{"id":"dd3ae77a-9c19-40f5-a5aa-73e03c021e0a","arxiv_id":"2505.09176","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"Combined HSC, Planck, ACT, and DESI data prefer a rapid late-time suppression of structure growth at roughly 2 sigma over LambdaCDM and baryonic or neutrino alternatives.","lead":"The authors fit models of suppressed late-time structure growth to a combination of Subaru cosmic shear, Planck CMB, ACT CMB lensing, and DESI BAO data. They find that rapid growth suppression, as in the DETG model, fits better than LambdaCDM or baryonic and neutrino alternatives, with modest statistical significance.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Fixed HSC-Y3 nuisance parameters at the LambdaCDM MAP can bias the DETG preference for beta>0; the claimed ~2sigma improvement should be re-examined with nuisance marginalization.","rationale":"I agree with the reader's identification of the most fragile premise. The central claim depends on the DETG model improving the joint fit, but the cosmic shear part of the likelihood is the only component that directly responds to late-time growth suppression, and its nuisance parameters are fixed to the LambdaCDM MAP. Since IA and photo-z can mimic a low-S8 signal, the beta preference could be substantially weakened by marginalizing them. This concern is concrete and testable with a straightforward re-analysis. The paper's other limitations, such as the look-elsewhere effect over p, the use of a naive Gaussian significance, and ignored cross-covariance, are real but secondary; even the paper's own Bayes factors (ln R <= 1.1) are weak. The verdict should remain conditional pending the nuisance-marginalized re-analysis, and the current conditional verdict already reflects this need.","tokens_in":18519,"tokens_out":7775,"duration_ms":82699,"concrete_test":"Re-run the joint analysis for DETG p=3,4,5,6 (and the gamma model) with the HSC-Y3 likelihood in which the shear calibration m_i, intrinsic alignment amplitude A_IA, and photometric redshift shift parameters Delta z_i are sampled with the priors from Li et al. (2023), while keeping the Planck, ACT DR6, and DESI likelihoods unchanged. Compare the resulting beta posterior mode and 68% interval and Delta chi^2/Delta N_p against those in Table II, making sure the LambdaCDM reference also marginalizes the same nuisance parameters. If the beta mode moves toward zero or Delta chi^2/Delta N_p drops below about 1-2 for all p, the reported preference for rapid late-time growth suppression is not robust. If the preference persists, this concern is retired.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section III.B states: 'we fix the parameters to account for the systematic effects to the values at the maximum a posteriori (MAP) LambdaCDM model found in Li et al.' That means the intrinsic alignment amplitude, photometric redshift shift parameters, and shear calibration parameters are not re-fit for the DETG or gamma models. The DETG model of Eqs. (5)-(6) acts essentially as a redshift-dependent rescaling of the linear power spectrum amplitude, and Fig. 3 shows its main effect is to lower the cosmic shear 2PCFs over the angular scales used. Such a suppression is highly degenerate with the nuisance parameters: a larger IA amplitude or a shifted source redshift distribution can mimic a similar reduction in the predicted xi+/-. Because the LambdaCDM reference has already used the best-fit nuisance values, while the DETG model is not allowed to adjust them, the Delta chi^2 values in Table II (Delta chi^2/Delta N_p ~ 4-6 for p=3-6) and the resulting '~2sigma' statement in Section V may overstate the model improvement. This is the most load-bearing vulnerability of the central claim; the multiple-testing over p and the ignored cross-covariance are related but secondary issues.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper investigates whether late-time suppression of structure growth can resolve the S8 tension. It jointly analyzes HSC-Y3 cosmic shear two-point correlation functions, Planck-2018 primary CMB, ACT DR6 CMB lensing, and DESI Y1 BAO, comparing flat LambdaCDM with several extensions: baryonic feedback (HMCODE16 and HMCODE20), massive neutrinos, the gamma growth index model, and the dark-energy-tracking-growth (DETG) model of Lin et al. The central result is that the DETG model with p = 3, 4, 5, or 6 improves the chi-squared per added parameter by roughly 4-6 relative to LambdaCDM, corresponding to a claimed ~2 sigma preference for beta > 0, while baryonic and neutrino extensions are not favored. The paper concludes that models with more rapid late-time growth suppression offer a better combined fit, but it tempers this with the modest Bayes factors and the phenomenological nature of the ansatz.","tokens_in":18807,"tokens_out":5320,"duration_ms":56032,"significance":"If the result holds, it would provide evidence that a redshift-dependent suppression of structure growth, triggered by dark-energy domination, can simultaneously satisfy early- and late-universe data better than LambdaCDM or baryonic/neutrino alternatives. The analysis is technically careful: it uses public likelihoods, a consistent set of scale cuts, and a uniform treatment of the datasets across models. It also honestly reports the Bayes factors, which are considerably less favorable than the chi-squared differences. However, because the central claim rests on a fixed-nuisance comparison and a selection over the discrete parameter p, the significance is conditional on these choices. The paper is a useful exploratory step but does not yet establish a robust detection.","major_comments":[{"comment":"The cosmic shear nuisance parameters (intrinsic alignment amplitude, photometric redshift shifts, shear calibration) are fixed to the MAP values from the LambdaCDM analysis of Li et al. This is a load-bearing assumption for the central claim. The DETG model modifies the growth factor and predominantly reduces the predicted cosmic shear 2PCFs at the scales used; a similar reduction can be produced by shifting these nuisance parameters. By fixing them to LambdaCDM MAP values, the reported Delta chi^2 values for DETG (p = 3-6) and the resulting '~2 sigma' statement in Section V are not a robust model comparison. The authors should marginalize over at least the intrinsic alignment and photo-z shift parameters for each extended model, or demonstrate that the preference for beta > 0 is insensitive to reasonable variations of these nuisances.","section":"III.B, Table II"},{"comment":"The DETG model is tested for six discrete values of p (1 through 6), and the conclusion of a '~2 sigma deviation' is based on the subset p = 3, 4, 5, 6 that shows the largest chi-squared improvements. This multiple-testing selection is not accounted for; the probability of finding at least one p with such an improvement by chance is not evaluated. The Bayes factors in the same table (ln R between 0.3 and 1.1, categorized as 'barely worth mentioning' by the authors themselves) do not support a strong preference. The paper should either treat p as a free parameter with a prior, apply a trials factor, or de-emphasize the chi-squared-based significance in favor of the Bayes-factor evidence.","section":"IV.D, Table II"},{"comment":"The analysis assumes the datasets are independent and ignores all cross-covariances (Section III.B: 'We assume the different datasets are independent and ignore the cross-covariance between them'). The ACT DR6 CMB lensing map overlaps substantially with the HSC-Y3 sky area, and the Planck CMB lensing signal is physically correlated with the HSC shear field. This approximation can bias the chi-squared differences used to rank models, especially at the 2-sigma level. The authors should at least assess the sensitivity by removing one dataset at a time or by adding a crude cross-correlation term, and discuss the expected impact on the Delta chi^2 values.","section":"III.B"},{"comment":"The DETG ansatz forces alpha(z) = 1 in the matter-dominated regime by construction, so any growth modification before dark-energy domination is excluded a priori. This is a strong model assumption that shapes the interpretation of the result: the data preference for beta > 0 is only a preference for a specific 'dark-energy-triggered' redshift dependence, not for growth suppression in general. The paper should explicitly acknowledge that the comparison with the gamma-index model does not cover models with earlier-onset suppression, and that the conclusion is limited to the chosen ansatz.","section":"II.B.2, Eq. (6)"}],"minor_comments":[{"comment":"There are typographical issues: 'within the the LambdaCDM framework' appears in the abstract and the introduction, and 'In this model, The linear growth rate' in Section II.B.1 should have a lowercase 'the'.","section":"Abstract, Section I, Section II.B.1"},{"comment":"The caption describes the diagonal panels as 'lower-left' for xi+ and 'upper-right' for xi-, which is confusing given the panel layout. Please clarify which panels correspond to which combination, or reorder the panels with clear subfigure labels.","section":"Figure 3"},{"comment":"The nuisance parameters fixed to the Li et al. MAP values (intrinsic alignment amplitude, photometric redshift parameters, shear calibration) are not listed in Table I or defined in the text. For reproducibility, provide the full set of fixed values or a precise reference to the table in Li et al. where they are given.","section":"III.B and Table I"},{"comment":"The statement that Delta chi^2 / Delta N_p of about 4-6 corresponds to 'more than a 2 sigma level under a naive Gaussian-like consideration' is not a standard significance statistic; the chi-squared difference should be interpreted with the effective number of degrees of freedom, and the Wilks or AIC/BIC criteria would be more appropriate. Please either justify the significance using a proper distribution or present it as a heuristic only.","section":"IV.D, Section V"},{"comment":"HMCODE16 is used to compute the nonlinear matter power spectrum for the DETG model, although the model modifies the linear growth factor. This is an approximation; the paper should mention that the halo-model calibration has not been validated for DETG perturbations and discuss the possible systematic uncertainty on the chi-squared results.","section":"II.C, II.B.2"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a careful application of an existing phenomenological growth-suppression model (DETG from Lin et al.) to a new combination of public datasets. Its main contribution is the comparative assessment across several extensions within one pipeline. The fixed-nuisance issue and the multiple-testing over p are, in my view, the two factors that most affect the credibility of the claimed ~2 sigma preference. I would encourage the editor to require that the authors re-run the DETG analysis with at least the dominant cosmic shear nuisances marginalized, and to report the global significance after accounting for the number of p values tried. If the preference survives those tests, the paper would be a solid contribution; if not, the conclusion should be softened accordingly."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a competent, transparent model-comparison paper, and the first application of the DETG growth-suppression model to the HSC-Y3 + Planck + ACT DR6 + DESI BAO combination. The central finding—that rapidly triggered late-time growth suppression improves the joint fit by Δχ²/ΔN_p ~ 4-6—is worth taking seriously. But the headline '~2σ' is not robust as stated. The analysis fixes the HSC cosmic shear nuisance parameters (intrinsic alignment, photo-z shifts, shear calibration) at the MAP values from the ΛCDM fit of Li et al. (2023). That is the load-bearing weakness: the DETG suppression is largely a redshift-dependent rescaling of the lensing amplitude, and a model that is not allowed to adjust IA or calibration parameters will spuriously look better if the true cosmology has different growth. The paper itself reports Bayes factors that only reach ln R ~ 1.1 for the best DETG cases, which it honestly calls 'barely worth mentioning.' So the evidence is a mild preference, not a solution.\n\nWhat it does well: the model comparison is systematic—baryonic feedback via HMCODE16 and HMCODE20, massive neutrinos, γ growth, and DETG with p fixed to various integers are all examined with the same data and likelihoods. The scale- and redshift-dependence arguments in Sec. II are clear, and Fig. 3 is useful. The use of public codes and likelihoods makes the analysis reproducible in principle, though no code is shipped.\n\nSecondary issues: the look-elsewhere effect over p (six values) is not accounted for in the 2σ statement; datasets are assumed independent with cross-covariance ignored; and the Δχ²/ΔN_p heuristic is a rough guide rather than a calibrated significance. None of these kill the paper, but they all point the same direction: the preference for β>0 is less secure than the abstract implies.\n\nBottom line: this is a serious paper with a real question, but the central claim needs stronger support. If I were the editor, I would send it to a referee—the analysis is careful enough to deserve a full review—but ask the referee to focus on nuisance marginalization and multiple-testing. The authors should re-run with shear nuisance parameters free, or at least justify why the MAP values remain valid. As is, it is a useful contribution to the S8 literature, not a resolution.","headline":"A careful but over-optimistic model comparison: the DETG preference for late-time growth suppression is real at a modest level but likely inflated by fixing shear nuisance parameters to the LambdaCDM MAP.","tokens_in":19347,"tokens_out":3265,"would_cite":false,"duration_ms":29902,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper claims that rapid late-time suppression of structure growth fits combined cosmic shear and CMB data better than LambdaCDM, hinting at a possible ~2 sigma deviation from General Relativity.","keywords":["S8 tension","cosmic shear","modified gravity","structure growth suppression","DETG model","growth index","CMB lensing","baryonic feedback"],"falsifier":"Refit the DETG models with the HSC nuisance parameters left free instead of fixed to the LambdaCDM maximum a posteriori values; if the posterior on $\\beta$ returns to $\\beta=0$ or the improvement over LambdaCDM disappears, the claimed ~2 $\\sigma$ deviation from General Relativity is not robust.","tokens_in":18298,"feed_emoji":"🔭","tokens_out":5804,"duration_ms":53697,"temperature":0.7,"pith_summary":"The paper asks whether the long-standing mismatch between the low clustering amplitude $S_8$ seen in cosmic shear and the higher value inferred from the early universe can be explained by suppression of structure growth at late times. It compares four ways to produce that suppression: baryonic feedback, massive neutrinos, a modified growth index $\\gamma$, and a two-parameter 'dark energy tracking growth' (DETG) model in which growth slows as dark energy comes to dominate. Fitting the combined HSC Year 3 cosmic shear, Planck CMB, ACT DR6 CMB lensing, and DESI BAO data, the DETG models with the most rapid late-time suppression ($p=3,4,5,6$) improve the fit relative to $\\Lambda$CDM at about the $2\\sigma$ level with $\\beta>0$, while baryonic and neutrino models do not. A sympathetic reader would care because this is a concrete, parameterized way to reconcile the $S_8$ tension with a genuine modification of gravity rather than a nuisance effect.","feed_headline":"Rapid late-time growth suppression eases S8 tension","feed_subtitle":"Combining HSC shear, Planck CMB, and ACT lensing favors a ~2 sigma deviation from GR.","key_machinery":"The load-bearing object is the DETG rescaling of the linear matter power spectrum, which enters every cosmological probe through the growth factor. The model rescales the linear matter power spectrum at each redshift by $\\alpha(z)=1-\\beta[\\Omega_{\\rm DE}(z)/\\Omega_{\\rm DE}(0)]^p$, where $\\beta$ sets the present-day suppression amplitude and $p$ sets how rapidly the suppression switches on as dark energy becomes dynamically important. This scale-independent rescaling is fed through the HMCODE16 nonlinear mapping so that cosmic shear and CMB lensing predictions inherit a distinct redshift dependence that the data can separate from baryonic feedback and neutrino free-streaming.","core_discovery":"The central claim is that the data prefer a scale-independent but redshift-dependent suppression of the linear growth factor that turns on only as dark energy begins to dominate, and that the suppression must evolve quickly toward the present day. The DETG model encodes this as $\\alpha(z)=[D_{\\rm DETG}(z)/D_{\\Lambda\\rm CDM}(z)]^2 = 1-\\beta[\\Omega_{\\rm DE}(z)/\\Omega_{\\rm DE}(0)]^p$, with $\\beta=0$ recovering $\\Lambda$CDM. For $p=3,4,5,6$ the posterior favors $\\beta>0$ with an improvement of $\\Delta\\chi^2/\\Delta N_p\\gtrsim 4$, which the authors read as a potential deviation from General Relativity at roughly $2\\sigma$; the $\\gamma$ growth-index model and the DETG model with $p=1,2$ are less preferred, and neither baryonic feedback nor varying neutrino mass resolves the tension.","pith_inferences":["The reported ~2 sigma preference depends on fixing the HSC nuisance parameters to the LambdaCDM maximum a posteriori values; if those parameters were allowed to float under modified-growth cosmologies, the central value of $\\beta$ could shift and the significance could drop.","The analysis assumes the datasets are independent and ignores cross-covariance between HSC shear, Planck, and ACT; including cross-terms could either weaken or strengthen the preference.","A clean discriminator would be a growth-rate measurement at $z\\sim 1$ to $2$: DETG with $p=5$ predicts nearly LambdaCDM growth there, while the $\\gamma$ model with $\\gamma=0.65$ predicts suppression already underway, so a high-redshift null result would separate the models.","If the preference survives in next-generation lensing surveys, it would motivate physical modified-gravity theories whose growth suppression tracks the dark-energy density rather than a purely empirical fit."],"forward_implications":["If the DETG interpretation is right, the $S_8$ tension is not a systematic error in cosmic shear but a real signal of modified gravity active only once dark energy dominates.","The preferred parameter region ($p=3\\text{--}6$, $\\beta>0$) predicts growth suppression that strengthens toward $z=0$, which can be checked with redshift-resolved growth measurements from galaxy clustering and CMB lensing cross-correlations.","Models with slower suppression, such as the $\\gamma$ model and DETG with $p=1,2$, fit markedly worse, so future data should discriminate the shape of the suppression rather than just its overall amplitude.","Because baryonic feedback and massive neutrinos do not produce the right redshift dependence, the analysis points away from within-$\\Lambda$CDM fixes and toward genuine late-time physics."],"supporting_citations":[{"why":"Supplies the HSC Year 3 cosmic shear two-point correlation functions and the maximum a posteriori nuisance parameter values used for the fixed systematic parameters.","marker":"[4]"},{"why":"Proposes the dark energy tracking growth (DETG) parameterization tested in this paper.","marker":"[17]"},{"why":"Provides the modified CAMB implementation of growth-index and growth-suppression models used for the modified gravity predictions.","marker":"[16]"},{"why":"Develops the growth-rate parameterization and modified CAMB used to compute linear power spectra for the modified growth models.","marker":"[52]"},{"why":"HMCODE16 supplies the nonlinear matter power spectrum mapping used for all models, including the DETG and neutrino cases.","marker":"[34]"},{"why":"Supplies the Planck 2018 primary CMB spectra used as the early-universe anchor in the joint likelihood.","marker":"[55]"},{"why":"Supplies the ACT DR6 CMB lensing power spectrum used as a late-time structure probe.","marker":"[56]"},{"why":"Provides the companion ACT DR6 lensing map and cosmological parameter analysis that defines the lensing data vector.","marker":"[57]"},{"why":"Supplies the DESI Year 1 BAO distance measurements used to constrain background geometry.","marker":"[58]"},{"why":"Establishes the baryonic-effect scale cuts and shows the S8 tension persists when baryonic effects are included.","marker":"[23]"}],"fun_headline_variants":["Modified gravity with late growth suppression eases S8 tension","Rapid late growth suppression favors modified gravity over LCDM","S8 tension eased by redshift-dependent growth suppression","Late-time growth dip in modified gravity eases S8","Late growth suppression in modified gravity preferred for S8"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The analysis fixes the HSC cosmic shear nuisance parameters, including intrinsic alignment, photometric redshift errors, and shear calibration, to the values fitted for LambdaCDM in Li et al., so if the true cosmology has strong late-time growth suppression, those nuisance parameters would shift and the reported preference for $\\beta > 0$ could be an artifact.","fun_headline_variants_meta":{"raw":{"variants":["Modified gravity with late growth suppression eases S8 tension","Rapid late growth suppression favors modified gravity over LCDM","S8 tension eased by redshift-dependent growth suppression","Late-time growth dip in modified gravity eases S8","Late growth suppression in modified gravity preferred for S8"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001063,"raw_usage":{"total_tokens":4495,"prompt_tokens":1019,"completion_tokens":3476,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":635,"completion_tokens_details":{"reasoning_tokens":3398}},"tokens_in":635,"tokens_out":3476,"duration_ms":22478,"temperature":1.0,"reasoning_tokens":3398,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T21:37:36.638149+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Refit the DETG models with the HSC nuisance parameters left free instead of fixed to the LambdaCDM maximum a posteriori values; if the posterior on $\\beta$ returns to $\\beta=0$ or the improvement over LambdaCDM disappears, the claimed ~2 $\\sigma$ deviation from General Relativity is not robust.","supporting_citations":[{"cited_title":"ij” denote tomographic bins; e.g. “ij","cited_arxiv_id":null,"evidence_quote":"Supplies the HSC Year 3 cosmic shear two-point correlation functions and the maximum a posteriori nuisance parameter values used for the fixed systematic parameters."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"HMCODE16 supplies the nonlinear matter power spectrum mapping used for all models, including the DETG and neutrino cases."}],"review_version":1}