and on the trend order; a low-frequency component can often be absorbed by a polynomial trend. The simulations do not provide a general identifiability proof. They also use only one realization per (model, n, SN) cell, so detection probabilities and false-positive rates are not estimated; Tables 1-7 report bootstrap parameter errors for a single dataset, not sampling variation across datasets. The extreme n and SN of Models 2, 5 and 7 are acknowledged in Section 4.4, further limiting the generality of the 'time span is irrelevant' conclusion.","section":"Section 4.4; Sections 3.2, 3.5, 3.7"}],"minor_comments":[{"comment":"The title contains a spacing typo: 'Transfo rm'. The affiliation contains 'Hels inki'. These should be corrected.","section":"Title and author affiliation"},{"comment":"Cross-references to figures are wrong: Section 3.3 refers to 'Figures 1a-d' when it means Figures 3a-d, and Section 3.4 similarly refers to 'Figures 1a-d' when it means Figures 4a-d. Section 3.6 refers to 'Equation 37' for the Model 6 formula, but the displayed equation is Eq. (36).","section":"Sections 3.3 and 3.4"},{"comment":"The abstract says DCM 'can not fail', while Section 3.5 explicitly states 'DCM can fail, just like any other time series analysis method'. This internal inconsistency should be reconciled. Also, Table 8 has malformed notation such as 'g1,,1,-1', which needs cleaning.","section":"Abstract and Section 3.5"}],"recommendation":"reject","confidential_remarks":"The manuscript should be checked for version integrity: the arXiv abstract advertises an El Niño/solar-forcing analysis that is entirely absent from the full text. The author may have submitted a different intended version; if the real-data analysis exists in a companion paper (Jetsu 2025), it needs to be included and properly summarized. As it stands, the submission is not self-contained and the central advertised claim is unverifiable. In addition, the Gauss-Markov argument in Section 4.4 is a correctness error that would require substantial rewriting, not a local edit."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know: the paper you are looking at is a methods comparison, not the El Niño discovery the abstract and title promise. The full text contains no ENSO data, no sunspot series, no fitted “Big wave”, no forecasts, no causal analysis. It is seven synthetic-data models (Tables 1–7) comparing the author's DCM with one DFT version (Horne & Baliunas 1986). The solar-forcing claims live entirely in the abstract, supported only by references to unpublished work (Jetsu 2025 and a submitted manuscript). So the headline result is unverifiable from the submitted text.\n\nWhat is actually in the paper has some merit. The DCM method is laid out clearly, including the frequency-symmetry trick that reduces the grid-search combinations, a Fisher-test model-selection procedure, and a Prediction-test on a held-out slice. The author ships data and control files on Zenodo, which deserves credit. The simulations make the intended point: DCM handles the four listed DFT limitations in these particular cases, and the Fisher-test and Prediction-test work on the synthetic example. The convergence behaviour in the tables is what you would expect from least squares as n and SN grow.\n\nThe soft spots are real but uneven. The biggest is the abstract gap: no amount of simulation can support a claim about the Pacific Ocean that is never analysed. That is a desk-reject-level mismatch for any journal. Second, the Gauss-Markov claim in Section 4.4 is not justified: Gauss-Markov applies to a fixed linear model, it says nothing about global optimization over a nonconvex frequency grid. The paper even shows DCM failing in Table 5 (n=10,000, SN=1,000), which contradicts the “cannot fail” language, though the author does admit the failure in the text. Third, all simulations are single realisations with no between-seed variability; some use absurd SN values (5,000,000). Fourth, the “WD-effect” is essentially the usual consistency of least squares, dressed up as a new “sees through time” property. Fifth, the unstable-model exclusion is applied post hoc; without pre-registered criteria it is hard to know how dependent the success rate is on that filter.\n\nWho is this for? People working on periodic-signal detection in unevenly spaced data with trends might find the DCM code and the CPU-reduction symmetry trick useful, and the comparison to Horne-Baliunas DFT is a legitimate but narrow benchmark. It deserves a serious referee only after the abstract, title, and body are aligned. As submitted, the gap between the claim and the evidence is too wide to send to referees without first demanding the actual El Niño analysis. If the author resubmits with real ENSO data, with the Gauss-Markov claim corrected, and with proper multiple-realisation simulations, it would be worth a second look.\n\nMy verdict: the methods are not worthless, but this draft is not reviewable in its current form. The editor should not quietly desk-reject the whole thing, but should require the El Niño content to actually appear before a referee spends time on it.","headline":"The body is a modest DCM-vs-DFT simulation study; the abstract's El Niño and solar-forcing claims have no support in the text, so the submission overreaches badly.","tokens_in":30378,"tokens_out":2596,"would_cite":false,"duration_ms":29912,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["62M10"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper claims its Discrete Chi-Square Method recovers periodic signals where the Discrete Fourier Transform fails, and that on El Niño records it reveals a solar-forced 'Big wave' rather than purely chaotic ocean-atmosphere dynamics.","keywords":["Discrete Chi-Square Method","Discrete Fourier Transform","time series analysis","periodic signal detection","El Niño","solar forcing","sunspot cycle","ill-posed problems"],"falsifier":"A decisive test: (1) inject into a simulated series a non-sinusoidal signal whose period exceeds the window and whose true frequency falls between two grid nodes of the DCM search, with a high-order trend and realistic noise; if the recovered minimum is not the injected signal at large n and low sigma, the window-dimension guarantee fails. (2) For the climate claim: run a coupled ocean-atmosphere El Niño model with no solar input and check whether it can generate the multi-periodic 'Big wave'; if it can, 'only solar forcing' is false. Either experiment is directly executable.","tokens_in":29508,"feed_emoji":"🌊","tokens_out":13884,"duration_ms":132465,"temperature":0.7,"pith_summary":"This paper aims to establish that a computationally heavy but conceptually simple strategy — fitting a huge catalogue of linear least-squares models, one per candidate frequency combination, and keeping the minimum — can pull periodic signals out of data windows where the Discrete Fourier Transform is known to fail: signals whose periods exceed the record length, signals riding on an unknown trend, close frequencies, and non-sinusoidal shapes. The method, the Discrete Chi-Square Method (DCM), claims a 'Window Dimension effect': detection is guaranteed once the sample is large or accurate enough, regardless of the time span, so it can 'see through time'. The abstract of this version goes further, asserting that on El Niño records DCM detects a multi-periodic 'Big wave' on the warming trend, produces deterministic forecasts that outperform official agencies, and that only solar forcing — the Pacific acting as a bolometer of the solar dynamo light curve — can explain the cooling at sunspot minima. The body supplied here develops the simulation-based case (seven synthetic data sets, all recovered by DCM and all missed by the comparison DFT); the El Niño application itself is announced rather than displayed. If true, the climatological part would push El Niño modeling toward incorporating astrophysical cycles and deterministic periodic prediction, not merely probabilistic chaos.","feed_headline":"El Niño's 'big wave' is solar-driven, method claims","feed_subtitle":"A least-squares search claims to find signals Fourier misses — and to beat official El Niño forecasts.","key_machinery":"The device that carries the argument is conditional linearization. The DCM model g(t) = h(t) + p(t) is nonlinear only because the frequencies f_i sit inside trigonometric arguments; once a tested frequency combination is fixed, every remaining parameter — harmonic amplitudes and trend coefficients — enters linearly, so ordinary least squares yields a unique, stable solution for that combination (the paper's well-posedness conditions C1–C3). A grid search over ordered frequency tuples (f1 > f2 > ... > fK1, exploiting permutation symmetry) produces a catalogue of linear fits; the minimum of z supplies starting values for a final nonlinear iteration. Model selection is then Fisher-test based, w","core_discovery":"The paper claims that the Discrete Chi-Square Method (DCM) — fitting g(t) = h(t) + p(t), the sum of K1 periodic signals (each with K2 Fourier harmonics) plus a polynomial trend — solves an ill-posed nonlinear fitting problem by brute force: hold the frequencies fixed, solve the now-linear least-squares problem uniquely for every tested frequency combination, keep the combination with the smallest z = sqrt(R/n) or sqrt(chi^2/n), then iterate. On seven simulated data sets designed to hit every known DFT failure mode, DCM recovers the injected periods, amplitudes, phases, and trend coefficients, while the Horne-Baliunas DFT fails in all seven. The claimed guarantee, the Window Dimension effect,","pith_inferences":["The body of this version contains only the simulation study (Models 1–7); the El Niño forecast, the 'Big wave' detection, and the solar-forcing conclusion are asserted in the abstract without displayed data, fitted models, or forecast windows in the supplied text. A reader should treat the climatological claims as announced results pending the companion analysis.","The paper's 'cannot fail' conclusion leans on the Gauss-Markov theorem harder than the theorem reaches: it covers the least-squares fit for each fixed frequency combination, not the global minimum over a high-dimensional grid of frequency combinations, which is a nonconvex search. The gap sits in Section 4.4's sentence that the model with the lowest R or chi^2 is 'eventually always found'.","The paper itself concedes in Section 4.4 that the n and signal-to-noise values of Models 2, 5, and 7 are extreme and unrealistic for most real data, so the practical margin over DFT on real-length, real-noise records is not established by the simulations shown.","The bolometer metaphor yields a checkable prediction independent of the next forecast: if the Pacific really tracks the solar dynamo light curve, the phases of the detected multi-periodic components should lock to sunspot-cycle extrema across independent ocean indices (Niño 3.4, SOI, thermocline depth) with consistent lags. Testing that phase locking would distinguish the solar-forcing claim from "],"forward_implications":["If DCM detects signals with periods longer than the observing window, then no record need be 'long enough' in the classical resolution sense: dense, accurate sampling can reveal cycles before they complete a single repeat.","If the El Niño analysis holds, forecasting would shift from probabilistic to deterministic — a fixed multi-periodic model fitted once and extrapolated — and the paper claims such forecasts outperform official agency outlooks.","If only solar forcing can cause the Pacific cooling at sunspot minima, then future El Niño models must couple astrophysical solar and planetary cycles to ocean-atmosphere dynamics instead of treating the system as autonomously chaotic.","The four DFT limitations (period longer than window, trend, close frequencies, non-sinusoidal shape) are presented as the reason such signals were never detected before in records like the sunspot series; wherever DFT is standard, DCM is claimed to be a drop-in replacement."],"supporting_citations":[{"why":"Formulates the DCM model and supplies the unstable-model (UM) examples the method's diagnostics build on.","marker":"Jetsu (2020)"},{"why":"The specific DFT algorithm DCM is benchmarked against; also the source of the DFT significance estimate.","marker":"Horne and Baliunas (1986)"},{"why":"The least-squares optimality result the paper calls the backbone of the whole method.","marker":"Gauß (1821)"},{"why":"Extended Gauss-Markov theorem invoked to justify least squares when data errors are not normal.","marker":"Wooldridge (2010)"},{"why":"Source of the spectral resolution, leakage, and too-short-window limitations (Equations 27 and 29).","marker":"Kay and Marple (1981)"},{"why":"Source of the trend-removal requirement (Equation 28) that hampers DFT.","marker":"Nerlove (1964)"},{"why":"Source of the pure-sine restriction (Equation 30) that DFT analyses assume.","marker":"Bretthorst (1988)"},{"why":"Bootstrap technique used for all DCM parameter and model-curve error estimates.","marker":"Efron and Tibshirani (1986)"},{"why":"F-distribution behind the Fisher-test that selects the best nested model.","marker":"Draper and Smith (1998)"},{"why":"Companion sunspot-record analysis cited as evidence that DCM finds real signals and that the Predictivity-test works on real data.","marker":"Jetsu (2025)"}],"fun_headline_variants":["Sun's dynamo seen in El Niño via new chi-square method","El Niño's big wave is solar, not chaos, study claims","Discrete chi-square finds solar forcing in El Niño","Pacific bolometer: El Niño tracks solar light curve","Solar-driven El Niño? New method says yes"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The load-bearing premise is that brute-force grid search over candidate frequencies is guaranteed to find the true signals once data are dense or accurate enough; the least-squares optimality theorem the paper leans on covers each individual linear fit, but the paper gives no proof that it extends to the global search over a high-dimensional, nonconvex frequency grid.","fun_headline_variants_meta":{"raw":{"variants":["Sun's dynamo seen in El Niño via new chi-square method","El Niño's big wave is solar, not chaos, study claims","Discrete chi-square finds solar forcing in El Niño","Pacific bolometer: El Niño tracks solar light curve","Solar-driven El Niño? New method says yes"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000257,"raw_usage":{"total_tokens":1496,"prompt_tokens":904,"completion_tokens":592,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":648,"completion_tokens_details":{"reasoning_tokens":511}},"tokens_in":648,"tokens_out":592,"duration_ms":6866,"temperature":1.0,"reasoning_tokens":511,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T12:26:48.332090+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive test: (1) inject into a simulated series a non-sinusoidal signal whose period exceeds the window and whose true frequency falls between two grid nodes of the DCM search, with a high-order trend and realistic noise; if the recovered minimum is not the injected signal at large n and low sigma, the window-dimension guarantee fails. (2) For the climate claim: run a coupled ocean-atmosphere El Niño model with no solar input and check whether it can generate the multi-periodic 'Big wave'; if it can, 'only solar forcing' is false. Either experiment is directly executable.","supporting_citations":[{"cited_title":"Discrete Chi-square Method for Detecting Many Signals","cited_arxiv_id":"2002.03890","evidence_quote":"Formulates the DCM model and supplies the unstable-model (UM) examples the method's diagnostics build on."},{"cited_title":": Econometric Analysis of Cross Section and Panel Data , 2nd edn","cited_arxiv_id":null,"evidence_quote":"Extended Gauss-Markov theorem invoked to justify least squares when data errors are not normal."},{"cited_title":", Tibshirani , R","cited_arxiv_id":null,"evidence_quote":"Bootstrap technique used for all DCM parameter and model-curve error estimates."}],"review_version":1}