{"id":"109c7eb9-5e72-4ac4-b5ef-dd0053dd1e86","arxiv_id":"1908.10369","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Nonthermal dark matter produced relativistically in an early matter-dominated era remains mostly relativistic at reheating, so free streaming suppresses small-scale structure and Lyman-alpha plus Milky Way satellite data bound its production velocity.","lead":"Dark matter born from a decaying particle during a brief early matter-dominated era does not cool enough: most of it is still travelling near light speed when normal expansion begins. This fast dark matter erases small clumps that would otherwise grow, and telescope data now place upper limits on how fast it could have been made.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lyman-alpha limits depend on linearly extrapolating a WDM-calibrated likelihood to gamma ~ -1.1, so the exact alpha (and gamma_D) bounds are unvalidated.","rationale":"The reader's conditional verdict already flags the transfer-function mapping and the extrapolation of Ref. [67]'s precomputed simulations as the weakest assumption, and I agree this is the load-bearing point for the quantitative corollary. I partially differ on emphasis: the difference in shape at T(k) < 0.1 is probably not the main problem, because the Lyman-alpha data probe scales where T(k) is still ~0.8 or higher for the allowed alpha values; the more concrete risk is that the authors' gamma ~ -1.1 lies far outside the WDM-like region of the simulation parameter grid, and the stated 'linear extrapolation' of the flux power template has no physically validated error bar. The qualitative central claim -- that most nonthermally produced dark matter remains relativistic at reheating and that EMDE-enhanced structure is erased -- is robust and independently supported by the analytic birth-time distribution and the simple velocity-redshift argument. The MW satellite constraint is also less sensitive to the shape mismatch because it enters through the half-mode scale. Thus the reader's CONDITIONAL verdict remains appropriate: the paper advances the subfield, but the precise excluded region should be revisited once the likelihood extrapolation is validated or the code is released.","tokens_in":20849,"tokens_out":8312,"duration_ms":95410,"concrete_test":"Run dedicated hydrodynamic simulations of the Lyman-alpha flux power spectrum for the paper's nonthermal transfer function at representative parameter values (alpha, beta, gamma) = (0.026, ~2.42, ~-1.1) and (0.011, ~2.51, ~-1.35), and compare the resulting flux power against the prediction obtained by linearly interpolating/extrapolating the Ref. [67] simulation template at the same parameters. If the difference exceeds the observational error bars, re-derive the alpha bounds using a new simulation grid that actually covers gamma ~ -1.1; if the resulting 68% and 95% C.L. limits on alpha shift by more than ~30%, the quoted gamma_D-TRH exclusion curves in Fig. 12 should be revised.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's quantitative constraint, e.g. gamma_D <~ 550 for TRH = 10 MeV, is set by the Lyman-alpha upper bound on the cutoff parameter alpha. That bound is obtained by feeding the fitting form T(k) = [1 + (alpha k)^beta]^gamma into the MCMC likelihood of Ref. [67], with beta and gamma locked to alpha-dependent analytic functions given in Fig. 11. The authors' own fits give gamma ~ -1.1 to -1.35, whereas the thermal WDM shapes that define the reference simulation grid have gamma = -4.46. This is not a small perturbation: the high-k slope of T(k) changes by roughly a factor of three relative to WDM. Ref. [67]'s precomputed hydrodynamic simulation grid was built for WDM-like models, and the text explicitly states that parameter values outside the template are linearly extrapolated. A linear extrapolation to gamma ~ -1.1 is an untested assumption, and the Lyman-alpha flux power spectrum at high k is sensitive to the shape of the suppression, not just the half-mode. The MW satellite bound and the free-streaming estimate depend mainly on the half-mode and are therefore more robust, but the headline Lyman-alpha exclusion inherits this extrapolation risk. No code or simulation products are released, so the likelihood evaluation cannot be independently checked. This does not threaten the paper's central qualitative claim that most EMDE-produced dark matter remains relativistic at reheating; it affects the precise excluded region in Fig. 12.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies nonthermal dark matter production during an early matter-dominated era (EMDE) preceding reheating. The authors model a decaying scalar field that produces both Standard Model radiation and dark matter particles with a common initial velocity v_D, and they derive the resulting dark matter velocity distribution and its evolution. Their central finding is that adiabatic cooling during the EMDE is inefficient: most dark matter particles are still relativistic at reheating, so their free streaming erases the perturbation growth that would otherwise occur during the EMDE. They also compute the matter power spectrum suppression using CLASS and fit the resulting transfer function to the form T(k)=[1+(alpha k)^beta]^gamma, with beta and gamma fixed as functions of alpha. Using a Lyman-alpha forest likelihood from Ref. [67], they derive bounds on alpha, translate them into constraints on the production Lorentz factor gamma_D as a function of reheat temperature T_RH, and find a comparable bound from Milky Way satellite abundances. The headline result is that for T_RH=10 MeV, dark matter must be produced with gamma_D lesssim 550.","tokens_in":21046,"tokens_out":3658,"duration_ms":41849,"significance":"If the quantitative bounds hold, the paper closes an important loophole in the EMDE-enhanced structure formation scenario: it shows that the most natural, non-fine-tuned production of dark matter during an EMDE yields relativistic particles whose free streaming destroys the linear-growth enhancement. The analytic derivation of the birth-time distribution and average velocity in Sections II and III is clean, self-contained, and robust, and it requires no external data. The Milky Way satellite bound is independent of the Lyman-alpha likelihood and depends mainly on the half-mode scale, making it a relatively robust quantitative result. The main weakness is that the headline Lyman-alpha exclusion relies on extrapolating a WDM-calibrated likelihood to a transfer-function shape very different from thermal WDM, so the precise excluded region in Fig. 12 is less certain than the central qualitative claim.","major_comments":[{"comment":"The quantitative Lyman-alpha bounds on alpha, and hence the headline constraint gamma_D lesssim 550 for T_RH = 10 MeV, depend on evaluating the Ref. [67] likelihood for transfer-function shape parameters that are far outside the WDM-like simulation grid. The paper itself states that parameter values outside the template are linearly extrapolated, and the fitted gamma values are approximately -1.1 to -1.35, whereas thermal WDM has gamma = -4.46. This is roughly a factor of three difference in the high-k slope of T(k), and Lyman-alpha flux power is sensitive to the shape of the suppression, not just the half-mode scale. The derived alpha bounds and the corresponding exclusion region in Fig. 12 therefore inherit an unvalidated extrapolation. The authors should either validate the likelihood for this shape with dedicated hydrodynamic simulations, or explicitly present the Lyman-alpha bounds as approximate and make the MW satellite bound the primary quantitative result.","section":"Section IV.B, Eq. (34) and Fig. 11"},{"comment":"The claim that the nonthermal transfer function is 'quite similar' to thermal WDM is based on matching the half-mode scale, but the two shapes differ substantially at T(k) lesssim 0.1 because of the different gamma. Since the Lyman-alpha likelihood is mapped through the full fitting form, the alpha approximately 2 alpha_WDM relation cannot by itself justify using the WDM-calibrated simulation grid. The MW satellite constraint in Section IV.C, which uses only khm through Eq. (38), is less sensitive to this shape difference and should be presented as the more robust channel; the paper's current emphasis on the Lyman-alpha bound gives a false impression of equal robustness for the two constraints.","section":"Section IV.B, Figs. 9-10"},{"comment":"The manuscript reports the Lyman-alpha bound as a firm result, but the reliance on linear extrapolation is disclosed only in a single sentence. No code or simulation products are released, so the likelihood evaluation for the extrapolated region cannot be independently checked. This is not a critique of the analytic derivation, but it is a load-bearing issue for the quantitative exclusion region: a reader cannot distinguish which part of Fig. 12 is anchored by actual simulations and which part is extrapolation. At minimum, the figures should mark the extrapolated region of parameter space, and the text should state how much of the reported bound lies outside the simulation grid.","section":"Section IV.B, paragraph on 'Whenever some of the parameters assume values not enclosed by the template'"}],"minor_comments":[{"comment":"There is a typo in 'Markov hain Monte Carlo' that should read 'Markov chain Monte Carlo.'","section":"Section IV.B"},{"comment":"The integral expression contains an unexplained factor 'sqrt(i)' in the elliptic-integral term; this may be a typographical artifact, but it should be clarified or corrected.","section":"Section III, Eq. (26)"},{"comment":"The caption reads '1 > a_D > a_D,2', which appears to be a typo; it should presumably be 'a_D,1 < a_D < a_D,2'.","section":"Section III, Fig. 3 caption"},{"comment":"The sentence 'We will see in Section IV that restrictions from small-scale structure on the parameter space of gamma_D and T_RH provide much stronger bounds' is followed by a period missing after 'relic abundance'; the text should be punctuated consistently.","section":"Section II.A"},{"comment":"Equation (35) gives alpha in units of Mpc/h but the numerical coefficient is written as 0.177 (lambda_fs/Mpc)^0.908 Mpc/h; it would be clearer to state explicitly that lambda_fs is in Mpc and alpha in Mpc/h.","section":"Section IV.B, Eq. (35)"}],"recommendation":"major_revision","confidential_remarks":"The paper's central physical conclusion is likely correct, but the headline Lyman-alpha exclusion is less certain than presented because it relies on linear extrapolation of a WDM-calibrated likelihood to a very different transfer-function shape. The MW satellite bound is more robust and should probably carry more weight in the final presentation. I also note that one coauthor (Murgia) is also an author of Ref. [67], which is the source of the likelihood; this is not improper, but it makes the absence of released code or validation runs more conspicuous."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe thing to know: this paper settles a question that's been hovering over EMDE dark matter. Ref. [51] assumed all dark matter was born at reheating; Miller et al. derive the actual birth-time distribution across the whole EMDE and show that for relativistic production the bulk of dark matter is still relativistic at reheating, so the much-touted EMDE perturbation enhancement is erased. That result is new, cleanly derived, and I think it's right.\n\nWhat's good: Sections II and III are a model of honest analytic cosmology. The integrals for average velocity and birth-time distribution are self-contained, the approximations are stated, and the qualitative conclusion is robust to the details—you'd need vD <~ 0.02 at birth to get even a percent of dark matter below v = 0.01 at reheating. The Milky Way satellite constraint (khm > 36 h/Mpc, leading to alpha < 0.026) is on solid ground because it depends mainly on the half-mode scale, not the detailed high-k shape.\n\nThe soft spot is the Lyman-alpha bound. The authors feed their transfer function fit into the precomputed MCMC likelihood of Ref. [67], using beta and gamma as analytic functions of alpha. But their gamma is about -1.1, while the thermal WDM templates in the simulation grid have gamma = -4.46. That's a factor-of-three difference in the high-k slope, and the text admits the likelihood is linearly extrapolated when parameters fall outside the template. The Lyman-alpha flux power spectrum is shape-sensitive, so the quoted gamma_D <~ 550 at TRH = 10 MeV carries real uncertainty. This doesn't threaten the central qualitative claim, but it means the precise excluded region in Fig. 12 should be treated as provisional until either the transfer function mapping is validated against dedicated simulations or the code is released. No code is shipped, so independent replication takes real work.\n\nBottom line: a substantive paper, worth a serious referee, and likely to be cited in the EMDE literature despite the Lyman-alpha caveat. I'd send it out; the referees should push for validation of the extrapolation or a clear statement that the Lyman-alpha bounds are indicative.","headline":"The paper convincingly shows that nonthermal dark matter produced during an EMDE remains relativistic at reheating, wiping out the EMDE perturbation enhancement; the Lyman-alpha-derived exclusion limits are real but less certain than they look because of the transfer-function shape extrapolation.","tokens_in":21649,"tokens_out":1752,"would_cite":true,"duration_ms":16859,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Dark matter produced during an early matter-dominated era is still relativistic at reheating, so free streaming erases the era's boost to small-scale structure, and observed structures bound the birth velocity ($\\gamma_D \\lesssim 550$ at…","keywords":["early matter-dominated era","nonthermal dark matter","free streaming","matter power spectrum","Lyman-alpha forest","Milky Way satellites","reheating temperature","warm dark matter"],"falsifier":"Detect structure on scales the EMDE is supposed to enhance—for instance a population of microhalos corresponding to comoving scales $\\lambda \\lesssim 30$ pc, or a matter power spectrum at $k \\gtrsim 10\\,h/\\mathrm{Mpc}$ that shows no suppression below the cold-dark-matter prediction—and the central claim that free streaming erases those perturbations is contradicted.","tokens_in":20565,"feed_emoji":"🌌","tokens_out":10627,"duration_ms":103157,"temperature":0.7,"pith_summary":"This paper considers dark matter produced nonthermally from the decay of a scalar field that dominates the universe during an early matter-dominated era (EMDE), so that all dark-matter particles start with a single relativistic velocity. The paper claims that despite adiabatic cooling, most of this dark matter is still relativistic at reheating, because new hot particles are created up to the end of the EMDE. The resulting free streaming erases the linear growth of small-scale perturbations that the EMDE would otherwise imprint. The paper then uses observations of the Lyman-$\\alpha$ forest and of Milky Way satellite galaxies to set upper limits on the birth velocity as a function of reheat temperature, e.g. $\\gamma_D \\lesssim 550$ for $T_{RH}=10$ MeV. This matters because it constrains the allowed mass hierarchy and decay properties of any model that produces dark matter this way.","feed_headline":"Nonthermal dark matter stays hot through the early matter era","feed_subtitle":"Its free streaming erases tiny structures; observations limit how fast it can be born.","key_machinery":"The load-bearing object is the birth-time distribution $f(a_D)$ of dark-matter particles, obtained from the comoving production rate $d\\hat{n}_\\chi/da_D \\propto \\rho_\\phi \\sqrt{a_D}/H_D$ during the EMDE. This distribution determines the average velocity $\\langle v^2\\rangle(a)$, the fraction of particles born early enough to be cold at reheating, and—after scaling by the combination $\\mu = \\gamma_D v_D a_{RH}/a_0$—the present-day momentum distribution that enters the transfer function. The transfer function itself is fitted with the ansatz $T(k) = [1 + (\\alpha k)^\\beta]^\\gamma$, with $\\beta$ and $\\gamma$ given as analytic functions of $\\alpha$, so the whole cosmological constraint reduces to a single cutoff scale $\\alpha$. A separate piece of machinery is the analytic free-streaming integral, expressed in terms of $\\mu$, which connects $\\alpha$ to the physical free-streaming length by a simple power law.","core_discovery":"The central discovery is that adiabatic cooling cannot rescue nonthermal dark matter in an EMDE. The average dark-matter velocity at reheating stays close to the velocity imparted at decay—for $v_D=0.99$, the rms velocity is still about 0.93—because the continuous creation of fresh relativistic particles offsets the redshift of older ones. Consequently, only a tiny fraction of the dark matter (about 0.15% for $v_D=0.5$) is slow enough at reheating to preserve the factor-of-ten perturbation enhancement that modes entering the horizon at $0.1 a_{RH}$ would otherwise enjoy, and mixed-dark-matter studies imply even that fraction is far too small to matter. On observed scales, the same free streaming suppresses power in the Lyman-$\\alpha$ forest and reduces the predicted number of Milky Way satellites. The resulting constraints are phrased in terms of the parameter combination $\\mu = \\gamma_D v_D a_{RH}/a_0$, which sets both the free-streaming length and the momentum distribution; for a reheat temperature of 10 MeV they require $\\gamma_D \\lesssim 550$.","pith_inferences":["Because the cutoff depends on the single combination $\\mu = \\gamma_D v_D a_{RH}/a_0$, any nonthermal production mechanism with the same birth-time distribution and the same $\\mu$ should show the same cutoff scale, so the bounds are transferable beyond the specific two-body decay considered.","If future Lyman-alpha measurements or satellite catalogs push the cutoff below $\\alpha \\approx 0.011\\,\\mathrm{Mpc}/h$, the allowed region would contract toward lower $\\gamma_D$; if they relax it, higher birth velocities would open up.","A broader, non-monoenergetic distribution of birth velocities could change the shape of the transfer function and shift the limits, depending on the high-velocity tail; this is a natural next step left by the paper.","If dark matter can exchange momentum with Standard Model particles after production, the velocity distribution would cool more efficiently, potentially reviving the EMDE microhalo signature; searches for microhalos at pc scales would then discriminate between this minimal model and models with interactions."],"forward_implications":["The EMDE enhancement to small-scale structure is preserved only for dark matter born with very low velocity; for a particle born at half the speed of light only about 0.15% of the population is slow enough at reheating ($v_{RH} < 0.01$), so the enhancement is effectively erased.","Lyman-alpha forest data bound the transfer-function cutoff to $\\alpha < 0.011\\,\\mathrm{Mpc}/h$ (68% C.L.) and $\\alpha < 0.026\\,\\mathrm{Mpc}/h$ (95% C.L.), which for a given reheat temperature translates into an upper limit on $\\gamma_D$; for $T_{RH}=10$ MeV, $\\gamma_D \\lesssim 550$.","Milky Way satellite counts give essentially the same 95% cutoff, $\\alpha < 0.026\\,\\mathrm{Mpc}/h$ (equivalently a half-mode scale $k_{hm} > 36\\,h/\\mathrm{Mpc}$), so the two independent small-scale probes agree.","In the absence of annihilations, matching the observed relic abundance requires the branching fraction $f$ of the decaying scalar into dark matter to be below about $10^{-4}$ in the allowed region, meaning annihilations or another depletion mechanism are needed to avoid fine-tuning.","The constraints can be recast as limits on the scalar decay rate $\\Gamma_\\phi$ and on the parent-daughter mass hierarchy ($m_\\phi = 2\\gamma_D m_\\chi$), connecting the cosmological bound to particle-physics parameters."],"supporting_citations":[{"why":"Establishes the EMDE linear-growth enhancement and the slow-velocity criterion that this paper's velocity distribution must satisfy.","marker":"[50]"},{"why":"Argued relativistic dark matter from scalar decay erases the enhancement but assumed production at reheating; the birth-time distribution here corrects that assumption.","marker":"[51]"},{"why":"Supplies the comoving production-rate procedure used to construct the dark-matter birth-time distribution.","marker":"[61]"},{"why":"The linear perturbation code used to compute the nonthermal transfer functions from the momentum distribution.","marker":"[65]"},{"why":"Provides the transfer-function fitting form and shape conventions used to compress the model to the cutoff parameter alpha.","marker":"[66]"},{"why":"Provides the Lyman-alpha likelihood, hydrodynamic simulation grid, and marginalized bounds on alpha that set the 68% and 95% limits.","marker":"[67]"},{"why":"Gives the Milky Way satellite bound on the half-mode mass that the paper converts into the independent alpha < 0.026 Mpc/h constraint.","marker":"[58]"},{"why":"Supplies the warm-dark-matter subhalo mass-function suppression formula linking the satellite abundance to the cutoff scale.","marker":"[72]"}],"fun_headline_variants":["Nonthermal dark matter stays hot, erasing small-scale structure","Early matter era leaves dark matter too fast for small structures","Observations cap speed of dark matter born in early matter era","Hot dark matter from EMDE free-streams away tiny structures","Free streaming of nonthermal dark matter limits satellite galaxies"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that the model's transfer function is faithfully represented by the fitting form $T(k) = [1 + (\\alpha k)^\\beta]^\\gamma$ with $\\beta$ and $\\gamma$ fixed as functions of $\\alpha$, and that the Lyman-$\\alpha$ likelihood computed from precomputed hydrodynamic simulations remains valid when extrapolated to this shape; if the true shape differs where $T(k)$ is small, the quoted $\\alpha$ and $\\gamma_D$ limits would shift.","fun_headline_variants_meta":{"raw":{"variants":["Nonthermal dark matter stays hot, erasing small-scale structure","Early matter era leaves dark matter too fast for small structures","Observations cap speed of dark matter born in early matter era","Hot dark matter from EMDE free-streams away tiny structures","Free streaming of nonthermal dark matter limits satellite galaxies"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000772,"raw_usage":{"total_tokens":3451,"prompt_tokens":1014,"completion_tokens":2437,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":630,"completion_tokens_details":{"reasoning_tokens":2354}},"tokens_in":630,"tokens_out":2437,"duration_ms":16298,"temperature":1.0,"reasoning_tokens":2354,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T10:45:47.170101+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Detect structure on scales the EMDE is supposed to enhance—for instance a population of microhalos corresponding to comoving scales $\\lambda \\lesssim 30$ pc, or a matter power spectrum at $k \\gtrsim 10\\,h/\\mathrm{Mpc}$ that shows no suppression below the cold-dark-matter prediction—and the central claim that free streaming erases those perturbations is contradicted.","supporting_citations":[],"review_version":1}