{"id":"f6e6663c-0945-4fc7-88bb-944957a169ab","arxiv_id":"2504.13156","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"The sourced gravitational wave background from axion inflation carries energy-density anisotropies up to O(10^{-1}), far larger than the vacuum contribution, due to modulation of sourced tensor modes by CMB-scale curvature perturbations.","lead":"Axion inflation, where a pseudoscalar inflaton amplifies gauge fields, is shown to produce gravitational wave anisotropies as large as O(10^{-1}) of the background energy density at interferometer frequencies. This makes the cosmological gravitational wave background from axion inflation potentially distinguishable from astrophysical foregrounds and opens a new observational target for upcoming detectors.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The O(10^{-1}) result hinges on a slow-roll-derived response factor applied in the strong backreaction regime; if backreaction modifies dξ/dφ0, the central prediction shifts.","rationale":"The reader's weakest_assumption is exactly the linear response relation (2.31) in the strong backreaction regime. My stress-test sharpens this: the load-bearing element is the numerical response coefficient dξ/dφ0, which the paper evaluates using the slow-roll background relation (2.33). In strong backreaction, the inflaton dynamics is modified by gauge-field backreaction, so the relation between ξ and φ0 is not the slow-roll one; the explicit assumption in Section 1 that results remain valid when expressed via dotφ does not establish that (2.33) holds. Since the central normalized correlator scales as the square of this factor, an O(1) change in dξ/dφ0 moves the prediction across the claimed 10^{-5}–10^{-1} window. This is a legitimate correctness risk, but it is an identified assumption rather than an internal inconsistency; the paper is transparent about it. I also note the angular integrals in Appendices B and C are quoted without derivation, a secondary numerical-verification issue. The proposed lattice test would settle whether the backreacted response matches the slow-roll estimate; until then, CONDITIONAL acceptance is appropriate. I therefore keep the reader's verdict unchanged.","tokens_in":31609,"tokens_out":7091,"duration_ms":70765,"concrete_test":"Use a lattice simulation code in the strong-backreaction regime (e.g., following Figueroa et al. PRL 131, 151003) to measure the response of the sourced GW power spectrum at interferometer-scale modes to a small change in the initial inflaton field value, φ0 → φ0+δφ (or equivalently a long-wavelength perturbation implemented via separate-universe patches). Extract d ln P_h,S / dφ0 from the fractional change in the sourced GW spectrum and compare it with 4π(dξ/dφ0), where dξ/dφ0 is computed from (2.33). If the two disagree by more than ~30%, the response factor in (4.28) is mis-estimated and the 2.4×10^{-1} upper bound in (4.29) is not supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central result (4.29) is proportional to the square of the response factor 2π(dξ/dφ0)(dotφ/H), with dξ/dφ0 taken from the slow-roll expression (2.33): dξ/dφ0 = (ϵ−η/2)/f. This derivation assumes the background equation of motion 3H dotφ ≃ −V′ holds at the time interferometer-scale GWs are sourced. But those modes are produced in the strong backreaction regime, where gauge-field production adds a significant source term to the inflaton equation of motion and the slow-roll relation is violated. The paper's Section 1 assumption that results written in terms of dotφ(t) remain valid does not by itself justify replacing the backreacted response dξ/dφ0 with the slow-roll expression (2.33). Since (2.34) uses this to bound the response factor as O(0.1–3), an order-one error in dξ/dφ0 directly changes the claimed range (4.29). If the backreacted response is suppressed or enhanced, the headline anisotropy changes by orders of magnitude. The linearity of δh in ζV (eq. 2.31) is also an assumption in a nonperturbative regime, but the specific weak point is the numerical value of dξ/dφ0 entering the response.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper computes the normalized two-point correlator of the gravitational-wave energy density, C_ΩΩ(k), in axion inflation, and splits it into vacuum, sourced-intrinsic, sourced-extrinsic, and sourced-fluctuation contributions. The central result is eqs. (4.27)–(4.29): the scale-invariant part of the sourced extrinsic correlator is about 9.8×10^-5 / δ^2 × (2π dξ/dφ0 φ̇0/H)^2, which with δ ∈ [0.06, 0.2] and the slow-roll response factor (2.34) gives a range 2.4×10^-5 to 2.4×10^-1. The paper argues that all other contributions, including the vacuum correlator, the intrinsic sourced correlator, and the subdominant extrinsic terms, are suppressed by (k/k_BR)^3 and are therefore negligible for CMB-scale anisotropy modes. It concludes that axion inflation can produce gravitational-wave anisotropies large enough to be observable by future detectors.","tokens_in":31793,"tokens_out":12706,"duration_ms":122329,"significance":"If the central result holds, the paper provides a concrete and falsifiable target for stochastic-gravitational-wave anisotropy searches and strengthens the case that axion inflation can be distinguished from astrophysical backgrounds. A notable strength is the normalization by Ω_GW,S, which removes the exponential e^{4πξ} sensitivity and makes the prediction depend only on δ and on the response factor, so the quoted range is meaningful rather than dominated by the large exponential. The paper is also transparent about its main assumption, stated in Section 1, and the appendices contain a substantial amount of the calculation. The main risk, as detailed below, is that the slow-roll expression for dξ/dφ0 is used inside the strong-backreaction epoch, where the background equation of motion is modified; this is a serious but correctable issue rather than a fatal inconsistency.","major_comments":[{"comment":"The headline value rests on evaluating the response factor R = 2π(dξ/dφ0)(φ̇0/H) with the slow-roll derivative dξ/dφ0 = (ϵ − η/2)/f, derived from 3H φ̇0 ≃ −V′. However, the sourced interferometer modes studied in Section 4 are produced near τ_BR, where gauge-field production adds a source term to the inflaton equation and the slow-roll relation is not guaranteed. The Section 1 statement that results written in terms of φ̇(t) remain valid addresses the amplitude h^0_{ij,S}, but not the derivative dξ/dφ0; the cited reference [41] does not establish (2.33) in the backreaction regime. Since (4.28) is quadratic in R and the quoted range (4.29) uses the O(0.1–3) bound of (2.34), an order-one change in dξ/dφ0 shifts the central prediction by orders of magnitude. Please derive or parametrize the backreacted response and propagate the resulting uncertainty.","section":"2.2–4.2.1, eqs. (2.31)–(2.34) and (4.28)"},{"comment":"The linear response relation δh_{ij,S} = R ζ_V h^0_{ij,S} is assumed to hold in the strong-backreaction regime, where the dynamics is nonperturbative. This relation is obtained from a first-order Taylor expansion of h^0 ∝ e^{2πξ}; in the backreacted background the response of the sourced tensor modes to long-wavelength ζ could be nonlocal, suppressed, or nonlinear in ζ. Because this relation is the mechanism that converts long-wavelength curvature perturbations into gravitational-wave anisotropies, the paper should state its regime of validity more carefully and, if possible, test it against lattice results such as those in [38,41].","section":"2.2, eq. (2.31), used in (4.7)–(4.10)"},{"comment":"The momentum integrals are truncated at p = k_BR with the statement that contributions from momenta larger than k_BR are negligible. Under the parameterization (4.18), modes with p > k_BR are sourced at τ > τ_BR, where ξ = ξ_BR, so they carry the full e^{4πξ_BR} enhancement and no (τ_BR/τ)^{2πδ} suppression. The text does not justify why these modes can be dropped; if they contribute, the (k/k_BR)^3 suppression in (4.21), (4.30)–(4.32) and the coefficient of the scale-invariant term (4.27) would need revision. Please provide a physical estimate of the p > k_BR region or extend the integration.","section":"4.1–4.2.2 and Appendices B–C"},{"comment":"The numerical angular integrals are quoted without derivation or method; examples include 1.4×10^4 in (B.4), 1.8×10^3 in (B.5), 4.4×10^4 in (C.3), and the closed-form results (C.5)–(C.12). Because the headline range (4.29) depends directly on I_S.I. through (4.27)–(4.28), these numbers are load-bearing. Please provide the angular integration procedure or a reproducibility notebook, and state the estimated error introduced by the 2^7-ordering truncation used throughout.","section":"Appendices B and C, eqs. (B.4)–(B.7), (C.3), (C.5)–(C.12)"}],"minor_comments":[{"comment":"There is a typo in the transfer-function argument: \\(\\hat T(|k+p_2)\\) should read \\(\\hat T(|k+p_2|)\\).","section":"4.2, eq. (4.22)"},{"comment":"The statement that the correlator of the fluctuations C^F_ΩΩ is 'much smaller' is not quantified; a one-line estimate, e.g., its parametric suppression by P_ζ^2 and R^4, would make the neglect more transparent.","section":"4, after eq. (4.10)"},{"comment":"The abstract and conclusions state the result as 'O(10^-1)', but the actual range in (4.29) is 2.4×10^-5 to 2.4×10^-1; using both endpoints would give a more accurate summary of the prediction.","section":"6, Conclusions"},{"comment":"Reference [30] is incomplete: it lists only the arXiv number without a title or journal, and reference [5] similarly lacks a title.","section":"References"},{"comment":"The angular part A_D is written down without derivation; a brief derivation using the polarization-sum identity (4.26) would improve reproducibility.","section":"4.2.1, eq. (4.25)"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of JCAP and likely to interest the axion-inflation and gravitational-wave communities. The central concern is the use of the slow-roll-derived response factor dξ/dφ0 in the strong-backreaction epoch; this should be addressed before publication, either by deriving the backreacted response or by introducing and propagating an explicit parameter for it. The appendices are substantial but need more detail on the angular integrations and on the k_BR cutoff before the quantitative claims can be independently checked."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper computes the auto-correlator of gravitational wave energy densities sourced by gauge fields in axion inflation, and the main result is new: a scale-invariant extrinsic term that, normalized by the fractional energy in sourced GWs at interferometer scales, can be as large as O(10^{-1}). The vacuum and intrinsic contributions are shown to be suppressed by (k/k_BR)^3, so the extrinsic term dominates. That is a genuine step beyond the existing literature, which had the vacuum correlator and the cross-correlation with curvature perturbations but not this auto-correlator.\n\nWhat the paper does well: the decomposition of the sourced four-point function into intrinsic, extrinsic, and fluctuation pieces is careful, and the analytic evaluation is detailed. The author is explicit about the main approximations: WKB gauge mode functions, saddle-point time integrals, dropping momentum orderings suppressed by 2^7, and a phenomenological parametrization of the backreaction-era growth of ξ with δ in the range 0.06–0.2. That is honest and lets a reader see where the numbers come from.\n\nThe soft spots are in proportion to how much they matter. The strongest caveat, also flagged in your stress-test note, is that the response factor 2π(dξ/dφ0)(dotφ/H), which enters the result squared, is derived from the slow-roll equation of motion. The paper assumes this remains valid in the strong-backreaction regime, and it cites lattice results to justify that the sourced GWs depend on dotφ, but that does not directly justify replacing the backreacted dξ/dφ0 with the slow-roll expression. If backreaction changes this derivative, the central prediction shifts. The paper states the assumption explicitly, so it is not hidden, but it is untested. Second, the angular integrals in Appendices B and C are quoted after \"numerical integration\" without the integrands or code; the main constant 4.4×10^4 in eq. (C.3) determines the amplitude, and I could not verify it independently. Third, the quoted range (4.29) spans five orders of magnitude, from 2.4×10^{-5} to 2.4×10^{-1}, driven almost entirely by δ. The abstract emphasizes the upper end, which is the optimistic, small-δ corner. That is worth keeping in mind when citing the result.\n\nNone of this is fatal. The calculation is self-contained, the central argument holds together, and the limitations are the usual kind for an analytic estimate in a nonperturbative regime. This paper deserves a serious referee: I would send it to review, and I would ask for more detail on the angular integrals and a sensitivity analysis around the backreaction assumption. The readers who get value from this are people working on gravitational wave anisotropies, axion inflation, and the astrophysical-versus-cosmological background distinction. I would cite it as the first computation of the sourced auto-correlator, with a caveat on the strong-backreaction response.","headline":"New calculation of the sourced GW energy-density auto-correlator in axion inflation with a headline O(10^{-1}) anisotropy that depends sensitively on a slow-roll-derived response factor applied in the strong-backreaction regime.","tokens_in":32411,"tokens_out":2362,"would_cite":true,"duration_ms":24298,"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":"Sourced gravitational waves from axion inflation can have order-0.1 anisotropies in their energy-density correlator, a level that would make the cosmic gravitational-wave background anisotropic enough for next-generation detectors.","keywords":["stochastic gravitational wave background","axion inflation","gravitational wave anisotropies","sourced tensor modes","gauge field production","inverse decay","extrinsic correlator","CMB cross-correlation"],"falsifier":"Run a lattice simulation of axion inflation through the strong-backreaction regime and measure the modulation of short-wavelength sourced tensor modes by a long-wavelength curvature perturbation; if the response deviates from $-2\\pi\\frac{d\\xi}{d\\phi_0}\\frac{\\dot\\phi_0}{H}$ at the level assumed in eq. (2.31), the predicted $O(10^{-1})$ extrinsic correlator collapses. Alternatively, an observational upper bound on the normalized GW anisotropy correlator below $10^{-3}$ at interferometer frequencies, in a model where sourcing is confirmed by a chiral signal, would contradict the central claim.","tokens_in":31304,"feed_emoji":"📡","tokens_out":8048,"duration_ms":70226,"temperature":0.7,"pith_summary":"This paper asks whether axion inflation leaves a detectable anisotropic pattern in the stochastic gravitational wave background. It argues that the dominant contribution is not the vacuum tensor modes, whose anisotropy correlator is tiny, but the extrinsic part of the sourced modes: the modulation of short-wavelength sourced gravitational waves by long-wavelength curvature perturbations. Normalized by the expected fractional energy density in sourced gravitational waves at interferometer frequencies, this correlator can reach O($10^{-1}$), with a central estimate of 2.4 times $10^{-1}$. If true, the gravitational-wave sky from axion inflation could be anisotropic enough for future detectors and for cross-correlation with CMB anisotropies, offering a way to tell a cosmological background from an astrophysical one.","feed_headline":"Axion inflation can drive large gravitational wave anisotropies","feed_subtitle":"A new calculation finds the sourced correlator can reach 0.24, putting the background within detector reach.","key_machinery":"The load-bearing relation is the linear response formula for sourced tensor modes, $\\delta h_{ij,S} = -2\\pi\\frac{d\\xi}{d\\phi_0}\\frac{\\dot\\phi_0}{H}\\zeta_V\\, h^0_{ij,S}$ (eq. 2.31), where $\\xi=\\dot\\phi_0/(2fH)$ is the gauge-field amplification parameter. It converts long-wavelength vacuum curvature perturbations $\\zeta_V$ into a multiplicative modulation of the short-wavelength sourced gravitational-wave amplitude. Substituting this into the four-point function of sourced tensor modes and applying Wick's theorem, the term labeled $C_D$ contains a delta function $\\delta(k-p_3)$ that cancels the $k^3$ prefactor in the correlator definition, producing a scale-invariant extrinsic contribution. The amplitude is set by the parametrization (4.18), in which $\\xi$ grows logarithmically before backreaction and then becomes nearly constant.","core_discovery":"The paper's main result is that the scale-invariant part of the sourced extrinsic correlator of the gravitational-wave energy density, computed in the large-momentum regime and normalized by the square of the fractional sourced energy, is $C^{S.I.}_{\\Omega\\Omega}(k)\\big|_{l.m.} \\simeq \\frac{9.8\\times10^{-5}}{\\delta^2}\\left(2\\pi\\frac{d\\xi}{d\\phi_0}\\frac{\\dot\\phi_0}{H}\\right)^2$, which lies between $2.4\\times10^{-5}$ and $2.4\\times10^{-1}$ for the backreaction growth parameter $\\delta$ in the range $0.06$--$0.2$ (eq. 4.29). The intrinsic sourced correlator, the subdominant extrinsic terms, and the vacuum correlator are all suppressed by $(k/k_{BR})^3$ with $k\\sim k_{CMB}$, making them unobservable. The paper concludes that this scale-invariant extrinsic term is the only relevant component of the sourced correlator and that axion inflation can generate gravitational-wave anisotropies within observational reach.","pith_inferences":["Editorial inference: the same modulation mechanism should operate in any scenario where short-wavelength sourced tensor modes are linearly modulated by long-wavelength curvature perturbations, so the $O(10^{-1})$ scale is not specific to the axion potential but to the combination $2\\pi(2\\epsilon-\\eta)\\xi/\\delta$; models with larger $\\xi$ or smaller $\\delta$ would predict even larger anisotropies.","Editorial inference: because the dominant term is scale-invariant and ties interferometer-scale tensor modes to CMB-scale scalars, the model predicts a specific cross-correlation between GW background anisotropies and CMB temperature anisotropies that could be measured even if the isotropic background is only marginally detectable.","Editorial inference: a direct lattice computation of the sourced tensor four-point function in the strong-backreaction regime would test whether the linear response relation (2.31) survives nonperturbatively, and would either confirm or revise the $2.4\\times10^{-1}$ upper value."],"forward_implications":["The dominant observable in the axion-inflation gravitational-wave sky is the anisotropic part of the sourced background, not the isotropic vacuum signal.","The vacuum and intrinsic sourced correlators are suppressed by $(k/k_{BR})^3$ and are unobservable, so a detected anisotropy at the predicted level would point specifically to gauge-field sourcing.","Anisotropies of order $10^{-1}$ in the normalized correlator fall within the reach of next-generation gravitational-wave observatories, making the cosmological background distinguishable from the astrophysical one through its angular structure.","The scale-invariant extrinsic term imprints a correlation between interferometer-scale gravitational waves and CMB-scale scalar perturbations, motivating searches for cross-correlations between GW anisotropies and CMB temperature maps."],"supporting_citations":[{"why":"The companion calculation of the curvature-perturbation to gravitational-wave-energy-density correlator that this paper extends, providing the decomposition into homogeneous and fluctuating sourced modes.","marker":"[26]"},{"why":"Establishes parity-violating sourced gravitational waves from a pseudoscalar inflaton and gives the sourced tensor power spectrum used for normalization.","marker":"[10]"},{"why":"Derives the amplified gauge-field mode functions and the sourced scalar and tensor spectra that the four-point calculation uses.","marker":"[11]"},{"why":"Shows sourced gravitational waves from axion inflation can reach ground-based interferometer sensitivities, motivating the interferometer-frequency normalization.","marker":"[17]"},{"why":"Argues sourced gravitational-wave production in the strong-backreaction regime depends only on the inflaton velocity, the assumption that lets the calculation extend into that regime; also supplies an example with $N_{BR}\\simeq 40$.","marker":"[41]"},{"why":"Studies resonant backreaction in axion inflation and gives the $N_{BR}\\simeq 10$ value and growth-rate estimate used in the parametrization (4.18).","marker":"[37]"},{"why":"Lattice study of the strong-backreaction regime showing that $\\xi$ stabilizes, justifying the nearly constant $\\xi$ after $\\tau_{BR}$.","marker":"[40]"},{"why":"Quantifies the intrinsic noise limits on detecting stochastic-background anisotropies, used as the observability threshold against which the $O(10^{-1})$ result is compared.","marker":"[27]"}],"fun_headline_variants":["Axion inflation boosts gravitational wave anisotropies to 0.24","Gravitational wave anisotropies from axion inflation up to 0.24","Axion inflation: large gravitational wave anisotropies within reach","Sourced gravitational wave anisotropies from axion inflation hit 0.24"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The linear response relation for sourced gravitational waves, derived perturbatively, is assumed to remain valid for interferometer-scale modes in the strong backreaction regime, where the inflaton-gauge-field system is nonperturbative.","fun_headline_variants_meta":{"raw":{"variants":["Axion inflation boosts gravitational wave anisotropies to 0.24","Gravitational wave anisotropies from axion inflation up to 0.24","Axion inflation: large gravitational wave anisotropies within reach","Sourced gravitational wave anisotropies from axion inflation hit 0.24"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000346,"raw_usage":{"total_tokens":1927,"prompt_tokens":1008,"completion_tokens":919,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":624,"completion_tokens_details":{"reasoning_tokens":837}},"tokens_in":624,"tokens_out":919,"duration_ms":7476,"temperature":1.0,"reasoning_tokens":837,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T12:13:23.641679+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run a lattice simulation of axion inflation through the strong-backreaction regime and measure the modulation of short-wavelength sourced tensor modes by a long-wavelength curvature perturbation; if the response deviates from $-2\\pi\\frac{d\\xi}{d\\phi_0}\\frac{\\dot\\phi_0}{H}$ at the level assumed in eq. (2.31), the predicted $O(10^{-1})$ extrinsic correlator collapses. Alternatively, an observational upper bound on the normalized GW anisotropy correlator below $10^{-3}$ at interferometer frequencies, in a model where sourcing is confirmed by a chiral signal, would contradict the central claim.","supporting_citations":[{"cited_title":"Intrinsic limits on the detection of the anisotropies of the Stochastic Gravitational Wave Background","cited_arxiv_id":"2301.08074","evidence_quote":"Quantifies the intrinsic noise limits on detecting stochastic-background anisotropies, used as the observability threshold against which the $O(10^{-1})$ result is compared."}],"review_version":1}