{"id":"b15ce34e-3ca0-4d01-b908-45999d0dcdb9","arxiv_id":"2504.18514","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"In single-field inflation, imposing a zero one-point function for the curvature perturbation leaves a finite, scheme-dependent one-loop correction to the two-point function that background renormalization cannot absorb.","lead":"This paper shows that in single-field inflation, subtracting the average of the curvature perturbation from the background does not remove its one-loop corrections to the power spectrum. A finite correction survives, but its value depends on the mathematical regularization scheme, so it is not a robust prediction.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The finite terms in Eqs. (30) and (36) are extracted by inserting the WKB form (19) into the dimensionally regularized k-integral; for a featured potential, non-adiabatic corrections to the mode functions can contribute to the finite remainder, so the claim that these terms are the unique…","rationale":"The paper presents a coherent background-renormalization argument, and the mechanism that one-point renormalization induces quadratic counterterms is plausible and internally consistent. The reader's verdict is CONDITIONAL, and my stress-test does not move that verdict. I agree with the reader that the WKB assumption is the weakest point, but I sharpen it: WKB is used not only to isolate UV divergences but also to define the finite remainder through Eq. (33). For the sharp-feature models that motivate the paper, non-adiabatic corrections to the mode functions around the feature scale can contribute finite terms that WKB misses. A purely algebraic re-derivation cannot settle this; a numerical mode-by-mode evaluation for a concrete featured potential is the decisive check. If that check shows agreement, the central claim is substantially supported. If it shows disagreement, the explicit finite terms and the scheme-dependence conclusion would need revision. Because the paper itself promises the comoving-gauge and general-scale calculation in the companion work, the appropriate status remains conditional rather than accepted or rejected.","tokens_in":12508,"tokens_out":21310,"duration_ms":243759,"concrete_test":"Choose a featured potential realizing a sharp USR-to-SR transition, as in Refs. [40,41]. Numerically solve the Mukhanov-Sasaki equation (18) with Bunch-Davies initial conditions for a sufficiently large set of k modes. Compute the dimensionally regularized k-integral in Eq. (33) in two ways: (i) using the exact Δ²_v(k,τ), and (ii) using the WKB expression (19). Then compare the resulting renormalized two-point functions with Eqs. (30) and (36). If the exact and WKB results differ by more than the numerical tolerance near the feature scale, the finite terms quoted in the paper are not robust; if they agree, the WKB extraction is validated for this relevant regime.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that background renormalization with ⟨ζ⟩=0 fixes quadratic counterterms and leaves a finite, scheme-dependent one-loop two-point function. The key step is Eq. (33), where the renormalized remainder is obtained from ∫ dk d/dk[(k/μa)^δ Δ²_v/(2εa²)]. The letter states that the subtraction (32) is exact and does not rely on the WKB approximation (19), but the subsequent evaluation of the finite part does insert (19) into this total-derivative integral. For a smooth slow-roll background the WKB expansion is reliable at high k, but for the featured potentials that motivate the paper, Δ²_v(k,τ) contains non-adiabatic corrections near the feature scale k_f ∼ aH. These corrections are not pure power laws and are not captured by the WKB asymptotic form. In dimensional regularization, finite non-WKB pieces of the form ∫ dk g(k), with g(k) integrable, are not projected out by the rule ∫ dk k^α=0 for α≠−3; they can shift the finite remainder. Therefore the explicit finite terms (30) and (36), and the conclusion that the finite term 'must be taken as it is', are not uniquely determined without either a numerical evaluation of the exact mode sum or the promised comoving-gauge companion calculation. This does not invalidate the overall renormalization mechanism, but it makes the headline numerical claim conditional on the validity of WKB for the entire k-integration.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper addresses the question of whether the renormalization condition ⟨ζ⟩=0 in single-field inflation fixes the counterterms needed to render the one-loop two-point function finite. The authors introduce two linear counterterms (wave function and potential) and show that the induced quadratic counterterms cancel the divergence of the cubic-exchange diagram, leaving a finite, regularization-scheme-dependent correction. They identify the cancellation with Maldacena's consistency condition and argue that the residual finite term cannot be absorbed by the quadratic counterterms because those are fixed by the one-point renormalization condition. The calculation is performed in flat slicing with N=1, N^i=0, and the finite terms are computed in both cutoff and dimensional regularization.","tokens_in":12764,"tokens_out":22227,"duration_ms":227517,"significance":"If correct, the paper establishes a conceptual point missing from the recent PBH/loop-correction debate: background renormalization with a one-point condition does not automatically make the two-point function scheme-independent, and additional quadratic counterterms (e.g., speed-of-sound or higher-derivative operators) are needed. The use of Maldacena's consistency relation as an exact subtraction identity is elegant, and the explicit scheme dependence in Eqs. (30) and (36) is a concrete testable claim. The strength of the paper is its clear framework and the cross-check via the consistency condition; the weakness is that the quantitative finite terms rely on the WKB asymptotic form for all k and on a truncated flat-gauge action. The result is important for the debate but should not be taken as the final word until the companion comoving-gauge calculation and a treatment of non-adiabatic modes are available.","major_comments":[{"comment":"The finite parts of the renormalized two-point function are evaluated by inserting the WKB asymptotic form (19) for Δ²_v into the k-integrals. Equation (33) is the clearest point: the subtraction (32) is exact, but the evaluation of the right-hand side uses (19) to replace Δ²_v by its UV expansion. For a featured potential, the mode function has non-adiabatic corrections near k_f ∼ aH, and these contribute to the finite part of the dimensionally regularized integral as ∫ dk g(k), which is not removed by the rule ∫ dk k^α=0. Hence the explicit finite terms (30) and (36), and the conclusion that the finite term 'must be taken as it is', are not uniquely determined by the calculation presented. The authors should either evaluate the exact mode sum (numerically or analytically for a specific potential) or explicitly restrict the claim to the UV contribution and defer the full finite part to the companion paper [82].","section":"Eqs. (19), (28), (30), (33), (36)"},{"comment":"The action used for the loop computation is truncated to flat slicing with N=1 and N^i=0, so metric perturbations in flat gauge are not included in the cubic and quartic vertices. For featured potentials with large η, these neglected terms can be of the same order as the kept terms, so the one-loop two-point function computed here may not be the full single-field result. Since the conclusion that the finite term is large and time-dependent is quantitative, the authors should either justify the truncation in the relevant regime or make the domain of validity explicit. The promised comoving-gauge calculation in [82] is welcome but does not by itself establish the present result.","section":"Eq. (3) and following text"},{"comment":"One renormalization condition, ⟨ζ⟩_ren = 0, cannot by itself fix the two functions δV(t) and δZ(t). The choice δV1 = -V3⟨δφ²⟩/2 below Eq. (10) is an additional assumption (or a second renormalization condition) that is not derived from ⟨ζ⟩=0. The authors should show that either (i) the residual freedom in the split does not affect the renormalized two-point function, or (ii) the background equation of motion (11) provides the second condition, and derive the resulting relation. Without this, the claim that the quadratic counterterms are 'completely fixed' by the one-point renormalization condition is not established.","section":"Eqs. (6)–(10) and the statement 'we have fixed the two counterterms'"}],"minor_comments":[{"comment":"The notation '⟨δφ(x, t)⟩bare = ⁄= 0' is garbled; it should read '⟨δφ(x, t)⟩_bare ≠ 0'.","section":"Eq. (5)"},{"comment":"The relation for \tilde{δZ} should be \tilde{δZ} = -⟨f(2)(ζ)⟩ given the preceding definitions, whereas the text states \tilde{δZ} = ⟨f(2)(ζ)⟩; please check the sign or state the convention explicitly.","section":"Text after Eq. (16)"},{"comment":"As written, the left-hand side is integrated over d^{3+δ}k but the right-hand side is not, making the equality dimensionally inconsistent; it should be an equality of integrands, or the right-hand side should carry the same integration.","section":"Eq. (32)"},{"comment":"The phrase 'non-linear symmetry of the curvature perturbation' is vague; 'non-linear gauge transformation' would be clearer.","section":"Abstract"},{"comment":"The step ∫ dk/k δ k^δ = -1 relies on a specific dimensional-regularization convention for power-law integrals; please state the convention explicitly to avoid sign confusion.","section":"Eq. (33)"},{"comment":"The discussion of the disagreement with the cutoff regularization in [74,75] is hard to follow; a short equation or a clear statement of the time argument of the UV cutoff would help.","section":"Footnote 2"}],"recommendation":"major_revision","confidential_remarks":"The paper is embedded in a contentious debate about PBH constraints and loop corrections. The sign typo in \tilde{δZ} and the missing integration in (32) are easy fixes. The deeper issues—the WKB evaluation of the finite term and the uniqueness of the counterterm split—require either a numerical mode-sum calculation or a clear restriction of the claims. The promised companion paper [82] is essential; if it does not confirm the comoving-gauge result, the conclusions may be gauge-dependent. I would recommend asking for the companion or a careful statement of scope before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know this paper makes one clean point in the PBH loop debate: if you impose the renormalization condition ⟨ζ⟩=0, the induced quadratic counterterms are fixed, and they cannot absorb the finite part of the one-loop two-point function. That finite part is scheme-dependent. This is a new result, not just a rehash of Pimentel–Senatore–Zaldarriaga. It directly addresses the question of whether background renormalization alone can make loop corrections scheme-independent, and the answer is no.\n\nWhat the paper does well: the derivation is systematic, the logic is clear, and it gives explicit results in both dimensional regularization and cutoff regularization. The use of Maldacena's consistency condition as a cross-check is legitimate, not circular. They are also honest about the p≪k limit and the flat-slicing gauge, and they defer the general case to a companion paper. The cancellation of the divergence is convincing, and the conclusion that the leftover finite term is not absorbable by the induced counterterms follows from the structure of the calculation.\n\nThe soft spots are real but not fatal. The algebra in Eqs. (28) and (35) is compressed; you have to trust the intermediate steps. More importantly, the explicit finite terms in Eqs. (30) and (36) are obtained by inserting the WKB asymptotic form (19) into the integral. The paper notes that the subtraction (32) is exact, but the evaluation of the finite remainder does rely on WKB. For featured potentials, non-adiabatic corrections near the feature scale could contribute to the finite part, so the numerical coefficients in those equations are not fully robust. The stress-test note is half-right: the scheme-dependence claim is still demonstrated, but the size and time-dependence of the finite term are conditional on WKB. The companion paper is essential to close this gap.\n\nThe conclusion that the finite term “must be taken as it is” is correct only within the restricted setup of background renormalization. The authors themselves acknowledge that extra quadratic counterterms (e.g., speed of sound) could further renormalize the two-point function, so the “must” is not a no-go theorem.\n\nOverall: this is a serious paper that deserves a referee. The central idea is sound, the limitations are stated, and the result matters for the active debate on PBH loop corrections. I would send it to peer review; it should not be desk-rejected. I would also cite it if I worked on inflationary loop corrections. For a reading group, it would generate good discussion about renormalization schemes and the role of consistency conditions, though it is specialized.","headline":"A focused, plausible demonstration that fixing the one-point function leaves a scheme-dependent finite one-loop two-point function in single-field inflation; the quantitative terms are conditional on the WKB mode function, but the central point survives.","tokens_in":13337,"tokens_out":2699,"would_cite":true,"duration_ms":28814,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Background renormalization in inflation leaves a finite, scheme-dependent one-loop correction to the curvature power spectrum that cannot be absorbed.","keywords":["single-field inflation","curvature perturbation","background renormalization","one-loop power spectrum","regularization scheme dependence","ultra-slow-roll inflation","primordial black holes","consistency relation"],"falsifier":"Evaluate the one-loop renormalized two-point function directly in the comoving gauge for a specific featured potential, without the $p\\ll k$ approximation. If the cutoff and dimensional-regularization results can be made identical after absorbing the finite remainder into the quadratic counterterms that the one-point renormalization condition induces, the paper's central claim fails.","tokens_in":12215,"feed_emoji":"🌌","tokens_out":9391,"duration_ms":86473,"temperature":0.7,"pith_summary":"To make sense of non-linear inflationary predictions, the background must be renormalized so that the curvature perturbation has zero expectation value. The paper shows that this background renormalization fixes two linear counterterms, which then induce quadratic counterterms through the non-linear gauge transformation between field variables. Those induced counterterms subtract only the divergent part of the one-loop two-point function; the remaining finite correction depends on the regularization scheme and cannot be absorbed, because the quadratic counterterms are already fully fixed by the one-point condition. This matters because one-loop power spectra are used to constrain phenomena such as primordial black hole formation, so a scheme-dependent residual changes what can be predicted without imposing further renormalization conditions.","feed_headline":"Inflation's one-loop spectrum depends on the regularization scheme","feed_subtitle":"Setting the curvature perturbation's average to zero fixes the counterterms, so the leftover term has to stay.","key_machinery":"The load-bearing object is the ultraviolet limit of the canonical perturbation variable, $\\lim_{k\\to\\infty}\\Delta_v^2(k,\\tau)=k^2/(4\\pi^2)+\\frac{1}{8\\pi^2}z''/z$ (equation 19), which fixes which parts of $\\langle\\delta\\varphi^2\\rangle$ are divergent in each regularization scheme. The argument then runs through the background renormalization $\\phi=(1+\\delta Z/2)\\phi_R$, $V=V_R+\\delta V$; the linear counterterms fixed by $\\langle\\zeta\\rangle=0$; the quadratic counterterms they induce; and the consistency relation (31) that identifies the cubic-exchange integral with the logarithmic derivative of the power spectrum, guaranteeing the divergence cancellation. The different treatment of time and wavenumber integrals in cutoff versus dimensional regularization is what produces the scheme-dependent finite terms.","core_discovery":"The authors consider single-field inflation in flat slicing, with the action expanded to quartic order, and impose the renormalization condition $\\langle\\zeta\\rangle=0$ on the comoving curvature perturbation. They introduce the background renormalization $\\phi=(1+\\delta Z/2)\\phi_R$, $V=V_R+\\delta V$, and show that the tadpole $\\langle\\delta\\varphi\\rangle$ is cancelled by the linear counterterms $\\delta V_1$ and $\\delta Z$ once $\\langle\\zeta\\rangle=0$ is enforced. These linear counterterms generate quadratic counterterms through the second-order gauge transformation $f^{(2)}(\\zeta)$, and the paper computes the one-loop two-point function in both cutoff and dimensional regularization. The divergent pieces cancel, as a manifestation of the single-field consistency relation, but the finite parts obtained in the two schemes differ, and because the quadratic counterterms are already determined the difference must be kept. The paper therefore concludes that background renormalization alone is insufficient and that additional quadratic counterterms, of the sound-speed or higher-derivative type, are needed to make the two-point function fully renormalized.","pith_inferences":["A practical resolution would be to promote the scheme dependence into a measured parameter, for instance by fixing the two-point renormalization condition to the CMB normalization at a pivot scale.","The same mechanism should affect the bispectrum and trispectrum, whose loop corrections will inherit scheme-dependent finite parts that require their own counterterms.","If the residual term is as large as the paper allows, the perturbative expansion may break down in strongly featured potentials, making a non-perturbative treatment necessary.","The superhorizon growth of the residual hints that loop corrections could mimic or contaminate signals attributed to primordial black holes or stochastic gravitational-wave backgrounds."],"forward_implications":["One-loop power spectra in single-field inflation are not fully determined until a renormalization condition is imposed on the two-point function itself, not just on the background.","The residual loop correction can be large and grow with time, so curvature perturbations can evolve significantly outside the horizon at loop level.","Primordial-black-hole abundance estimates that rely on one-loop corrections inherit a regularization-scheme dependence unless extra counterterms are introduced.","The divergence cancellation provides a direct check of the single-field consistency relation, since any mismatch would leave an uncancelled ultraviolet divergence.","The same logic extends to higher-point functions, whose loop corrections will generically not be fully subtracted by background renormalization."],"supporting_citations":[{"why":"It supplies the non-linear gauge transformation $f^{(2)}(\\zeta)$ and the consistency relation used to cancel the divergence.","marker":"[79]"},{"why":"It first argued that linear counterterms enforcing the one-point condition induce quadratic counterterms within the effective field theory of inflation.","marker":"[78]"},{"why":"It provides the ultraviolet WKB limit of the perturbation variable used to separate divergent and finite parts in both schemes.","marker":"[80]"},{"why":"It supplies the cutoff-regularization procedure for the time and wavenumber integrals in second-order perturbation theory.","marker":"[81]"},{"why":"It details the dimensional-regularization subtraction that extracts finite parts from log-divergent integrals in these correlators.","marker":"[71]"},{"why":"It represents the alternative $\\langle\\delta\\varphi\\rangle=0$ renormalization condition that the paper argues is insufficient.","marker":"[74]"}],"fun_headline_variants":["Background renormalization can't fix inflation's one-loop","Inflation's leftover loop term undoes pure background renormalization","Scheme-dependent residue survives inflation background renormalization","Extra counterterms needed after inflation backreaction renormalization","Maldacena saves divergences but finite term stays scheme-bound"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The calculation assumes that the ultraviolet behaviour of the perturbation variable follows the WKB expansion in equation (19) in both regularization schemes, and that the flat-slicing action with $N=1$, $N^i=0$ captures all relevant non-linear interactions; if either assumption fails, the explicit finite terms in equations (30) and (36) would not follow.","fun_headline_variants_meta":{"raw":{"variants":["Background renormalization can't fix inflation's one-loop","Inflation's leftover loop term undoes pure background renormalization","Scheme-dependent residue survives inflation background renormalization","Extra counterterms needed after inflation backreaction renormalization","Maldacena saves divergences but finite term stays scheme-bound"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000455,"raw_usage":{"total_tokens":2256,"prompt_tokens":889,"completion_tokens":1367,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":505,"completion_tokens_details":{"reasoning_tokens":1283}},"tokens_in":505,"tokens_out":1367,"duration_ms":10597,"temperature":1.0,"reasoning_tokens":1283,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T10:14:08.849722+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Evaluate the one-loop renormalized two-point function directly in the comoving gauge for a specific featured potential, without the $p\\ll k$ approximation. If the cutoff and dimensional-regularization results can be made identical after absorbing the finite remainder into the quadratic counterterms that the one-point renormalization condition induces, the paper's central claim fails.","supporting_citations":[],"review_version":1}