{"id":"fd6ff9da-1d48-4d1a-afd9-8e768cf1dfc2","arxiv_id":"2502.01729","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"ALP production from a hot Abelian plasma is computed with full 1PI-resummed propagators, giving positive rates at all momenta and revealing dominant timelike-timelike photon contributions at very soft ALP momenta.","lead":"This paper computes how axion-like particles are produced from hot photon-like plasma using a more complete version of the gauge boson propagator, finding the production rate stays positive at all momenta. At very soft momenta, a channel involving two timelike gauge bosons, previously missed, can dominate and affect dark matter structure formation bounds.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The p≲g^4T TT-dominance claim is not yet a leading-order result: the calculation omits the last-row 2PI diagrams of Fig. 2, which correct the photon width that sets the TT scaling.","rationale":"The central claim has two parts: positivity of the ALP production rate, and TT dominance at very soft momenta. The positivity part is robust: the full 1PI-resummed spectral densities are positive, no matching/subtraction is used, and any omitted cut diagrams would add positively. The TT-dominance part, however, rests on a truncation that the authors themselves describe as incomplete in exactly the regime where the claim is made. At the end of Sec. 3 they state that the last-row diagrams of Fig. 2 are needed for a consistent leading-order computation of soft ALP production, and the conclusion repeats that the behavior 'needs to be confirmed by a consistent leading order computation'. This is not an external standard imposed on the paper; it is the paper's own stated limitation. The technical reason this omission is load-bearing is that the TT contribution in Eq. (B.12) is governed by the one-loop photon width ΓTT: for p≲pc the self-energy scales as p^2/ΓTT. The omitted diagrams are precisely two-loop corrections to that width (and to the ALP-photon vertex), so they can change the TT amplitude, the crossover momentum pc, and the TT/SS ratio at the same parametric order. The fitted coefficient κI in the SS estimate and the absence of released code are lesser concerns; the SS scaling is checked numerically and the fitted constant is an estimate, not the basis of the dominance claim. The hard-momentum results, the O(10%) HTL comparison, and the positive rates across all momenta remain credible. The reader's CONDITIONAL verdict is therefore appropriate, and no verdict change is needed.","tokens_in":22212,"tokens_out":10607,"duration_ms":112166,"concrete_test":"Perform the missing leading-order computation: include the last-row diagrams of Fig. 2 (at minimum the two-loop transverse photon self-energy, or the LPM-resummed fγ→fγa contribution) and recompute Π^< from Eq. (3.7) for g1=0.35, z=0, at p/T=10^-3 and 10^-4. Compare the TT and SS components with Fig. 6 and the ΓTT≈5×10^-4T crossover. If the TT-over-SS excess and the total rate change by less than ~30%, Sec. 4(iv) stands; if the missing contribution shifts ΓTT by O(1) or changes the p-scaling, the dominance claim must be restricted to p≳pc or retracted.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's own Sec. 3 limitation statement is the load-bearing issue. The one-loop ALP self-energy with two 1PI-resummed photon propagators, Eq. (3.7), is used to claim in Sec. 4(iv) that for p≲g1^4T TT-type contributions dominate, potentially by another order of magnitude. But Sec. 3 states that the last row of Fig. 2 (soft ALP emission from external photons, e.g. fγ→fγa) is part of the leading-order contribution to the rate in exactly the soft regime, and that these diagrams are left out. The omission is not an independent channel correction: those diagrams are two-loop corrections to the photon self-energy / ALP vertex, and the TT contribution in Eq. (B.12) is controlled by the photon width ΓTT that they correct. Hence the TT amplitude, the crossover scale pc=ΓTT(k=0), and the TT/SS ratio are all determined at a level that is not demonstrably leading order. Positivity is not threatened—missing cut diagrams are positive—but the flagship soft-momentum dominance claim is explicitly provisional; the abstract's 'potentially' and the conclusion's 'needs to be confirmed by a consistent leading order computation' correctly flag this, yet the Sec. 4 item (iv) wording can be read as an established result.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper computes the thermal production rate of axion-like particles coupled to a U(1) gauge boson, using the full 1PI-resummed photon propagator in both lines of the one-loop ALP self-energy instead of HTL-resummed propagators. The authors show that the resulting rate is positive for all momenta, avoiding the negative rates obtained in matching/subtraction schemes. They decompose the rate into timelike-timelike (TT), timelike-spacelike (TS), and spacelike-spacelike (SS) contributions, and identify the momentum regimes in which each dominates. The main new quantitative claim is that at very soft momenta p ≲ g^4 T the TT contribution, enabled by the finite photon width, may exceed the SS contribution by another order of magnitude. They use these rates to update the ALP abundance and average momentum, finding only percent-level changes relative to previous work.","tokens_in":22448,"tokens_out":3557,"duration_ms":40364,"significance":"If the central soft-momentum claim survives scrutiny, the paper provides a framework for positive, physically consistent ALP production rates over the full momentum range, and it sharpens the understanding of the soft-momentum regime that determines warm-dark-matter constraints. Strengths of the paper include a careful derivation of the self-energy expression in Eq. (3.7), an explicit demonstration of gauge independence, comparison against the external benchmark of Ref. [11] and against the independent result of Ref. [54], and an unusually frank discussion of the diagrams that are not included. The omission of the last-row diagrams in Fig. 2, however, makes the flagship TT-dominance claim provisional rather than a complete leading-order result, and this needs to be reflected more consistently in the abstract and in Section 4.","major_comments":[{"comment":"The paper itself states that the diagrams corresponding to soft ALP emission from external photons (e.g. f γ → f γ a) are part of the leading-order contribution to the rate for p ≲ m_V^2/T and that these diagrams are not included. This omission is not a minor higher-order correction: the TT contribution in Eq. (B.12) is proportional to the photon width Γ_TT, and the crossover scale p_c = Γ_TT(k=0) is the same width that the omitted diagrams correct. Consequently, the magnitude of the TT contribution, the determination of p_c, and the TT/SS comparison in Sec. 4 item (iv) are all computed at a level that is not demonstrably leading order in exactly the regime where the TT claim is made. The authors acknowledge this in the conclusion, but Sec. 4 item (iv) and the abstract's 'potentially exceeding' wording do not carry the same caveat. This issue must be addressed, either by performing the missing calculation or by explicitly reframing the TT-dominance claim as a partial/beyond-LO estimate whose leading-order status remains open.","section":"Sec. 5 vs. Sec. 4 item (iv)"},{"comment":"The conclusion states that the approach 'does not strictly apply to ALP momenta p < g^4 T, where the quasiparticle description breaks down and hydrodynamics become necessary,' yet Sec. 4 item (iv) presents TT dominance for p ≲ g^4 T as one of the main findings, and the abstract highlights the p ≲ g^4 T region as a potentially order-of-magnitude enhancement. These statements need to be harmonized. If the formalism is not valid below g^4 T, then the abstract and Sec. 4 should present the p ≲ g^4 T behavior as an extrapolation of the 1PI-resummed calculation, explicitly outside the strict validity range, rather than as an established production channel.","section":"Sec. 4 and abstract"},{"comment":"The scaling estimate for the TT contribution in Appendix B.2 relies on the narrow-width pole approximation and on approximating the width Γ_TT by the one-loop photon self-energy at zero momentum. This width controls both the piecewise p-scaling in Eq. (B.12) and the crossover momentum p_c. Because the omitted vertex/self-energy corrections enter at the same level as the width that sets the TT scaling, the numerical value p_c ≈ 5 × 10^-4 T and the resulting dominance window in Sec. 4 item (iv) should be labeled as an estimate rather than a prediction until the missing diagrams are included.","section":"Appendix B.2, Eqs. (B.5)-(B.12)"}],"minor_comments":[{"comment":"The numerical factor κ_I = 8 is fitted to the numerical result, not derived. The p^{4/3} scaling is analytic, but the text should state explicitly that the prefactor 93 in Eq. (B.4) is a numerical fit, since the phrase 'using HTL resummed propagators' might otherwise suggest a parameter-free result.","section":"Appendix B.1, Eq. (B.3)"},{"comment":"The approximation of replacing the 1PI-resummed propagator by the HTL form for timelike momenta at hard ALP momentum p > T is acceptable because hard momenta are not the focus, but this should be stated in the main text near Fig. 6 rather than only in the footnote, since Fig. 6 is used for the TS comparison.","section":"Sec. 4, footnote 6"},{"comment":"The notation Π_{TT}(p) = p^0 for p ≳ p_c is dimensionally confusing; the exponent should be written as a constant with respect to p (e.g. ∼ const. or p^0) and the p^2 branch clarified, since Eq. (B.12) is not obviously identical in notation.","section":"Eq. (3.14)"},{"comment":"The text states 'p ≲ 10^{-2} T ≈ g_1^4 T' and later reports Γ_TT ≈ 5 × 10^{-4} T, but the figure axis extends to p/T = 10^{-3}. The relation between the quoted p_c and the lowest plotted momentum should be made explicit to avoid the impression that the p^2 branch is visible in Fig. 6 when it is only visible in Fig. 9b.","section":"Sec. 4, Fig. 6 discussion"},{"comment":"There are several typographical and grammatical slips, e.g. 'if Fig.2' in the caption of Fig. 2, 'the the ALP self energy' near the top of Sec. 4, and 'goverened' in Eq. (2.2). These do not affect the physics but should be corrected.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The paper is honest about its limitations, and the core positivity result plus the hard/soft momentum comparison are likely solid. The main concern is that the abstract and Sec. 4 highlight the p ≲ g^4 T TT dominance as a central finding, while the paper's own Sec. 3 and Sec. 5 state that the calculation is not a complete leading-order result in that regime and that the needed diagrams are deferred. This is fixable within the manuscript's scope by rephrasing the claims as estimates and explicitly marking the p < g^4 T region as outside the formal validity of the calculation, or by adding the missing diagrams. I therefore recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. The paper fixes a real, known problem: matching-based computations of ALP/axion production from Abelian gauge fields gave unphysical negative rates at soft momenta. Using the full 1PI-resummed photon propagator in the ALP self-energy, the authors get positive rates at every momentum with no matching or subtraction scheme. And the paper's second headline — timelike-timelike (TT) dominance at p ≲ g^4 T — is a genuinely new finding, but it is explicitly provisional, and the authors say so in their own Sec. 3 and conclusion.\n\nCredit where due. The derivation of Eq. (3.7) is careful, the gauge-independence argument is clean, and the SS scaling in the HTL regime reproduces the independent Laine–Vuorinen–Zhu result — a good sanity check. The paper is unusually explicit about its approximations, including the ones that cut against its own headline. The phenomenological impact is modest (integrated quantities shift by a few percent; ⟨p⟩ changes by O(1–10%)), which is an honest result: the average is dominated by hard momenta, so Lyman-alpha bounds move little. The soft-momentum shape of the distribution does change, which is what makes the TT finding matter for keV ALP work.\n\nThe soft spot, in proportion. The p ≲ g^4 T TT-dominance claim is not yet a demonstrable leading-order result. The omitted last row of Fig. 2 — 2↔3 processes like fγ → fγa — are two-loop corrections to the photon self-energy and ALP vertex, i.e., corrections to the very width ΓTT that sets the TT scaling in Eq. (B.12), the crossover pc = ΓTT(k=0), and the TT/SS ratio. So the flagship claim rests on an incomplete truncation at exactly the momentum where it dominates. The missing cut diagrams are positive, so the positivity result stands; the concern is completeness, not sign. I'd also flag that Sec. 4 item (iv) is worded as if the TT dominance were established, which is stronger than the abstract's \"potentially\" and the conclusion's \"needs to be confirmed by a consistent leading order computation\" — the conclusion has the right tone, item (iv) does not. Minor: no code is shipped, and the κ_I = 8 prefactor in the SS scaling is fitted to their own numerics (fine for a scaling estimate, not a prediction).\n\nWho it's for: anyone doing thermal ALP/axion production, freeze-in, or soft resummation in hot plasmas. It deserves a serious referee. I'd send it out and ask the authors to either compute the missing 2↔3 contribution or consistently flag the p ≲ g^4 T claim as provisional throughout.","headline":"A careful, self-aware full-propagator calculation that cures negative rates and finds a new timelike-timelike channel at p ≲ g^4 T, but that flagship claim is provisional because the omitted 2↔3 diagrams correct exactly the photon width that sets it.","tokens_in":23052,"tokens_out":6691,"would_cite":true,"duration_ms":56072,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["14.80.Va","95.35.+d"],"model":"deepseek-v4-flash","headline":"Using the full 1PI-resummed gauge-boson propagator in the ALP self-energy, the paper shows thermal production rates stay positive at all momenta and that a timelike-timelike channel dominates at the softest scales, p ≲ g⁴T.","keywords":["axion-like particles","thermal production rate","1PI resummation","hard thermal loops","soft momentum","freeze-in dark matter","Lyman-alpha constraints","gauge boson spectral density"],"falsifier":"Complete the leading-order computation by adding the missing $2\\to 3$ diagrams with soft ALP emission from external photons (e.g., $f\\gamma\\to f\\gamma a$) in the same 1PI-resummed framework and recompute the rate at $p \\lesssim g^4 T$: if the timelike-timelike channel no longer dominates, or if the total rate develops a negative region, the paper's central claims fail. A cheaper test is to rerun the calculation at a smaller gauge coupling, say $g_1 \\approx 0.1$, so the windows between $g^4 T$, $g^2 T$, and $gT$ widen, and check the predicted turnover of the timelike-timelike contribution from a momentum-independent plateau to $p^2$ scaling at $p \\approx \\Gamma_{TT}$, with positivity at every momentum.","tokens_in":21993,"feed_emoji":"⚛️","tokens_out":15331,"duration_ms":128877,"temperature":0.7,"pith_summary":"This paper aims to fix a known pathology in thermal production rates for feebly interacting particles: when hard and soft momentum regions are computed separately and matched at an arbitrary scale, the ALP interaction rate turns negative at soft momenta, which is unphysical. The authors argue that the cure is to stop approximating the gauge-boson propagator by hard-thermal-loop (HTL) forms and instead use the full 1PI-resummed propagator in both legs of the one-loop ALP self-energy, which is valid at every momentum scale and needs no matching or subtraction. They show that this yields positive rates everywhere and exposes a production channel — two timelike gauge bosons whose finite width relaxes kinematic constraints — that can dominate at the softest momenta ($p \\lesssim g^4 T$) and exceed the previously known spacelike-spacelike channel by another order of magnitude. A careful reader would care because the soft part of the momentum distribution feeds the Lyman-$\\alpha$ constraints on keV-scale ALP warm dark matter, and the paper updates the abundance and average momentum accordingly, although the integrated shifts are only at the percent level.","feed_headline":"Photon width reveals new dominant channel for softest ALP production","feed_subtitle":"Full resummed photon propagators cure negative rates and reshape the keV ALP dark-matter momentum distribution.","key_machinery":"The central object is the 1PI-resummed gauge-boson Wightman propagator, decomposed into transverse, longitudinal, and momentum-parallel polarizations, with spectral densities built from the full one-loop photon self-energy including the vacuum term. Feeding both propagators in their full resummed form into the ALP self-energy integral is what carries the argument: the imaginary parts of the full self-energy do not vanish for timelike momenta, so the timelike pole acquires a width that relaxes the kinematic suppression of the timelike-timelike channel, while the gauge-dependent parallel polarization cancels in the contraction, making the whole rate gauge independent. The clean separation into timelike-timelike, timelike-spacelike, and spacelike-spacelike channels — with scalings $p^2$ below the width, exponential suppression, and $p^{4/3}$, respectively — is what organizes the momentum regimes and underlies the soft-rate estimates.","core_discovery":"Using the full form of the 1PI-resummed abelian gauge-boson propagator in both legs of the ALP self-energy, rather than HTL-approximated propagators matched to free ones at an intermediate scale $gT \\ll k_\\star \\ll T$, removes the unphysical negative interaction rates that plagued previous calculations: the collision term $\\Pi^<(p)$ is positive for all momenta without any matching or subtraction procedure. The new channel appears because the full spectral densities give the timelike photon a finite width — the HTL imaginary parts vanish for $K^2 > 0$ — which relaxes the kinematic constraints on production from two timelike gauge bosons. The resulting rate separates into clear regimes: for $p \\gtrsim g T$ timelike-spacelike ($t$-channel $2\\leftrightarrow 2$) scatterings dominate; for $g^4 T \\lesssim p \\lesssim g^2 T$ two spacelike bosons dominate and scale as $p^{4/3}$; and for $p \\lesssim g^4 T$ two timelike bosons take over, with the self-energy nearly momentum-independent until $p$ drops below the zero-momentum photon width $\\Gamma_{TT} \\approx 5\\times 10^{-4} T$, below which it scales as $p^2$. The authors stress that this hierarchy holds within their truncation and that a consistent leading-order result for the soft regime still requires the diagrams with soft ALP emission from external photons, which are left for future work.","pith_inferences":["Carried over to the gravitino and sterile-neutrino production problems, where the same negative rates appear, the full-propagator cure is likely a general feature of abelian sectors, while non-abelian cases would confront the gauge dependence the authors sidestep here.","The timelike-timelike dominance implies a flatter low-momentum tail in the freeze-in spectrum than the cut-based estimate suggests; even though the integrated average momentum shifts by only a few percent, the soft tail itself could be probed by momentum-sensitive small-scale structure observables such as 21-cm or strong-lensing statistics, which the paper does not study.","A decisive near-term extension is to include the omitted soft-emission diagrams; if they preserve the positivity and the timelike-timelike dominance, the method becomes the default tool for abelian freeze-in rates, and if they do not, only the positive-rate statement for the one-loop truncation itself survives.","The turnover scale $p_c \\approx \\Gamma_{TT}$ carries a specific coupling dependence, so scanning $g_1$ downward should move the plateau-to-$p^2$ turnover to smaller $p/T$; the paper's fixed $g_1 = 0.35$ setup leaves that prediction untested."],"forward_implications":["The full-propagator method eliminates the matching scale $k_\\star$ and the subtraction or tuned-mass schemes, so the ALP production rate is positive at every momentum without additional input parameters.","The timelike-timelike channel, completely absent in the HTL approximation because of the vanishing width, becomes the dominant production mechanism for $p \\lesssim g^4 T$, and neglecting it underestimates the soft-momentum interaction rate by more than an order of magnitude.","The updated ALP distribution function leaves the number density essentially unchanged, $n_{\\rm 1PI}/n_{\\rm HTL} = 0.97$ and $n_{\\rm 1PI}/n_{\\rm Cut} = 1.09$, and lowers the average momentum to $\\langle p/T\\rangle \\approx 3.06$, which translates into a percent-level weakening of the Lyman-$\\alpha$ lower bound on the ALP mass as warm dark matter.","Because the contraction cancels the gauge-dependent polarization, the same treatment applies without modification to any ALP or feebly interacting particle coupled to an abelian gauge field — QED photons, dark photons, or $U(1)_Y$ — making positivity and the channel hierarchy a general abelian feature.","The authors position the result as the first step toward full leading-order accuracy for $g^4 T \\lesssim p \\lesssim g T$, and note that below $p \\approx g^4 T$ the quasiparticle description breaks down and hydrodynamics become necessary."],"supporting_citations":[{"why":"Supplies the two-photon-loop ALP self-energy master formula and its cut-based reduction into longitudinal-transverse and double-transverse channels, which the paper evaluates with full propagators.","marker":"[9]"},{"why":"The previous ALP production calculation whose matching ('Cut') scheme yields negative soft-momentum rates; it is the distribution-function comparison target and the source of the Lyman-alpha motivation.","marker":"[11]"},{"why":"The recent improved subtraction and tuned-mass scheme for thermal ALP production that this work's full-propagator method is designed to supersede.","marker":"[30]"},{"why":"Introduces the matching-scale prescription that separates HTL-resummed from free propagators, the source of the negative-rate problem addressed here.","marker":"[5]"},{"why":"Earlier observation that spacelike-spacelike configurations dominate the soft rate; the paper reproduces that scaling and adds the timelike-timelike channel.","marker":"[54]"},{"why":"Previous freeze-in application of full 1PI-resummed propagators, providing the methodological template and the hard-momentum comparison point.","marker":"[38]"},{"why":"The dilepton-production study showing that 1PI resummations of this type can miss soft emission from external particles, the exact caveat the paper attaches to its own result.","marker":"[55]"},{"why":"The first HTL soft-rate computation in which the matched-propagator approach produced negative rates, framing the problem this paper solves.","marker":"[3]"}],"fun_headline_variants":["Full propagators fix ALP rates, reveal new soft channel","Timelike photon width enables dominant soft ALP production","Curing negative rates: new ALP production channel found","Resummed photons reshape keV ALP dark matter momentum","Full gauge propagators: positive ALP rates, new soft regime"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The paper assumes that the one-loop ALP self-energy with both gauge-boson propagators fully 1PI-resummed already gives the leading-order soft-momentum rate, yet it explicitly acknowledges that a class of three-loop diagrams — soft ALP emission from external photons, the last row of Fig. 2 — is missing and would be needed for a consistent leading-order computation in that regime.","fun_headline_variants_meta":{"raw":{"variants":["Full propagators fix ALP rates, reveal new soft channel","Timelike photon width enables dominant soft ALP production","Curing negative rates: new ALP production channel found","Resummed photons reshape keV ALP dark matter momentum","Full gauge propagators: positive ALP rates, new soft regime"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000224,"raw_usage":{"total_tokens":1527,"prompt_tokens":1079,"completion_tokens":448,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":695,"completion_tokens_details":{"reasoning_tokens":365}},"tokens_in":695,"tokens_out":448,"duration_ms":4928,"temperature":1.0,"reasoning_tokens":365,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T14:39:15.206867+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Complete the leading-order computation by adding the missing $2\\to 3$ diagrams with soft ALP emission from external photons (e.g., $f\\gamma\\to f\\gamma a$) in the same 1PI-resummed framework and recompute the rate at $p \\lesssim g^4 T$: if the timelike-timelike channel no longer dominates, or if the total rate develops a negative region, the paper's central claims fail. A cheaper test is to rerun the calculation at a smaller gauge coupling, say $g_1 \\approx 0.1$, so the windows between $g^4 T$, $g^2 T$, and $gT$ widen, and check the predicted turnover of the timelike-timelike contribution from a momentum-independent plateau to $p^2$ scaling at $p \\approx \\Gamma_{TT}$, with positivity at every momentum.","supporting_citations":[{"cited_title":"Braaten and M","cited_arxiv_id":null,"evidence_quote":"The first HTL soft-rate computation in which the matched-propagator approach produced negative rates, framing the problem this paper solves."}],"review_version":1}