{"id":"91c17aa2-6798-4bfe-9c1f-a820c17c2c08","arxiv_id":"2608.08602","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Small-instanton contributions can dominate the QCD axion potential in KSVZ models with many vector-like quarks, raising m_a to tens of MeV and shifting the m_a-g_aγγ relation.","lead":"This paper estimates how tiny instantons, quantum fluctuations of the strong force, can change the mass of the axion in a broad set of KSVZ models with extra heavy quarks. It finds that in some spectra the axion mass can grow to tens of MeV and that the usual relation between axion mass and photon coupling breaks, opening new search targets.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Aligned-potential premise is the weakest point: a relative phase between QCD and small-instanton potentials would change the quoted m_a values and generically induce a CP-violating minimum.","rationale":"Reading the paper in good faith, the instanton-NDA machinery is applied consistently: the endpoint exponents quoted in Sec. V match the group-theoretic zero-mode counting, the perturbativity bounds in Tab. I are internally coherent, and the UV enhancement powers follow from the stated closures. The reader's weakest_assumption identifies the same load-bearing premise: the aligned-potential assumption in Eq. (4.1). Since the paper explicitly labels this as an assumption and lists relative phases as beyond scope, this is not an internal inconsistency, but it is the least secure step linking chi_SI to the quoted axion masses and to the claim that the strong-CP solution is preserved. The proposed phase-scan test would settle whether the numbers are robust or whether they require a dedicated alignment mechanism. Because the qualitative central claim that small-instanton contributions can dominate in some KSVZ spectra is well supported by the power counting even under O(1) NDA uncertainties, the conditional verdict stands unchanged.","tokens_in":30468,"tokens_out":27769,"duration_ms":315450,"concrete_test":"For the benchmark spectra R13^(2) and S4 at Lambda = 5e11 GeV and y_Q = 1, compute the one-instanton phase delta from the product of complex VQ mass/Yukawa insertions after setting theta_QCD = 0 in the minimal Lagrangian, or scan delta uniformly over [0, 2 pi). Then minimize V(theta) = -chi_QCD cos theta - chi_SI cos(theta - delta), with chi_SI taken from Sec. III, and compute the physical mass m_a^2 = V''(theta_min)/f_a^2 and the induced theta_min. If m_a deviates by more than a factor of two from the aligned values for any Lagrangian-compatible delta, or if theta_min exceeds 1e-10 for O(1) delta, the headline numbers and the no-CP-violation statement are contingent on the alignment assumption rather than generic outcomes of the atlas.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central numerical claims depend on Eq. (4.1), where the QCD and small-instanton susceptibilities are added linearly as V(a) = -(chi_QCD + chi_SI) cos(theta-bar + a/f_a). Equal periodicity follows from both terms arising from SU(3) instantons, but alignment of the minima is not derived. In the one-instanton amplitude of Eq. (2.18), the VQ mass and Yukawa insertions carry complex phases; summing instanton and anti-instanton amplitudes then yields -chi_QCD cos theta - chi_SI cos(theta - delta) for a relative phase delta, not Eq. (4.1). The paper states the alignment assumption in Secs. I and IV and lists 'relative phases between the QCD and ultraviolet potentials' as beyond scope in Sec. V, so it is an acknowledged but unquantified premise. This matters most for the headline benchmarks R13^(2) and S4, where chi_SI/chi_QCD is enormous: for generic delta of order one, the minimum is displaced by an O(1) amount, producing a strong-CP violation, and Eq. (4.5) no longer gives the physical mass shift. The qualitative statement that small-instanton effects can dominate the curvature is robust, but the specific values 71.2 MeV and 65.3 MeV, and the claim that the standard relation is simply modified without a CP problem, are conditional on delta = 0.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper computes small-instanton contributions to the axion potential in KSVZ models with vector-like quarks, using the instanton-NDA power counting of Ref. [48]. It classifies the allowed zero-mode closures for a single VQ, for multiple identical copies, and for sets of distinct VQ representations, and it imposes one- and two-loop gauge perturbativity up to the Planck scale. The endpoint exponent Delta = b3 + r - 4 determines whether each contribution is UV or IR dominated, and the authors identify the strongest UV enhancements as (M_Pl/Lambda)^{31/3} for a single VQ and (M_Pl/Lambda)^{13} for identical copies and distinct VQ sets. Assuming equal periodicity and aligned minima of the QCD and small-instanton potentials, the paper then derives the enhanced axion mass and the resulting axion-photon coupling, reporting benchmark values such as m_a = 71.2 MeV for two copies of R13 and m_a = 65.3 MeV for the S4 set, and displaying contours in the (m_a, g_aγγ) plane against current experimental bounds.","tokens_in":30700,"tokens_out":7671,"duration_ms":83573,"significance":"The paper provides a systematic and internally coherent atlas of small-instanton effects in KSVZ axion models. The zero-mode counting, the endpoint classification, and the perturbativity filter are presented explicitly, and the group-theoretic determinants are collected in an appendix, which makes the calculation reproducible from the text. The identification of the strongest UV scalings, (M_Pl/Lambda)^{31/3} and (M_Pl/Lambda)^{13}, is a useful result that is likely robust in ordering even if the numerical prefactors are NDA estimates. If the aligned-potential premise holds, the modified m_a-g_aγγ relation opens new search regions beyond the conventional QCD-axion band, which is of genuine phenomenological interest. The main weaknesses are that the headline masses and couplings depend on an acknowledged but unquantified phase-alignment assumption, and that the numerical benchmarks are order-one NDA estimates presented without an error budget. The paper is therefore best read as a model-space survey with indicative predictions rather than as a precision calculation.","major_comments":[{"comment":"The central numerical claims rest on the assumption that the QCD and small-instanton potentials have equal periodicity and aligned minima, so that V(a) = -(χ_QCD + χ_SI) cos(θ̄ + a/f_a). The text acknowledges in Secs. I and V that relative phases require a model-dependent treatment, but this is exactly the premise on which Eq. (4.5) and the benchmark masses 71.2 MeV and 65.3 MeV depend. In the one-instanton amplitude of Eq. (2.18), the mass and Yukawa factors can carry complex phases; summing instanton and anti-instanton amplitudes generically gives -χ_QCD cos θ - χ_SI cos(θ - δ) for a relative phase δ, not Eq. (4.1). For the spectra with χ_SI/χ_QCD ≫ 1, an O(1) δ displaces the minimum by an O(1) amount, induces strong CP violation, and changes the mass-shift formula. The qualitative statement that small instantons can dominate the curvature is robust, but the quantitative atlas and the claim that the standard relation is simply modified without a CP problem are conditional on δ = 0. The authors should either derive or quantify this alignment, for example by scanning the phase and showing which conclusions survive, or clearly mark all m_a benchmarks and g_aγγ contours as applying only in the strictly aligned limit.","section":"Sec. IV A, Eq. (4.1); Secs. I, V"},{"comment":"The instanton-size integral is evaluated with a single threshold Λ = m_σ over the whole range 1/M_UV ≤ ρ ≤ 1/Λ, with b3 fixed. In the mass plots, M_Q is scanned over many orders of magnitude, including values well above Λ. Once M_Q exceeds Λ, the vector-like quark is no longer active above its own mass, so b3 and the RG-invariant scale Λ_G should be matched at M_Q as well; using a single threshold changes both the endpoint exponent Δ and the prefactor. As written, the contours in Figs. 3 and 4 for M_Q ≫ Λ are not derived from the stated matching procedure. The authors should either restrict the scan to M_Q ≤ Λ or extend Eq. (2.20) to piecewise intervals with matching at both M_Q and m_σ.","section":"Sec. II, Eq. (2.20); Sec. IV A"},{"comment":"There is an internal inconsistency in the relation between M_Q, y_Q, and Λ. From Eq. (3.4), m_σ = sqrt(2 λ_Φ) v and M_Q = y_Q v / sqrt(2). With λ_Φ = 1 and the identification Λ = m_σ, one has v = Λ and hence y_Q = sqrt(2) M_Q/Λ, whereas Sec. IV A states that 'Eq. (3.4) gives y_Q = 2 M_Q/Λ' and Fig. 5 uses M_Q = y_Q Λ/2. The factor sqrt(2) propagates into the quoted benchmark masses and couplings. The paper should state the exact convention used and recompute the affected numbers, or explain why the relation differs from Eq. (3.4).","section":"Sec. III, Eq. (3.4); Sec. IV A"},{"comment":"The numerical predictions are NDA estimates with no assessment of the order-one uncertainties in the determinant prefactors C3, the loop factors (2π/α_s)^6, and the identification of Λ with m_σ. The paper quotes m_a = 71.2 MeV and m_a = 65.3 MeV as if they were sharp values. Because χ_SI appears with powers such as (M_UV/Λ)^13, an O(1) change in Λ changes the enhancement by a large factor, and the quoted benchmark masses are correspondingly fragile. The abstract and Sec. V should present these benchmarks as order-of-magnitude indicators, or include an explicit error budget, before they are used to define search targets.","section":"Sec. IV A, Figs. 3-6"}],"minor_comments":[{"comment":"The sentence 'while the axion-photon of g_aγγ remains controlled by f_a and the anomaly ratio of E/N' is grammatically incomplete and should read 'while the axion-photon coupling g_aγγ remains controlled by f_a and the anomaly ratio E/N'.","section":"Abstract"},{"comment":"The phrase 'the scalar field is an active propagating propagating degree of freedom' contains a duplicated word; it should be 'an active propagating degree of freedom'.","section":"Sec. II, after Eq. (2.16)"},{"comment":"The authors explicitly state that the maximum multiplicities and distinct-VQ sets are fixed at Λ = 5×10^11 GeV and then reused for other values of Λ in the mass and coupling plots. This is an acknowledged limitation, but its quantitative impact on the displayed contours is not estimated; a sentence quantifying the expected shift of the perturbativity bounds would help.","section":"Sec. IV A, first paragraph"},{"comment":"The figures use colored symbol legends that may be hard to distinguish in monochrome print; a labeled legend or distinct line styles for the individual R_i spectra would improve readability.","section":"Figs. 3 and 4"}],"recommendation":"major_revision","confidential_remarks":"The paper is a useful systematic survey, and the internal NDA machinery is presented carefully. However, the headline mass values and the new regions in the (m_a, g_aγγ) plane are conditional on the unquantified alignment of the QCD and ultraviolet potentials, and the threshold treatment has technical gaps that affect the numerical contours. I would be willing to accept after a revision that either quantifies the phase alignment or explicitly restructures the claims as strictly aligned-limit results, and that fixes the threshold and parameter-relation issues."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper is a systematic extension of the instanton-NDA program to an atlas of KSVZ spectra: single VQ, identical copies, distinct sets, with two-loop perturbativity cuts. What's genuinely new is the catalog itself—the endpoint exponents, the (M_Pl/Λ)^13 maximal scalings for identical copies and distinct sets, and the concrete (m_a, g_aγγ) trajectories. The internal arithmetic checks out: I re-derived the endpoint exponents for R13^(2) and R1^(25) and they match. The authors are honest about the framework coming from Csáki et al. and about what they did not compute.\n\nThe soft spot is the aligned-minima assumption. Eq. (4.1) adds QCD and small-instanton susceptibilities as if the two cosine potentials share a minimum. The stress-test is right that this is not derived: phases in the m_Q and y_Q insertions of Eq. (2.18) can rotate the small-instanton term relative to the QCD term, giving V = -χ_QCD cos θ - χ_SI cos(θ - δ). For the headline benchmarks (two copies of R13, set S4), where χ_SI/χ_QCD is huge, a generic δ displaces the minimum by O(1), so the quoted masses—71.2 MeV and 65.3 MeV—and the strong-CP-free conclusion are conditional on δ=0. The authors flag this in Secs. I, IV, and V, so it is acknowledged rather than hidden. But the abstract and summary present the numbers without the caveat, which will mislead casual readers. A referee should ask for a phase scan or at least an explicit statement that δ is a free parameter that can ruin the CP-conserving benchmark.\n\nThe other limitations are milder and standard for NDA: no error bars on order-one coefficients, a single-threshold approximation (Λ=m_σ) with M_Q varied independently, and fixed copy counts at Λ=5×10^11 GeV. None of these undercuts the qualitative message—small instantons can dominate the axion curvature in plausible KSVZ spectra, modifying the standard mass–coupling relation. The paper is a useful map of where that happens, with concrete targets for searches outside the QCD band.\n\nWho this is for: axion model-builders and experimental colleagues who want to know where heavy KSVZ spectra might land. It deserves a serious referee. I would want the phase treatment strengthened before publication, but the core atlas and the qualitative conclusion should survive.\n\nRecommendation: send to peer review; ask for a quantitative discussion of misaligned phases and a more careful statement of the benchmark conditions.","headline":"A systematic KSVZ instanton atlas with genuinely new UV-sensitivity benchmarks, whose headline mass numbers rest on the acknowledged but unquantified aligned-minima assumption.","tokens_in":31304,"tokens_out":1955,"would_cite":true,"duration_ms":21181,"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":"Small-instanton effects from vector-like quarks can enhance the KSVZ axion mass to tens of MeV at fixed decay constant, breaking the standard QCD axion relation while leaving the axion-photon coupling unchanged.","keywords":["small instantons","KSVZ axion models","vector-like quarks","axion mass","axion-photon coupling","strong CP problem","instanton zero modes","gauge perturbativity"],"falsifier":"A direct computation of the one-instanton determinant and the fermion zero-mode overlap integrals for the fully Yukawa-closed vertex of the two-copy $R_{13}$ spectrum at $\\Lambda=5\\times10^{11}$ GeV with $y_Q=1$ would settle the claim: if the exact result comes out orders of magnitude below the NDA estimate, the predicted $m_a\\simeq71.2$ MeV collapses.","tokens_in":30205,"feed_emoji":"⚛️","tokens_out":17111,"duration_ms":149744,"temperature":0.7,"pith_summary":"This paper tries to establish that small-size instantons, the same topological fluctuations that generate the QCD axion potential, can be dramatically enhanced in KSVZ axion models by heavy vector-like quarks, and can then dominate the axion mass. The enhancement is controlled by how the instanton's fermion zero modes are saturated: each scalar-Yukawa loop used in place of two mass insertions shifts the instanton-size integral toward the ultraviolet, turning a suppressed correction into a power of $M_{\\rm Pl}/\\Lambda$ as large as $(M_{\\rm Pl}/\\Lambda)^{13}$ in the strongest multi-quark spectra. Scanning fourteen gauge-perturbative representations, the authors find that, if the QCD and small-instanton potentials are aligned, the physical axion mass can rise to 71.2 MeV at fixed decay constant (two copies of the octet-doublet representation $R_{13}$), while the axion-photon coupling is unchanged. If true, the standard QCD relation among mass, decay constant, and photon coupling no longer defines the full axion search band, and unswept regions of the $(m_a,g_{a\\gamma\\gamma})$ plane become legitimate targets.","feed_headline":"Small instantons can lift KSVZ axion masses to 71 MeV","feed_subtitle":"At fixed decay constant the photon coupling is unchanged, so axion searches should look beyond the QCD band.","key_machinery":"The machinery is the one-instanton vertex and the naive-dimensional-analysis (NDA) power counting for saturating its fermion zero modes. For every VQ set, the number of zero-mode pairs is $2|N|=2T(R_3)\\dim(R_2)$, and a closure with $k$ scalar-Yukawa loops and $r$ mass insertions must satisfy $2|N|=2k+r$; the exponent $\\Delta=b_3+r-4$ then decides whether the size integral is infrared-dominated ($\\Delta>0$), logarithmic, or ultraviolet-dominated ($\\Delta<0$). The fully Yukawa-closed term carries the largest power of $M_{\\rm UV}/\\Lambda$ but is weighted by $(y_Q^2/16\\pi^2)^k$, and the one-instanton determinant prefactors $C_3(R_3,R_2)$ complete the susceptibility estimate. The atlas of fourteen perturbative VQ representations, with copy-number limits fixed by two-loop gauge running to the Planck scale, supplies these inputs for every spectrum.","core_discovery":"The authors claim that small-instanton contributions to the axion potential in KSVZ models are governed by a counting rule: a vector-like quark (VQ) representation with $2|N|$ conjugate zero-mode pairs is closed by $k$ scalar-Yukawa loops and $r$ mass insertions with $2|N|=2k+r$, and the instanton-size integral is ultraviolet-dominated when $\\Delta=b_3+r-4<0$. Because each Yukawa loop lowers $\\Delta$ by two, maximally Yukawa-closed contractions give the largest UV enhancement: $(M_{\\rm Pl}/\\Lambda)^{31/3}$ for a single color-15 VQ, $(M_{\\rm Pl}/\\Lambda)^{13}$ for two copies of the octet doublet $R_{13}$, and the same power 13 for the distinct sets $S_3$-$S_7$. Under the stated assumption that the QCD and small-instanton potentials have equal periodicity and aligned minima, the total potential is $V(a)=-(\\chi_{\\rm QCD}+\\chi_{\\rm SI})\\cos(\\bar{\\theta}+a/f_a)$, so the axion mass at fixed $f_a$ is enhanced: the largest values quoted are 3.43 eV for a single VQ, 71.2 MeV for two copies of $R_{13}$, and 65.3 MeV for the $N_{\\rm DW}=22$ assignment of the distinct set $S_4$. The axion-photon coupling, by contrast, stays determined by $f_a$ and $E/N$, and the paper exhibits contours in the $(m_a,g_{a\\gamma\\gamma})$ plane that leave the conventional QCD-axion band.","pith_inferences":["A natural next step, not taken in the paper, is to quantify the aligned-minima assumption: if the two potentials carry a relative phase, the same machinery predicts a displaced minimum and an induced electric dipole moment roughly proportional to $\\chi_{\\rm SI}/\\chi_{\\rm QCD}$, which neutron-EDM searches could bound.","The counting rule transfers directly to other axion constructions: adding more PQ-charged scalars or larger gauge representations would generate extra Yukawa closures, so similar atlases could be drawn for other axion models with even larger UV exponents.","For axion dark matter, an enhanced mass at fixed $f_a$ changes the relation between decay constant and relic abundance, so a spectrum that seems excluded in the standard band could reappear as a heavier axion with the same photon coupling in higher-frequency haloscope searches.","The NDA estimates are estimates of the leading size; computing the exact fermion zero-mode overlap integrals for a fully Yukawa-closed vertex would convert the quoted masses into sharper numbers, and could lower them significantly."],"forward_implications":["Small-instanton contributions can become the dominant source of the axion mass: once $\\chi_{\\rm SI}\\gtrsim\\chi_{\\rm QCD}$, the mass shift follows $\\Delta m_a/m_{a,\\rm QCD}\\simeq\\sqrt{\\chi_{\\rm SI}/\\chi_{\\rm QCD}}$ and the axion can weigh tens of MeV with $f_a$ fixed.","At fixed $f_a$ and anomaly ratio $E/N$, $g_{a\\gamma\\gamma}$ is unchanged, so the enhanced mass moves a model horizontally in the $(m_a,g_{a\\gamma\\gamma})$ plane: two copies of $R_{13}$ reach $m_a=71.2$ MeV and the distinct set $S_4$ reaches 65.3 MeV.","The strongest enhancements are $(M_{\\rm Pl}/\\Lambda)^{31/3}$ for a single VQ and $(M_{\\rm Pl}/\\Lambda)^{13}$ for identical copies and distinct VQ sets, so fully Yukawa-closed spectra are the ones that deviate most from the QCD relation.","Several predicted contours lie outside the conventional QCD-axion band but inside currently unexcluded regions; for example, the set $S_2=\\{R_{11},R_{13}\\}$ with $(E/N,N_{\\rm DW})=(34/33,22)$ gives $|g_{a\\gamma\\gamma}|\\simeq3.16\\times10^{-14}$ GeV$^{-1}$ at $m_a\\simeq81.5$ eV.","The atlas also identifies one-VQ spectra that stay close to the QCD relation (e.g., $R_{11}$, $R_{12}$, $R_{13}$) alongside those that leave it ($R_{14}$, $R_{15}$, and multi-copy cases), so the framework classifies models by the size of their expected deviation."],"supporting_citations":[{"why":"Supplies the instanton-NDA power-counting rules that fix how zero modes are saturated and which endpoint of the size integral dominates.","marker":"[48]"},{"why":"Provides the one-instanton measure and the $C_{N_c}$ determinant prefactor entering every susceptibility estimate.","marker":"[36, 58]"},{"why":"Gives the index-theorem zero-mode counting $n_f=2T(R_f)$ that fixes the required saturation.","marker":"[60]"},{"why":"Defines the VQ representations and anomaly coefficients that form the atlas of spectra.","marker":"[53, 54]"},{"why":"Generates the two-loop gauge running used to impose Planck-scale perturbativity and set copy numbers.","marker":"[59]"},{"why":"Provides the QCD topological susceptibility and the standard QCD axion mass formula that the small-instanton term modifies.","marker":"[37, 38]"},{"why":"Establishes the semiclassical small-instanton contribution to the axion potential that this paper extends to a full KSVZ atlas.","marker":"[40, 41]"},{"why":"Supplies the experimental and astrophysical exclusion curves against which the new $(m_a,g_{a\\gamma\\gamma})$ regions are shown to be unexcluded.","marker":"[64]"}],"fun_headline_variants":["Small instantons boost KSVZ axion mass to 71 MeV","Axion mass leaps to 71 MeV via small instantons","Small-instanton axions: mass up, coupling same","KSVZ axion mass hits 71 MeV, photon coupling fixed"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the QCD and small-instanton potentials have the same periodicity and their minima coincide, so their strengths simply add; a relative phase would change the mass-shift formula and generically produce a CP-violating minimum.","fun_headline_variants_meta":{"raw":{"variants":["Small instantons boost KSVZ axion mass to 71 MeV","Axion mass leaps to 71 MeV via small instantons","Small-instanton axions: mass up, coupling same","KSVZ axion mass hits 71 MeV, photon coupling fixed"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000821,"raw_usage":{"total_tokens":3716,"prompt_tokens":1194,"completion_tokens":2522,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":810,"completion_tokens_details":{"reasoning_tokens":2447}},"tokens_in":810,"tokens_out":2522,"duration_ms":20525,"temperature":1.0,"reasoning_tokens":2447,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T04:31:17.542907+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct computation of the one-instanton determinant and the fermion zero-mode overlap integrals for the fully Yukawa-closed vertex of the two-copy $R_{13}$ spectrum at $\\Lambda=5\\times10^{11}$ GeV with $y_Q=1$ would settle the claim: if the exact result comes out orders of magnitude below the NDA estimate, the predicted $m_a\\simeq71.2$ MeV collapses.","supporting_citations":[{"cited_title":"cajohare/axionlimits: Axionlimits,","cited_arxiv_id":null,"evidence_quote":"Supplies the experimental and astrophysical exclusion curves against which the new $(m_a,g_{a\\gamma\\gamma})$ regions are shown to be unexcluded."}],"review_version":1}