{"id":"cdc159de-ca14-4c7e-9d64-4a3161d77484","arxiv_id":"2502.02636","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"BOSS full-shape data do not tighten Planck bounds on velocity-dependent dark matter-baryon scattering, but adding a DES S8 prior yields a greater than 2 sigma preference for a velocity-independent cross section and lowers the S8 tension from 2.59 sigma to 1.47 sigma.","lead":"Dark matter that scatters off ordinary matter with a velocity-dependent cross section leaves a scale-dependent imprint on the growth of cosmic structure. This paper analyzes galaxy clustering from BOSS together with Planck and a weak lensing prior, and finds a mild, around 2 sigma, preference for such scattering in the velocity-independent case, plus a reduced S8 tension.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 2σ IDM preference rests on equating a Gaussian S8 prior with the full DES-Y3 likelihood; this equivalence is untested for IDM's scale-dependent power suppression.","rationale":"The reader's CONDITIONAL verdict is well aligned with my reading. The central claim depends on a compressed DES prior whose equivalence to the full DES-Y3 likelihood is asserted, not demonstrated, for the specific IDM models considered. This is the most load-bearing concern because it directly controls whether the headline 2σ preference is evidence for scattering or an artifact of the prior. The proposed full-likelihood check would settle it. I do not see a reason to reject the paper; the analysis is careful and the limitations are transparently stated, including the approximate n=-2, fχ=10% case and the IR-resummation regularization in Appendix G. The concern strengthens the need for the robustness check but does not change the reader's conditional verdict.","tokens_in":32785,"tokens_out":3699,"duration_ms":38856,"concrete_test":"Re-run the Planck+BOSS+n=0 analysis with the DES-Y3 3x2pt likelihood (or its public likelihood) evaluated on IDM power spectra, replacing the S8=0.776±0.017 prior. If the σ0 posterior retains a 95% lower bound above ~0.2×10^-26 cm^2, the preference is robust; if it becomes consistent with zero, the headline claim reduces to a prior effect. This is a finite computation with the public DES likelihood and the modified CLASS code.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 3 asserts that 'Imposing a prior on S8 is equivalent to adding the complete DES-Y3 dataset to our analysis.' This equivalence is the load-bearing step: without the prior, Planck+BOSS shows no preference for interactions, and in fact the n=0 posterior broadens (Table 2). For the equivalence to hold, the DES-Y3 likelihood must depend on cosmology only through S8. The cited robustness examples (ΛCDM, WDM, EDE) are not sufficient, because IDM with n=0, fχ=100% produces a scale-dependent cutoff in P(k) at k≳0.1 h/Mpc (Fig. 1), altering the shape of the lensing kernel over the k-range DES probes, not merely its amplitude. If DES-Y3 contains shape information beyond S8, a full likelihood could constrain σ0 differently, and the 2σ preference (Δχ²=-6.02, Table 1) could be a prior artifact rather than evidence for scattering. The authors acknowledge that the full DES likelihood is left to future work, which is precisely the gap.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constrains dark matter-proton elastic scattering with a velocity-dependent momentum-transfer cross section sigma = sigma_0 v^n using modified CLASS and CLASS-PT, Planck 2018 CMB data, BOSS DR12 full-shape galaxy clustering, and a Gaussian prior on S8 from DES-Y3. It explores n = -4, -2, 0, 2, 4 for 100% interacting DM and n = -2, 0, 2, 4 for a 10% interacting fraction. The main findings are that BOSS full-shape data do not significantly change the Planck-only bounds on sigma_0, that the n=0 model with the DES S8 prior yields a >2 sigma preference for non-zero sigma_0 (marginalized maximum sigma_0 = 1.47e-26 cm^2, 95% lower bound 0.234e-26 cm^2), and that this model reduces the Planck-DES S8 tension from 2.59 sigma in LCDM to 1.47 sigma (Table 4).","tokens_in":101,"tokens_out":6049,"duration_ms":122875,"significance":"If the central claim holds, the paper provides the first EFT-of-LSS analysis of velocity-dependent DM-baryon scattering and offers concrete evidence that a scale-dependent suppression of the linear matter power spectrum at k approximately greater than 0.1 h/Mpc may be preferred by current CMB, galaxy clustering, and weak lensing data. The analysis is built on validated public tools (modified CLASS, CLASS-PT, MontePython), publicly available likelihoods, full posterior tables, and explicit discussion of limitations, including the approximate treatment of the n=-2, f_chi=10% case. The main result, however, rests on the substitution of the full DES-Y3 likelihood by an S8 prior, an equivalence that is not validated for IDM's scale-dependent power suppression; consequently the headline preference is conditional on that untested step.","major_comments":[{"comment":"The statement 'Imposing a prior on S8 is equivalent to adding the complete DES-Y3 dataset to our analysis' is load-bearing, because without the prior the Planck+BOSS analysis shows no preference for interactions and the n=0 posterior actually broadens (Table 2). The equivalence requires that the DES-Y3 likelihood be sensitive to cosmology only through S8. For n=0, f_chi=100%, IDM produces a scale-dependent suppression of P(k) beginning around k approximately greater than 0.1 h/Mpc (Fig. 1), which changes the shape of the lensing kernel over the k-range that DES probes, not just its amplitude. The cited robustness examples (LCDM, WDM, EDE) do not have this feature, and the paper itself defers the full DES-Y3 likelihood to future work. The >2 sigma preference (Table 1, Delta-chi^2 = -6.02 for Planck+BOSS+DES) could therefore be a prior artifact rather than evidence for scattering; please validate the S8 sufficiency for IDM or replace the prior with the full DES-Y3 likelihood.","section":"Section 3"},{"comment":"The reported 'preference' is conditional on fixed parameters. In all runs m_chi is fixed (1 MeV for f_chi=100%; 1 MeV, 1 GeV, 10 GeV for f_chi=10%) and f_chi is fixed to 100% or 10%, so the marginalized preference is not marginalized over the interaction parameters of the model. Because the sigma_0 bounds scale with m_chi and the S8 shift depends on f_chi, a full parameter scan and model comparison (for example, Bayesian evidence over f_chi and m_chi) is needed to support the abstract's claim of a >2 sigma preference for non-zero interactions in a velocity-independent model. The authors acknowledge this at the end of Section 5, but this limitation directly qualifies the central claim and should be reflected in the presentation of the result.","section":"Section 4.1 and Table 3"},{"comment":"The regularisation that replaces P(k) with an analytic power law when the spectral index N < -4 is asserted to 'not bias' the predicted galaxy power spectrum, but no quantitative validation is shown. Figure 38 only displays pre- and post-regularisation spectra; it does not demonstrate that the BAO wiggles or the predicted P_gg multipoles at BOSS scales (k <= 0.2 h/Mpc) are unaffected for the n >= -2, f_chi=100% models. Since the BOSS data visibly change the n=0 posterior (Table 2), this unchecked approximation could affect the very model for which the >2 sigma preference is claimed.","section":"Appendix G"}],"minor_comments":[{"comment":"The text says the S8 tension is reduced by 40% in the n=0 case, while the Table 4 caption says '~50% lower'; the actual reduction from 2.59 sigma to 1.47 sigma is 43%. Please make these statements consistent.","section":"Section 4.2 and Table 4"},{"comment":"In the n=-2, f_chi=100%, Planck+BOSS+DES block, the S8 row reports 95% lower/upper values '0.764 / 0.7967', which appear to be the sigma_8 entries; please check the alignment of the 95% bounds.","section":"Table 6"},{"comment":"The normalization factor c_n in Eq. (2.2) is garbled in the typeset formula; please rewrite it unambiguously.","section":"Eq. (2.2)"},{"comment":"The caption says the cross section is at its '5-sigma limit from BOSS', but the text elsewhere uses 95% CL upper limits; please clarify which quantity is shown.","section":"Figure 7 caption"},{"comment":"For the n=-2, f_chi=10% model, the paper explicitly states that the EFT is not valid below z approximately 8 and that the results are approximate, yet Table 3 reports Delta-chi^2 values for this model without an 'approximate' marker; please add a note in the table caption or mark those entries.","section":"Appendix A and Table 3"}],"recommendation":"major_revision","confidential_remarks":"The paper is within JCAP's scope and the analysis is reproducible and generally careful. The main concern is not the statistical methodology itself but the load-bearing substitution of the DES-Y3 likelihood with an S8 prior in a model with scale-dependent power suppression; this should be either validated or relaxed before the headline 'preference for non-zero interactions' is stated. The fixed m_chi and f_chi parameters further limit the strength of the claim. The self-citation to the authors' previous work [30] is appropriate and does not indicate circularity. The paper also acknowledges external results (eROSITA, combined KiDS/DES) that may challenge the interpretation, which is commendable."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"What you should know: this is a solid and unusually transparent full-shape analysis. The genuinely new item is the first BOSS DR12 full-shape EFT treatment of velocity-dependent DM-baryon scattering, scanning n = -4, -2, 0, 2, 4 and also fractional interacting DM with f_chi = 10%. The method uses standard machinery (modified CLASS, CLASS-PT, MontePython), the code is linked, and appendices document the tight-coupling approximation for positive n and a regularisation fix for IR resummation when the power spectrum is strongly suppressed. They also openly mark the n = -2, f_chi = 10% case as approximate because it changes growth down to z ~ 8. That is real, reproducible work.\n\nWhere I would put the pressure: the advertised ~2 sigma preference for n = 0 appears only when DES-Y3 is represented by a Gaussian prior on S8 = 0.776 +/- 0.017. Planck alone and Planck+BOSS do not prefer interactions; adding BOSS actually broadens the n = 0 sigma_0 posterior. The paper's justification that the S8 prior is equivalent to the full DES likelihood is asserted, not demonstrated. The cited support is that S8 is stable across LCDM, WDM, and EDE, but the n = 0 IDM model has a scale-dependent cutoff at exactly the scales DES probes. A full DES-Y3 likelihood could easily move the sigma_0 posterior, and the 2 sigma preference is therefore conditional on an untested compression. Since the authors present it as a preference, not a detection, and explicitly say the full DES calculation is future work, I read this as a well-flagged limitation rather than a fatal flaw. Still, it is the referee's main request: either a full DES-Y3 run or a much stronger demonstration of compression-independence.\n\nOther soft spots are minor and also explicit: m_chi and f_chi are fixed, with no full model selection over them; the fractional results depend on benchmark choices. The citation pattern is fine; the self-reference to [30] is for context and benchmark choices, not circular. The stress-test concern survives reading.\n\nThis paper is for cosmologists following the S8 discussion and for anyone using EFT-based LSS to constrain DM microphysics. It deserves a serious referee; I would send it out. But I would not let the 2 sigma claim pass without the DES likelihood robustness check, or at least without a much more prominent caveat that the preference is prior-driven until then. Recommend revise with that check as the central request.","headline":"Careful, first full-shape BOSS EFT analysis for velocity-dependent DM-baryon scattering; the n=0 preference is a well-flagged but prior-dependent result that needs a full DES-Y3 check before it carries weight.","tokens_in":33583,"tokens_out":4246,"would_cite":true,"duration_ms":43183,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["95.35.+d","98.80.-k"],"model":"deepseek-v4-flash","headline":"Combined CMB, galaxy clustering, and weak lensing data favor dark matter that scatters off protons at the roughly 2-sigma level.","keywords":["dark matter-baryon scattering","interacting dark matter","effective field theory of large-scale structure","BOSS full-shape galaxy clustering","S8 tension","Planck CMB","cosmic shear prior","velocity-dependent cross section"],"falsifier":"Re-run the joint Planck+BOSS analysis replacing the DES $S_8$ prior with the full DES-Y3 likelihood and its correlations; if the preference for nonzero $\\sigma_0$ falls below roughly $2\\sigma$ or disappears, the central claim is falsified.","tokens_in":32518,"feed_emoji":"🌌","tokens_out":9132,"duration_ms":87812,"temperature":0.7,"pith_summary":"The paper argues that the standard picture of collisionless cold dark matter may be incomplete: when Planck CMB, BOSS galaxy clustering, and DES weak lensing are analyzed together, a model where dark matter elastically scatters off protons with a velocity-independent cross section is preferred over CDM at the $\\gtrsim2\\sigma$ level. The preferred cross section is around $\\sim 1.5\\times 10^{-26}\\,\\mathrm{cm}^2$ when all dark matter interacts, and a similar preference appears when only 10% of dark matter interacts. This matters because the model suppresses small-scale structure and lowers the $S_8$ tension between Planck and DES from about $2.6\\sigma$ to about $1.5\\sigma$, without worsening the Hubble tension. Adding BOSS full-shape data alone does not change the Planck-only bounds on the interaction cross section; the preference emerges only after DES weak lensing information is included as a prior on $S_8$.","feed_headline":"Dark matter scattering gets a 2-sigma preference from cosmic data","feed_subtitle":"Adding DES lensing to Planck and BOSS cuts the S8 tension nearly in half, hinting dark matter is not collisionless.","key_machinery":"The engine is the effective field theory of large-scale structure applied to a Boltzmann solver modified to include dark-matter-baryon momentum transfer. The scattering rate is set by a power-law cross section $\\sigma = \\sigma_0 v^n$, with the index $n$ encoding the velocity dependence; for a given $n$, the linear matter power spectrum acquires a scale-dependent suppression at small scales. The EFT then supplies the one-loop nonlinear matter and galaxy power spectra needed to interpret BOSS clustering, with galaxy bias and counterterms treated as nuisance parameters. Because the dark-matter-baryon scattering in these models acts mainly before recombination, late-time dynamics can be treated as CDM-like with an altered initial spectrum, and the paper adds a regularization for infrared resummation when the spectrum is steeply suppressed.","core_discovery":"The paper's central claim is that a velocity-independent elastic scattering between dark matter and protons, with a momentum-transfer cross section around $\\sigma_0 \\approx 1.47\\times 10^{-26}\\,\\mathrm{cm}^2$ (95% lower bound $0.234\\times 10^{-26}\\,\\mathrm{cm}^2$) when all dark matter interacts, improves the joint fit to Planck CMB, BOSS galaxy clustering, and DES weak lensing data relative to collisionless CDM. The improvement corresponds to $\\Delta\\chi^2_{\\rm min} = -6.02$ for the $n=0$ model, and the $S_8$ tension between Planck and DES drops from $2.59\\sigma$ in $\\Lambda$CDM to $1.47\\sigma$ in the interacting model. The paper also shows that adding BOSS full-shape data alone does not tighten the 95% confidence intervals on the cross section obtained from Planck alone; the preference appears once the DES-Y3 weak lensing constraint is included. Fractional scenarios in which 10% of dark matter interacts with protons show a similar $\\gtrsim2\\sigma$ preference for $n=0$ and $n=2$ models.","pith_inferences":["An implicit prediction is that the linear power-spectrum suppression is frozen in before recombination, so any late-time probe that measures the matter spectrum at two different redshifts should see the same transfer-function shape; a redshift-dependent cutoff would contradict the model's mechanism.","If the S8-prior equivalence were replaced by the full DES-Y3 likelihood and the preference vanished, the current $2\\sigma$ signal would be exposed as a prior-driven artifact rather than evidence for scattering.","The benchmark masses used in the paper (1 MeV, 1 GeV, 10 GeV) bracket parameter space that future sub-GeV direct-detection experiments and cosmological small-scale surveys could test simultaneously, giving a concrete target for experimental design."],"forward_implications":["If the central claim is correct, collisionless CDM is disfavored at roughly $2\\sigma$ by the combined data, meaning the concordance model would need to include dark-matter-proton scattering.","The $S_8$ tension between Planck and DES is reduced from $2.59\\sigma$ to $1.47\\sigma$ in the $n=0$, $f_\\chi=100\\%$ model, a quantitative resolution of one cosmological tension.","Adding BOSS full-shape data does not tighten the 95% upper limits on the cross section from Planck alone, which means the constraining power of current galaxy-clustering likelihoods on these models is limited.","Fractional interacting dark matter with 10% of DM scattering and $n=0$ or $n=2$ is compatible with Milky Way satellite counts while still improving the fit, so the model survives small-scale structure constraints that exclude the all-DM case.","Because $H_0$ is unchanged in these models, resolving $S_8$ with scattering does not worsen the Hubble tension, unlike many other proposed $S_8$ solutions."],"supporting_citations":[{"why":"Prior work by the same collaboration that first found an S8-tension preference for dark-matter-baryon scattering; this paper extends it with BOSS full-shape data.","marker":"[30]"},{"why":"Provides the treatment of the relative bulk velocity and the momentum-transfer rate used in the modified Boltzmann equations.","marker":"[12]"},{"why":"Supplies the perturbation-theory extension that computes the one-loop matter and galaxy power spectra for the EFT analysis.","marker":"[53]"},{"why":"Source of the DES-Y3 weak-lensing constraint that enters as the S8 prior driving the nonzero-cross-section preference.","marker":"[50]"},{"why":"The EFT-based BOSS full-shape analysis pipeline and likelihood that this paper applies to interacting dark matter.","marker":"[39]"},{"why":"Planck 2018 CMB temperature, polarization, and lensing likelihoods used for the early-universe constraints.","marker":"[58]"},{"why":"BOSS DR12 galaxy catalog and clustering measurements used as the large-scale structure data.","marker":"[33]"},{"why":"Milky Way satellite abundance constraints that exclude the all-dark-matter-interacting scenario while leaving fractional models alive.","marker":"[19]"}],"fun_headline_variants":["Dark matter may scatter off protons, cosmic data hint","S8 tension eases if dark matter interacts with baryons, data suggest","2-sigma preference for dark matter-baryon scattering from combined cosmic data","Velocity-independent dark matter scattering preferred by combined data sets","CMB, galaxy, and lensing data hint at dark matter-baryon collisions"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that compressing DES-Y3 lensing into a Gaussian prior on $S_8$ is equivalent to using the full DES likelihood for these interacting-dark-matter models, so that if $S_8$ is not as model-independent for IDM as it is for CDM, warm dark matter, and early dark energy, the apparent $2\\sigma$ preference for scattering could be an artifact of the prior.","fun_headline_variants_meta":{"raw":{"variants":["Dark matter may scatter off protons, cosmic data hint","S8 tension eases if dark matter interacts with baryons, data suggest","2-sigma preference for dark matter-baryon scattering from combined cosmic data","Velocity-independent dark matter scattering preferred by combined data sets","CMB, galaxy, and lensing data hint at dark matter-baryon collisions"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001341,"raw_usage":{"total_tokens":5512,"prompt_tokens":1070,"completion_tokens":4442,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":686,"completion_tokens_details":{"reasoning_tokens":4349}},"tokens_in":686,"tokens_out":4442,"duration_ms":31849,"temperature":1.0,"reasoning_tokens":4349,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T11:38:17.629757+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-run the joint Planck+BOSS analysis replacing the DES $S_8$ prior with the full DES-Y3 likelihood and its correlations; if the preference for nonzero $\\sigma_0$ falls below roughly $2\\sigma$ or disappears, the central claim is falsified.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Prior work by the same collaboration that first found an S8-tension preference for dark-matter-baryon scattering; this paper extends it with BOSS full-shape data."},{"cited_title":"Boddy, V","cited_arxiv_id":null,"evidence_quote":"Provides the treatment of the relative bulk velocity and the momentum-transfer rate used in the modified Boltzmann equations."},{"cited_title":"Aghanim, Y","cited_arxiv_id":null,"evidence_quote":"Planck 2018 CMB temperature, polarization, and lensing likelihoods used for the early-universe constraints."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"BOSS DR12 galaxy catalog and clustering measurements used as the large-scale structure data."},{"cited_title":"Nadler, V","cited_arxiv_id":null,"evidence_quote":"Milky Way satellite abundance constraints that exclude the all-dark-matter-interacting scenario while leaving fractional models alive."}],"review_version":1}