{"id":"b6e1001f-1c2e-403a-85a9-a58149b0bbc7","arxiv_id":"1908.02668","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Cosmological data limit a pre-recombination decaying dark matter fraction to at most about 2.7%, and future kSZ measurements could provide an independent test.","lead":"This paper studies what happens if a small fraction of dark matter decays into invisible, massless particles, and it updates the observational limit on that fraction. It also proposes that future measurements of the kinetic Sunyaev-Zel'dovich effect could independently test this decaying dark matter scenario.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The kSZ forecast in Sec 3.4 relies on a non-linear transfer function calibrated only at Gamma=1/31 Gyr^-1 and extrapolated to all decay rates, so the claimed kSZ constraints are not yet supported.","rationale":"The reader's weakest_assumption correctly identifies the kSZ non-linear transfer function extrapolation as the most load-bearing soft spot. I agree. The paper's primary constraint on f_dcdm (Sec 4.1, Table 4) comes from Planck 2015 + BAO + RSD and is computed at the linear level; it does not use T_NL. That bound is a standard MCMC result and I see no internal inconsistency in the perturbation equations or the likelihood. The footnote reporting a large chi2 improvement for a long-lived model with f_dcdm=20% is interesting but not decisive for the short-lived bound, because the joint analysis correctly weights CMB+BAO against RSD. The kSZ analysis, however, is advertised as a new contribution (Sec 5, point 4), and its quantitative predictions (Figs 12, 17, 18) depend on Eq 3.14 with a T_NL that is only validated at Gamma=1/31 Gyr^-1. The paper explicitly flags this limitation and substitutes a fixed-Gamma approximation. The direction of the resulting bias is not shown to be conservative; for short-lived models it could over-suppress the kSZ signal and artificially improve the agreement with the SPT measurement. Therefore the kSZ-based claims should be considered conditional pending a test of this approximation. The central 2.73% bound remains unaffected, so the verdict stays CONDITIONAL rather than being upgraded or rejected.","tokens_in":35917,"tokens_out":8894,"duration_ms":88376,"concrete_test":"Recompute the kSZ spectra in Fig 12 and the SPT-kSZ likelihood in Sec 4.2 replacing the fixed long-lived DCDM T_NL with two bounding choices: T_NL=1 (linear, no non-linear boost) and the standard Halofit T_NL for LambdaCDM. If C_kSZ at l=3000 for the representative models in Fig 12 shifts by more than the SPT 1-sigma error (1.3 microK^2), or if the 2-sigma allowed region in the (f_dcdm, Gamma_dcdm) plane changes qualitatively, the fixed-Gamma T_NL assumption is load-bearing and the kSZ forecasts should be revised.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The kSZ power spectrum in Sec 3.4 is computed from Eq. 3.15, which depends on Eq. 3.14 and thus on the non-linear transfer function T_NL(k). The paper adopts T_NL from Ref. [69], which is explicitly calibrated only for Gamma_dcdm <= 1/31 Gyr^-1 (about 1.05e-4 Mpc^-1), and then evaluates it at that single fixed Gamma for every model, including short-lived models with Gamma up to 3000 Mpc^-1. The paper states that this is a conservative assumption that T_NL would not deviate considerably for larger Gamma, but no evidence is given for that assumption. For short-lived DCDM, the decaying component is gone well before recombination and the late-time matter distribution should approach ordinary CDM plus dark radiation, so the true T_NL is likely closer to standard Halofit than to the long-lived DCDM fit used here. Using the long-lived T_NL would over-suppress small-scale power, over-suppress the kSZ signal, and bias the comparison with SPT in Sec 4.2 in favor of DCDM. The statements that SPT 'prefers the presence of the decaying DM' and that future kSZ measurements will provide 'further constraining power' (Secs 4.2 and 5) are therefore not robust to this extrapolation. The primary 2.73% bound on f_dcdm from Planck+BAO+RSD (Table 4) does not use T_NL and remains unaffected.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies a cosmological model in which a fraction f_dcdm of the cold dark matter decays into invisible massless 'dark radiation' with decay rate Gamma_dcdm. Using a modified CAMB/CosmoMC pipeline, the authors compute the effects on the CMB, matter growth, the baryon velocity field, and the kinetic Sunyaev-Zel'dovich (kSZ) effect. They report an updated 68% upper limit f_dcdm < 2.73% for short-lived dark matter (Gamma_dcdm/H0 > 1e4) from Planck 2015 + BAO + RSD data, a weaker limit of about 0.94% for the long-lived case, and argue that the current SPT kSZ measurement prefers a nonzero decaying fraction while future kSZ surveys would provide additional constraining power. The paper also presents physical interpretations of the signatures through the evolution of the Weyl potential.","tokens_in":36308,"tokens_out":3016,"duration_ms":35204,"significance":"If the main constraint result is correct, the paper provides a useful update on the allowed fraction of decaying dark matter, tightening the short-lived bound by roughly a factor of two relative to earlier work. The use of a standard MCMC pipeline with stated priors, external datasets (Planck, BAO, RSD, SPT), and convergence criteria is a strength, as is the transparent separation of the background, linear perturbation, and observational analyses. The kSZ investigation is interesting and potentially valuable, but its quantitative conclusions currently rest on an unvalidated extrapolation of a nonlinear transfer function, so the kSZ-related forecasts and the claim that SPT data 'prefer the presence of decaying DM' are not yet supported. The primary Planck+BAO+RSD bound does not depend on that assumption and remains the most robust result.","major_comments":[{"comment":"The kSZ power spectrum is computed using the nonlinear transfer function T_NL from Ref. [69], which the paper itself states is calibrated only for Gamma_dcdm <= 1/31 Gyr^-1. The calculation evaluates this function at a single fixed Gamma = 1/31 Gyr^-1 for all decay rates, including short-lived models with Gamma up to 3000 Mpc^-1. The paper calls this a 'conservative assumption' but provides no evidence that T_NL does not deviate considerably for larger Gamma. Physically, for short-lived DCDM the dark matter is essentially gone before recombination, so the late-time matter distribution should approach standard CDM plus dark radiation; using the long-lived DCDM fit likely over-suppresses small-scale power and hence over-suppresses the kSZ signal. This directly affects Fig. 12, the comparison with SPT in Sec. 4.2, and the conclusion that future kSZ measurements will tightly constrain DCDM. The authors should either validate the approximation (e.g., with N-body simulations for representative short-lived models), quantify the systematic uncertainty, or explicitly reframe the kSZ results as an idealized forecast and soften the associated claims.","section":"Sec. 3.4, Eq. (3.14) and Fig. 12"},{"comment":"The kSZ constraint analysis fixes all cosmological parameters to Planck 2015 best-fit values and varies only omega_c^ini, f_dcdm, and Gamma_dcdm. The resulting 2D/3D contours are therefore not a full posterior, and the statement that the SPT best-fit value deviates from the 'Planck2015+BAO+RSD'-inferred one at the 2 sigma level compares two very different constructions: a fixed-parameter chi-square map versus a marginalized MCMC posterior. The paper should clarify what the contours represent, avoid interpreting them as equivalent to a joint constraint, and ideally perform a full MCMC sampling including the kSZ likelihood rather than a conditional scan. This is important because the claim of a possible tension between kSZ and CMB data rests on this comparison.","section":"Sec. 4.2, Figs. 17 and 18"},{"comment":"The headline bound f_dcdm < 2.73% for the short-lived model is presented as an improvement over f_dcdm < 5.26% from Planck2015+BAO alone. This is internally consistent, but the paper does not discuss how sensitive the limit is to the assumed flat prior on log10(Gamma_dcdm/Mpc^-1) in the range 0.5-3.5. Since the posterior for Gamma_dcdm is only weakly constrained and the limit on f_dcdm is quoted as a 68% upper limit, a brief prior-sensitivity check (e.g., a wider or linear prior) would strengthen the claim that the bound is driven by the data rather than by the prior range.","section":"Sec. 4.1, Table 4"}],"minor_comments":[{"comment":"The text uses 'redshift distortion' where 'redshift-space distortion' (or 'redshift distortions') is standard; please correct the terminology consistently.","section":"Abstract and Sec. 1"},{"comment":"The phrase 'with a considerion of DM decay' contains a typo; it should be 'with a consideration of DM decay'.","section":"Sec. 2.1"},{"comment":"The upper limit of the kSZ integral is given as z* = 12, but the text earlier refers to t* as 'before the reionization' and the paper includes contributions up to the present. Please clarify that the z* = 12 cutoff is a numerical convergence choice and that it includes the post-reionization contribution, as the sentence 'any contributions from the higher ones can be safely neglected' suggests.","section":"Sec. 3.4, Eq. (3.15)"},{"comment":"The caption says 'bottom-left' and 'bottom-right' for panels (b) and (c); please renumber the panels or refer to them by their labels (b) and (c) to avoid ambiguity.","section":"Fig. 13 caption"},{"comment":"The transfer function W_g(k) is set to unity, but its meaning as a baryon smoothing function and the impact of this choice on the kSZ prediction should be stated explicitly, since the paper later emphasizes the importance of nonlinear corrections.","section":"Sec. 3.4, around Eq. (3.13)"}],"recommendation":"major_revision","confidential_remarks":"The paper is a solid but incremental contribution: the central constraint update is useful and the kSZ direction is interesting, but the kSZ quantitative claims are not yet robust because of the extrapolated nonlinear transfer function. The authors should be encouraged to either obtain a short-lived DCDM N-body calibration or clearly downgrade the kSZ results to a forecast with stated modeling uncertainty. The comparison of the SPT-kSZ contours with the full CMB+BAO+RSD constraints should also be redone or carefully qualified. The use of Planck 2015 rather than Planck 2018 is acceptable given the submission date, but the authors might mention the limitation."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: the short-lived DCDM bound f_dcdm < 2.73% from Planck15+BAO+RSD looks like a solid, standard MCMC result. The equations are properly attributed to Poulin, Serpico and Lesgourgues and to Audren et al., the priors and convergence criteria are stated, and the data combination is a genuine step beyond the existing 5.26% limit. The RSD section is straightforward and clearly presented.\n\nThe advertised kSZ part is the soft spot. Section 3.4 takes the non-linear transfer function T_NL from Enqvist et al. 2015, which is calibrated for Gamma <= 1/31 Gyr^-1, and evaluates it at that single Gamma for all models, including short-lived ones with Gamma up to ~3000 Mpc^-1. The paper calls this conservative, but that assumption is doing real work. For short-lived DCDM the matter distribution should approach standard CDM after the decay, so the long-lived T_NL likely over-suppresses small-scale power, over-suppresses the kSZ signal, and biases the SPT comparison in favor of DCDM. The claim that SPT 'prefers the presence of the decaying DM' is not robust to this extrapolation. The primary bound in Table 4 does not use T_NL and remains unaffected.\n\nTwo smaller points. The footnote reporting chi2 28.07 vs 11.51 for f_sigma8 refers to a 20% DCDM illustration model that the joint analysis excludes; it is not a contradiction but could mislead if taken out of context. And since the kSZ forecast is advertised as new, the authors should either validate T_NL with a quick N-body run or present the forecast as a sensitivity study with a clear caveat.\n\nWho this is for: people working on decaying DM constraints will want the updated bound. The kSZ forecast is a plausible first estimate, not a final word. It deserves a serious referee, with the kSZ section flagged for revision.","headline":"The central decaying-DM bound is defensible; the kSZ forecast rests on an unvalidated extrapolation and should be treated as preliminary.","tokens_in":36804,"tokens_out":2116,"would_cite":true,"duration_ms":22616,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["98.80.-k","95.35.+d"],"model":"deepseek-v4-flash","headline":"A decaying slice of dark matter is capped at 2.73 percent of the total, and the kinetic Sunyaev–Zel'dovich effect gives a new way to test it.","keywords":["dark matter decay","dark radiation","cosmic microwave background","redshift-space distortions","kinetic Sunyaev-Zel'dovich effect","cosmological constraints","H0 tension","sigma8 tension"],"falsifier":"Run N-body simulations of the decaying-cold-dark-matter model with $\\Gamma_{\\rm dcdm}\\gtrsim1\\,{\\rm Mpc}^{-1}$ and compare the resulting non-linear matter power spectrum with the fixed-rate fitting formula used here; if the mismatch exceeds the SPT kSZ measurement uncertainty, the predicted $C^{\\rm kSZ}_{\\ell=3000}$ suppression and the derived constraints would shift accordingly.","tokens_in":35706,"feed_emoji":"🌌","tokens_out":6139,"duration_ms":60360,"temperature":0.7,"pith_summary":"This paper asks how much of the dark matter could be an unstable component that decays into massless, invisible 'dark radiation,' and what that decay would do to the sky. Using the full Boltzmann hierarchy for the radiation produced by the decay, it computes the gravitational imprints on the CMB temperature spectrum, on the growth of structure measured through redshift-space distortions, and on the baryon velocity field that generates the kinetic Sunyaev–Zel'dovich effect. Combining Planck 2015, BAO, and RSD data, it tightens the 1σ bound on the short-lived decaying fraction from $f_{\\rm dcdm}\\lesssim5.26\\%$ to $f_{\\rm dcdm}\\lesssim2.73\\%$; for long-lived decays the RSD data do not improve the constraint. The paper also argues that future kSZ surveys will provide a further, independent handle on the model.","feed_headline":"Dark matter decay capped at 2.7 percent of the cosmic total","feed_subtitle":"Redshift-space distortion data tighten the old bound by half; the kSZ effect offers an independent probe.","key_machinery":"The engine of the calculation is a two-parameter fractional decaying-cold-dark-matter model ($f_{\\rm dcdm}$, $\\Gamma_{\\rm dcdm}$) in which the unstable component decays into massless, collisionless dark radiation, with the full Boltzmann hierarchy for that radiation evolved alongside the standard photon, baryon, neutrino, and CDM sectors. The decay acts through source terms proportional to $\\Gamma_{\\rm dcdm}$ in the dark-radiation perturbation equations; physically, the decay lowers the matter density and suppresses the Weyl potential $|\\Phi|$, and this suppression is what boosts the Sachs–Wolfe and integrated Sachs–Wolfe contributions to the CMB low-$\\ell$ spectrum, lowers $f\\sigma_8$, and reduces the baryon velocity power spectrum that enters the kSZ signal. To isolate decay-specific signatures, the illustrative calculations hold the CMB angular sound-horizon scale $\\theta_{MC}$ fixed and adjust the dark-energy density accordingly, while the Monte-Carlo analysis leaves $\\theta_{MC}$ free.","core_discovery":"The central claim is that a fractional cold dark matter component decaying into dark radiation before recombination is allowed only at a level below $f_{\\rm dcdm}\\lesssim2.73\\%$ (68% CL) for $\\Gamma_{\\rm dcdm}/H_0\\gtrsim10^4$, when Planck 2015 CMB + BAO + redshift-space distortion data are combined. For long-lived decays the current data do not improve the existing bound, with $f_{\\rm dcdm}\\lesssim0.94\\%$ from the same combination, and the model can only slightly reduce the $H_0$ and $\\sigma_8$ tensions, leaving them near $3\\sigma$. The paper further claims, for the first time, a detailed investigation of the kinetic Sunyaev–Zel'dovich effect as a probe of this model: the decay suppresses baryon peculiar velocities and therefore the kSZ power spectrum at $\\ell=3000$, so the SPT measurement prefers a nonzero decaying fraction while future kSZ surveys would constrain the parameter space independently.","pith_inferences":["If the kSZ preference for a non-zero decaying fraction is real, the same suppression should appear in other velocity-sensitive probes, such as pairwise kSZ measurements or velocity-reconstruction statistics; a cross-check with those would test the DCDM interpretation.","The near-degeneracy of $f_{\\rm dcdm}$ and $\\Gamma_{\\rm dcdm}$ in long-lived models suggests that any claimed bound on the fraction depends on the assumed lifetime; reporting constraints on the product $f_{\\rm dcdm}\\Gamma_{\\rm dcdm}$ as well would make comparisons between papers cleaner.","Because DCDM and a higher neutrino mass both suppress structure and kSZ, future kSZ data will need joint fits with neutrino mass to avoid misattributing one effect to the other.","The fixed-$\\theta_{MC}$ illustrations imply that some of the low-$\\ell$ CMB boost is a bookkeeping effect of adjusting $\\rho_\\Lambda$; model comparisons that do not fix $\\theta_{MC}$ may see smaller low-$\\ell$ signatures."],"forward_implications":["Short-lived decays (most DCDM gone before recombination) are pinned to $f_{\\rm dcdm}<2.73\\%$ at 1σ, cutting the previous bound by roughly a factor of two when RSD data are added.","Long-lived decays remain weakly constrained, and RSD data do not sharpen the bound; the $f_{\\rm dcdm}$–$\\Gamma_{\\rm dcdm}$ degeneracy persists because observables respond mainly to the product $f_{\\rm dcdm}\\Gamma_{\\rm dcdm}$.","The DCDM model does not substantially ease the $H_0$ and $\\sigma_8$ tensions; they persist at about the $3\\sigma$ level.","Dark-matter decay suppresses the kSZ power spectrum at $\\ell=3000$, and the current SPT measurement favours a non-zero decaying fraction, implying future kSZ surveys can act as an independent test of the model."],"supporting_citations":[{"why":"Supplies the fractional DCDM model and its background/perturbation equations, which the paper extends to RSD, bulk flow, and kSZ observables.","marker":"[37]"},{"why":"Provides the N-body-calibrated non-linear transfer function $T_{NL}$ used to compute the DCDM-suppressed kSZ power spectrum.","marker":"[69]"},{"why":"Introduces the gauge-invariant perturbation variables ($m_{\\rm cont}$, $m_\\psi$, $m_{\\rm shear}$) and the treatment of dark-radiation perturbations adopted in the hierarchy.","marker":"[35]"},{"why":"The numerical cosmological solver modified to evolve the DCDM background and perturbation equations.","marker":"[71]"},{"why":"Planck 2015 CMB temperature and polarization likelihoods used in the joint constraints.","marker":"[1]"},{"why":"SPT measurement of the kSZ power at $\\ell=3000$ used to derive the kSZ-based constraints.","marker":"[93]"},{"why":"Source of the $f\\sigma_8$ growth-rate data compilation used in the RSD likelihood.","marker":"[50]"},{"why":"BOSS DR12 RSD measurements providing several of the high-redshift $f\\sigma_8$ data points.","marker":"[80]"}],"fun_headline_variants":["Dark matter decay bound halved to 2.7% with new data","kSZ effect to probe decaying dark matter in future surveys","Decaying dark matter limited to 2.73% for short-lived cases","Dark matter decay: slightly eases Hubble and sigma8 tensions","New constraints halve allowed dark matter decay to 2.7%"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The calculation assumes that the non-linear clustering correction fitted from simulations for very long-lived dark-matter decay remains accurate for the much faster decay rates studied here, even though the code evaluates it at a single fixed slow rate.","fun_headline_variants_meta":{"raw":{"variants":["Dark matter decay bound halved to 2.7% with new data","kSZ effect to probe decaying dark matter in future surveys","Decaying dark matter limited to 2.73% for short-lived cases","Dark matter decay: slightly eases Hubble and sigma8 tensions","New constraints halve allowed dark matter decay to 2.7%"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001016,"raw_usage":{"total_tokens":4344,"prompt_tokens":1057,"completion_tokens":3287,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":673,"completion_tokens_details":{"reasoning_tokens":3194}},"tokens_in":673,"tokens_out":3287,"duration_ms":22001,"temperature":1.0,"reasoning_tokens":3194,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:38:55.387629+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run N-body simulations of the decaying-cold-dark-matter model with $\\Gamma_{\\rm dcdm}\\gtrsim1\\,{\\rm Mpc}^{-1}$ and compare the resulting non-linear matter power spectrum with the fixed-rate fitting formula used here; if the mismatch exceeds the SPT kSZ measurement uncertainty, the predicted $C^{\\rm kSZ}_{\\ell=3000}$ suppression and the derived constraints would shift accordingly.","supporting_citations":[],"review_version":1}