{"id":"4a351e3b-6459-4064-833a-4497ddeaab26","arxiv_id":"2501.03327","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Debye mass resummation reduces bound-state dark matter depletion by up to a factor of two relative to fixed-order NLO, changing relic abundance predictions by a few percent.","lead":"This paper computes how the thermal Debye mass of a dark photon affects the formation and destruction of dark matter bound states in the early universe. It finds that properly resumming Debye effects gives smaller corrections to the predicted dark matter abundance than the standard fixed-order calculation, reducing the depletion by up to a factor of two.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central numbers rest on an HTL expansion whose expansion parameter is O(1) for the nf=2 benchmark: T/m_D ≈ 1.09, so the m_D∼ΔE rates and relic-density shifts are not yet controlled quantitatively.","rationale":"The Pith Reader identified the marginal satisfaction of T ≫ m_D as the weakest assumption, and my stress test lands on the same point. This is genuinely load-bearing because every quantitative rate in the new m_D∼ΔE regime is computed with leading-order HTL propagators whose validity requires T ≫ q0, |q| ∼ m_D. For the n_f = 2, α = 0.1 benchmark, T/m_D ≈ 1.09, so the expansion parameter is O(1) and the numerical percentages quoted in the abstract are not protected by a systematic small parameter. I did not find an internal inconsistency in the EFT derivation itself: the scale separation argument is coherent, the cancellation of divergences is explicitly shown, and the qualitative conclusion that fixed-order NLO overestimates the rates is well supported by the structure of the calculation. The concern is therefore about the numerical robustness of the central claim, not about its logical soundness. Because the reader already rendered a CONDITIONAL verdict on precisely this basis, and because the proposed check would either validate or refute the quantitative impact, I do not change the verdict. The condition is: demonstrate that the full one-loop polarization, rather than its HTL limit, does not materially change the computed rates and relic-density shifts for the quoted benchmark points.","tokens_in":44784,"tokens_out":11802,"duration_ms":120003,"concrete_test":"Recompute the m_D∼ΔE self-energy (3.23)–(3.40) using the full one-loop retarded polarization tensor (B.11)–(B.14), retaining the exact q0/|q| dependence, and resumming the propagator as in (A.7), instead of using the HTL-expanded forms (B.20)–(B.23). Do this for n_f = 2, α = 0.1 and also for n_f = 1, α = 0.1, then recompute figures 6, 7, 11 and 12. If the resulting relic-density ratios R_ΩDM shift by less than about 0.5 percentage points, the marginal hierarchy is not numerically damaging; if they shift by 1–2 percentage points, or the factor-of-five rate ratios change substantially, the quantitative conclusions require a smaller-coupling benchmark or a quantified uncertainty band.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim depends on the condition (2.3), T ≫ m_D, stated equivalently as sqrt(4π n_f α/3) ≪ 1. The HTL-resummed propagators (B.31) and (B.32), used to derive the m_D∼ΔE self-energy (3.40) and hence eqs. (3.43) and (3.49), are obtained from the one-loop polarization expanded in q0, |q| ≪ T (appendix B.2). But the benchmarks used for the headline percentages are n_f = 1, 2 with α(2M) = 0.1. For n_f = 2, m_D/T = sqrt(8π·0.1/3) ≈ 0.92, so T/m_D ≈ 1.09; the expansion parameter (m_D/T)^2 ≈ 0.84 is not small. The paper explicitly excludes the strongly coupled case T ∼ m_D (sections 3.2 and 5), yet n_f = 2 sits essentially at that boundary. At this value, neglected next-to-leading corrections to the HTL polarization are formally O(1), so the claimed factor-of-five rate suppression and the 4.8–6.3% versus 7.3–11.0% relic-density differences could shift at the order of the effect itself. No scale-variation or higher-order estimate is supplied to bound this. The qualitative direction (resummation reduces the NLO overestimate) is plausible, but the quantitative reach of the abstract is hostage to this hierarchy.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a thermal EFT treatment of bound-state formation, dissociation, and bound-to-bound transitions for heavy Dirac fermion dark matter interacting with a dark photon and nf light dark fermions. It assumes the hierarchy M ≫ sqrt(MT) ≫ Mα ≫ T and T ≫ m_D, and analyzes the regimes m_D ≫ ΔE, ΔE ≫ m_D, and m_D ∼ ΔE, deriving closed-form or numerically integrable expressions for the rates after HTL-resumming the dark photon propagator. The main quantitative result is that the fixed-order NLO treatment overestimates the inelastic rates by up to a factor of five near freeze-out and therefore over-predicts dark matter depletion, while Debye-resummed rates give relic-density corrections of 2.5–3.5% (1S only) and 4.8–6.3% (n ≤ 2 with transitions) versus 3.7–6.0% and 7.3–11.0% at NLO. The calculation is first-principles, with explicit verification of scale cancellation in the key self-energy result, and it goes beyond earlier work by treating the m_D ∼ ΔE window without an assumed hierarchy.","tokens_in":45031,"tokens_out":10208,"duration_ms":100412,"significance":"If the quantitative claims are controlled, this is an important step for thermal DM freeze-out computations: it provides the first pNRQED-based treatment of the Debye-mass scale in the regime where m_D ∼ ΔE, gives explicit closed forms suitable for numerical evaluation (eqs. (3.30), (3.37), (3.43), (3.49)), and demonstrates that the fixed-order NLO treatment can misestimate depletion. The derivation is systematic and scale-cancellation is checked; the paper also explicitly includes excited states and bound-to-bound transitions, and it makes a falsifiable comparison of resummed versus fixed-order relic densities. The main weakness is that the quantitative reach of the headline benchmark numbers is limited by the strength of the hierarchy T ≫ m_D, as detailed in the major comments.","major_comments":[{"comment":"The central quantitative comparison relies on the assumption (2.3), T ≫ m_D, which is stated to hold weakly for the benchmarks. For n_f = 2 and α(2M) = 0.1, one has m_D/T = sqrt(4π n_f α/3) ≈ 0.92, so (m_D/T)^2 ≈ 0.84 and the HTL expansion parameter is of order one. The resummed propagators (B.31) and (B.32) are leading-order in this expansion, so neglected corrections are formally O(1) with respect to the very effects that produce the quoted 4.8–6.3% versus 7.3–11.0% difference in the relic density. The authors themselves exclude T ∼ m_D in §3.2 and §5. A revision should either restrict the quantitative claims to benchmark points with a clearly small m_D/T (for example m_D/T ≲ 0.3), or provide an estimate of the next-order thermal corrections and demonstrate numerically that the quoted percentages are stable under them. This is a load-bearing issue because it affects the central numerical message, not just the presentation.","section":"§2 (eq. (2.3)), §3.1.4 (eqs. (3.43), (3.49)), §5, Figs. 11–12"},{"comment":"The sentence 'The NLO terms in the second lines of eqs. (3.41), (3.42) and (3.43) are suppressed by nf α/π with respect to the LO radiative photon emission' is not correct as stated. In eq. (3.41), for example, the second line contains (m_D/(2ΔE_p^n))^2 times a bracket of order one with no explicit nf α/π factor. In the regime T ≫ m_D ≫ ΔE this term scales as (m_D/ΔE)^2 relative to the LO term and can be parametrically large, as the Landau-damping contribution in figure 5 indeed shows. The claim of suppression should be restricted to the vacuum-polarization term in the first line of these equations, or reworded to distinguish the parametric enhancement (~(T/ΔE)^2) from the coupling suppression. As written, the sentence contradicts the paper's own numerical finding that bath-particle scattering dominates at high temperatures.","section":"§3.2, paragraph after eq. (3.43)"},{"comment":"The quantitative boundary of the quoted numbers is not stated consistently. The abstract and §3.2 quote a factor of up to five overestimation of the rates, while the conclusions say 'a factor of order three'. Moreover, §5 states that the parameter choices satisfy 'T > m_D' at all times, but the formal hierarchy used is T ≫ m_D, and for n_f = 2 the initial value is already very close to T ≈ m_D. Please state the precise quantitative criterion used to define the validity range of the quoted numbers, including the effect of the one-loop running coupling on m_D/T at the lower temperatures shown in the figures.","section":"§5, abstract, and figure captions"}],"minor_comments":[{"comment":"The condition written as 'sqrt(π n_f α) ∼ 1' differs from the hierarchy condition (2.3), which is sqrt(4π n_f α/3) ≪ 1; include the factor 4/3 for consistency.","section":"§3.2, final paragraph"},{"comment":"The expression '−2/3 α r_i (ΔE)^2 r_i T' is hard to parse because r_i and ΔE are operators; please insert brackets or state explicitly that this is a matrix element between scattering and bound states.","section":"Eq. (3.10)"},{"comment":"The statement that using (3.43) over the whole vrel range versus splitting the integral changes the result by at most 4% is useful, but the text does not explain why the splitting effect becomes smaller in the effective cross section; a one-sentence explanation of the compensation in eq. (4.2) would help.","section":"§3.2, figure 4 discussion"},{"comment":"The function X1(ΔE_p^n, μ) is defined only by reference to (C.12); since X2 is written explicitly in (C.17), it would be clearer to display X1 explicitly as well.","section":"Appendix C, eq. (C.16)"},{"comment":"The statement that the hierarchy (2.2) holds 'to a good extent' at freeze-out is overly optimistic for α(2M)=0.1, where Mα/T ≈ 2.5 at T ≈ M/25; a brief comment on the size of the resulting Mα/T suppression would be appropriate.","section":"§1 and §2"}],"recommendation":"major_revision","confidential_remarks":"The paper is a serious, technically careful EFT calculation and is within the scope of JHEP. The main issue is quantitative: the headline benchmark comparison is made close to the boundary of the assumed weak-coupling hierarchy, so the quoted percentage differences are not yet controlled. I believe the authors can address this by re-scoping the quantitative claims, adding a next-order estimate, or both, and therefore I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, this is a genuinely new calculation: Debye-resummed bound-state formation and dissociation rates in the regime T ≫ m_D ~ ΔE, with the closed-form integrals X_l and X_t in eqs. (3.43) and (3.49), have not appeared before, even in the quarkonium literature. Second, the headline numbers sit at the edge of the paper's own validity condition, so the percentages should be read as indicative rather than precise.\n\nThe calculation is well done. The self-energy is split by scale (T, m_D, ΔE), the UV and IR divergences cancel between the pieces, and the µ-independence of eq. (3.43) is checked numerically. The physics message is clear: at high T the 2→2 Landau-damping scatterings dominate bound-state formation, fixed-order NLO overestimates the rates by up to a factor of five, and the overestimate largely cancels between formation and dissociation in the effective cross section, leaving relic-density shifts of a few percent. The comparison with the earlier fixed-order treatment is fair, and the citations to the prior framework papers (their refs. [39, 40], plus [32, 80]) look right.\n\nThe soft spot is real and is the one the stress-test flags. The assumed hierarchy is T ≫ m_D, equivalent to sqrt(π n_f α) ≪ 1. For the benchmark α(2M) = 0.1 this gives sqrt(π·0.1) ≈ 0.56 for n_f = 1 and sqrt(2π·0.1) ≈ 0.79 for n_f = 2, so the expansion parameter (m_D/T)^2 is 0.42 and 0.84 respectively. The authors explicitly exclude T ~ m_D as strongly coupled, yet the n_f = 2 benchmark sits essentially on that boundary. Neglected next-order HTL corrections are formally O(1) there, so the factor-of-five suppression and the 4.8–6.3% versus 7.3–11.0% comparison could shift at the order of the effect itself. The authors acknowledge this in sections 3.2 and 5 but do not quantify it: no scale-variation band, no estimate of the dropped terms. The qualitative direction is robust — fixed order over-depletes, resummation reduces the correction. The quantitative reach of the abstract is not fully controlled.\n\nSmaller items. The integrals X_l and X_t are defined but never evaluated in the text, and no code or data is shipped, so the numerics are not independently checkable without reimplementing. The abstract and figures put the fixed-order overestimate at a factor of five, while the conclusions say a factor of order three; minor, but the two should agree. Using the resummed rate over the whole v_rel range, including where T < ΔE, is an approximation; the authors check it (≤4% in the rate, <0.1% in the relic density), so I count that as minor.\n\nWho gets value: anyone computing thermal DM freeze-out with bound states, and the finite-T quarkonium crowd. It deserves a serious referee. I would send it to review, with the referee asked to push for a higher-order estimate or scale-variation band on the headline numbers, and for the numerics behind X_l and X_t.","headline":"Genuinely new Debye-resummed bound-state rates in the m_D ~ ΔE regime, with clean scale separation, but the headline benchmarks sit at the edge of the assumed T ≫ m_D hierarchy, so the few-percent numbers are indicative rather than controlled.","tokens_in":45683,"tokens_out":7320,"would_cite":true,"duration_ms":58305,"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":"During freeze-out, bound-state formation and dissociation must use Debye-resummed photon propagators; fixed-order NLO overestimates the rates by up to a factor of five.","keywords":["dark matter relic density","bound-state formation","Debye mass","hard thermal loop resummation","non-relativistic effective field theory","thermal freeze-out","Landau damping","dark photon"],"falsifier":"Compute the bound-state formation rate from the full retarded photon propagator without expanding in $q_0/T$ or imposing $T \\gg m_D$, at $m_D/T \\approx 0.92$; agreement with eq. (3.43) at the claimed few-percent level would confirm the resummation, while a significant deviation would show the weakly-coupled hierarchy is the weak point.","tokens_in":44516,"feed_emoji":"🌑","tokens_out":11456,"duration_ms":93179,"temperature":0.7,"pith_summary":"This paper tries to establish that, for heavy fermionic dark matter in a U(1) dark sector with light fermionic bath particles, the rates controlling bound-state formation, dissociation, and bound-to-bound transitions during freeze-out cannot be reliably computed order by order when the temperature is much larger than the emitted-photon energy. In that regime the thermal photon propagator develops a Debye mass, and the paper shows that resumming this mass through hard-thermal-loop propagators changes the thermally averaged cross sections and widths: the fixed-order next-to-leading-order calculation overestimates them by up to a factor of five near freeze-out. The consequence is quantitative: using Debye-resummed rates reduces the predicted correction to the dark matter relic density from 3.7-6.0% to 2.5-3.5% for the 1S ground state alone, and from 7.3-11.0% to 4.8-6.3% when n <= 2 bound states and transitions are included. A sympathetic reader would care because these are the percent-level shifts that decide whether a given dark matter mass and coupling point is consistent with the observed relic abundance.","feed_headline":"Debye resummation cuts dark-matter depletion corrections by half","feed_subtitle":"Near freeze-out, fixed-order NLO overestimates bound-state rates by up to 5x; Debye resummation is the correct treatment.","key_machinery":"The load-bearing object is the Debye-resummed dark photon propagator, obtained by hard-thermal-loop resummation of the longitudinal and transverse polarization tensors. In the real-time Schwinger-Keldysh formalism the heavy-pair self-energy splits into contributions from the scale $T$, the Debye scale $m_D$, and the binding-energy scale $\\Delta E$; the resummed propagator makes the infrared divergences and renormalization-scale dependence cancel between the scales, yielding finite expressions. In the general case $m_D \\sim \\Delta E$ the result is carried by the two integrals $X_l$ and $X_t$, which encode longitudinal Landau-damping scattering, transverse photo-emission, and transverse bath-particle scattering channels.","core_discovery":"The paper's central claim is that in the scale hierarchy $M \\gg \\sqrt{MT} \\gg M\\alpha \\gg T \\gg m_D$, with no assumed ordering between the Debye mass $m_D$ and the Coulombic binding energy $M\\alpha^2$, the correct bound-state formation cross section, dissociation width, and bound-to-bound transition widths are the Debye-resummed ones in eqs. (3.43), (3.49) and their transition analogues. These expressions interpolate between the limits $m_D \\gg \\Delta E$ and $\\Delta E \\gg m_D$ through two finite numerical integrals $X_l(\\Delta E/m_D)$ and $X_t(\\Delta E/m_D)$. When the rates are thermally averaged and inserted in the effective Boltzmann equation, the fixed-order NLO calculation is found to overestimate the individual rates by up to a factor of five near freeze-out, yet the residual effect on the relic density is only 2.5--3.5% for the 1S state and 4.8--6.3% for $n \\leq 2$ with bound-to-bound transitions, because bound-state formation and dissociation enter the effective cross section largely as a ratio.","pith_inferences":["Inference: the same Debye-resummed machinery should apply to non-abelian dark sectors and to coannihilating colored states, where a gluon Debye mass arises without extra light fermions; the closed-form integrals here provide a template, but the non-abelian case needs its own calculation.","Inference: because fixed-order NLO and Debye-resummed corrections meet at the boundary between the $T \\gg \\Delta E$ and $T \\lesssim \\Delta E$ regimes, the clean interpolation used over the whole $v_{\\rm rel}$ range deserves a dedicated matched expansion; the paper shows the numerical impact is below 0.1% on the relic density, but a systematic all-order matching would remove the residual ambiguity.","Inference: since the companion recoil corrections [40] have the opposite sign, combining the two calculations could push the total correction below the 1% observational accuracy of the relic density, which would make the benchmark predictions testable by future precision measurements.","Inference: a direct comparison with full non-equilibrium quantum-field-theory calculations of the rates in the region where $m_D \\sim \\Delta E$ would test whether the factor-of-five gap between the NLO and resummed results is quantitatively correct."],"forward_implications":["At temperatures near freeze-out, the fixed-order next-to-leading-order bound-state formation cross section and dissociation width exceed the Debye-resummed results by a factor of up to five, so earlier NLO-based depletion estimates are systematically high.","With Debye resummation the relic-density correction is 2.5--3.5% for the 1S state alone and 4.8--6.3% including $n \\leq 2$ states with transitions, roughly half the fixed-order NLO values.","In ionization equilibrium the enhancements of bound-state formation and dissociation largely cancel in the effective cross section, which is why the net relic-density effect is at the few-percent level rather than the order-of-magnitude level of the individual rates.","The resummation has a larger impact on the 2S and 2P excited states than on the ground state, so any treatment including a full tower of bound states should use the Debye-resummed rates."],"supporting_citations":[{"why":"Supplies the non-relativistic effective field theory framework and leading-order rates that this paper extends with Debye mass resummation.","marker":"[39]"},{"why":"Provides the companion center-of-mass recoil calculation whose corrections the paper compares against and finds to be of opposite sign.","marker":"[40]"},{"why":"Provides the fixed-order NLO bound-state formation computation that the paper shows overestimates the rates when Debye resummation is required.","marker":"[32]"},{"why":"Contained the earlier estimate that Debye mass effects reduce the rates by a logarithmic factor, which the paper now computes explicitly.","marker":"[80]"},{"why":"Supplies the HTL-resummed heavy-pair self-energy machinery at finite temperature used throughout the hierarchy analysis.","marker":"[55]"},{"why":"Provides the heavy-quarkonium-in-quark-gluon-plasma calculation underlying the limiting cases $m_D \\gg \\Delta E$ and $\\Delta E \\gg m_D$.","marker":"[79]"},{"why":"Gives the original bound-state formation treatment for thermal relic dark matter that serves as the leading-order baseline.","marker":"[14]"},{"why":"Supplies the effective-cross-section implementation beyond the no-transition approximation used for excited states with bound-to-bound transitions.","marker":"[33]"}],"fun_headline_variants":["Debye resummation corrects dark-matter depletion estimates","Fixed-order NLO overestimates dark-matter rates by 5x","Dark-matter relic density barely shifts with Debye resummation","Debye effects trim dark-matter freeze-out rate errors","Resummed Debye mass yields accurate early-universe rates"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument rests on the assumed hierarchy $T \\gg m_D$, which keeps the dark plasma weakly coupled; at the benchmark $n_f=2$, $\\alpha(2M)=0.1$ the ratio is $m_D/T \\approx 0.92$, so the hierarchy is only marginal and the quoted percent corrections could shift if the plasma is closer to strongly coupled.","fun_headline_variants_meta":{"raw":{"variants":["Debye resummation corrects dark-matter depletion estimates","Fixed-order NLO overestimates dark-matter rates by 5x","Dark-matter relic density barely shifts with Debye resummation","Debye effects trim dark-matter freeze-out rate errors","Resummed Debye mass yields accurate early-universe rates"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00099,"raw_usage":{"total_tokens":4239,"prompt_tokens":1029,"completion_tokens":3210,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":645,"completion_tokens_details":{"reasoning_tokens":3135}},"tokens_in":645,"tokens_out":3210,"duration_ms":20385,"temperature":1.0,"reasoning_tokens":3135,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T21:52:53.960294+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the bound-state formation rate from the full retarded photon propagator without expanding in $q_0/T$ or imposing $T \\gg m_D$, at $m_D/T \\approx 0.92$; agreement with eq. (3.43) at the claimed few-percent level would confirm the resummation, while a significant deviation would show the weakly-coupled hierarchy is the weak point.","supporting_citations":[{"cited_title":"Effective field theories for dark matter pairs in the early universe: center-of-mass recoil effects","cited_arxiv_id":"2402.12787","evidence_quote":"Provides the companion center-of-mass recoil calculation whose corrections the paper compares against and finds to be of opposite sign."},{"cited_title":"Bound-state effects on dark matter coannihilation: Pushing the boundaries of conversion-driven freeze-out","cited_arxiv_id":"2112.01499","evidence_quote":"Supplies the effective-cross-section implementation beyond the no-transition approximation used for excited states with bound-to-bound transitions."}],"review_version":1}