{"id":"1f331124-b3bd-4490-b7f0-d3b5d2bfb85f","arxiv_id":"2505.03940","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Two-point correlators of chiral/anti-chiral operators in SU(N) N=2 gauge theories with a non-zero beta function, computed by Feynman diagrams in flat space, match sphere-localization matrix model results exactly through two loops for generic matter representations.","lead":"This paper tests whether a mathematical trick called supersymmetric localization on a sphere can compute ordinary flat-space correlation functions in quantum field theories that are not conformally invariant. It finds that, up to two loops and within a perturbative regime, the two approaches agree exactly, provided carefully accounted evanescent terms are included.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The two-loop match depends on unproved absence of operator mixing; a nontrivial renormalization matrix would change the coefficient in (2.58).","rationale":"The reader's weakest assumption is the same one I judge most load-bearing. The explicit two-loop pole check is strong evidence that no divergent operator mixing is needed at this order, but it does not by itself settle finite same-charge mixing or a possible finite rotation between the field-theory basis and the matrix-model normal-ordered basis. The matching of the 3 zeta(3) C^(2) coefficient is nontrivial and independently derived on both sides, which supports the central claim and keeps the verdict at CONDITIONAL rather than REJECT. No internal inconsistency was found in the color identities or the diagram counting. The proposed check would determine whether footnote 10 is a theorem or a gap.","tokens_in":24299,"tokens_out":31006,"duration_ms":318930,"concrete_test":"Perform a two-loop operator-renormalization analysis in a generic non-conformal N=2 theory with representation R, keeping an explicit mixing matrix Z_{n,m} for operators of equal total charge. Compute the 1/epsilon poles and finite parts of <O_n(x) O_m(0)> in a minimal-subtraction scheme with all counterterm insertions included, and check whether Z_{n,m} reduces to delta_{n,m}(Z_{g*})^n. A sharp case is the off-diagonal pair O_4 and O_{2,2}: reproduce (B.2) by direct Feynman calculation with operator counterterms; a mismatch at order g^8 would invalidate the no-mixing assumption and the claimed match.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central formula (2.58) is obtained by renormalizing only the coupling: footnote 10 asserts that the explicit g_B^n factor in (2.5) lets all UV singularities be absorbed into Z_{g*}. That assertion is not established. In a non-conformal theory an operator made of n elementary fields can acquire a wavefunction renormalization and can mix with same-charge multi-trace operators, so the correct statement is a matrix Z_{n,m}, not a single Z_{g*}. The two-loop diagrams in Section 2 check that the 1/epsilon poles are those of the geometric progression, but they do not determine the finite operator counterterms needed to match the matrix-model normal-ordered operators in (3.18). The matrix model is built with the same no-mixing assumption, so the two computations may agree because both omit the same physics. If mixing exists, C^(2) in (2.59) would acquire extra representation- and partition-dependent terms, and off-diagonal correlators such as G*_{4,(2,2)} would deviate from (B.2).","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper considers SU(N) N=2 gauge theories with massless hypermultiplets in a generic representation R and a non-vanishing beta function. For chiral/anti-chiral operators O_{\\vec n} defined with an explicit overall factor g_B^n, the authors compute the two-point function in flat space by Feynman diagrams in dimensional regularization up to order g_B^{2n+4}. They show that the UV poles arrange into a geometric series that is removed by the one-loop coupling renormalization, while the interference of evanescent O(epsilon) terms with the 1/epsilon poles produces finite two-loop corrections. The resulting renormalized correlator is G^* = g^{2n} G^{(0)} [1 + (g^2/8\\pi^2)^2 3\\zeta(3) C^{(2)} + O(g^6)] with C^{(2)} = 2n(C_R i_R - N^2) + \\hat G_{\\vec n,\\vec m}/G^{(0)}_{\\vec n,\\vec m}. The same correlator is computed from the localization matrix model on S^4 using a diagrammatic expansion with the quartic interaction (g^2/8\\pi^2)^2 \\zeta(3)/2 \\,\\mathrm{Tr}'\\,a^4. The two independent computations agree exactly with the same coefficient C^{(2)}. The paper also gives explicit low-dimension examples and three-loop matrix-model predictions in Section 4.","tokens_in":24458,"tokens_out":11394,"duration_ms":116428,"significance":"If the central claim holds, this is an important extension of supersymmetric localization to non-conformal N=2 theories: it shows that, in the regime \\Lambda \\ll 1/R \\ll \\mu, S^4 localization reproduces flat-space correlators of protected local operators up to two loops. The agreement is nontrivial because it involves the cancellation of the N\\beta_0 terms through the color identities (3.31), and no constant is fitted. The matrix-model diagrammatic method is efficient and yields falsifiable higher-order predictions such as (4.4)-(4.5). The main caveat is that both computations rely on the same assumption about the absence of independent operator renormalization and mixing, which is not established in the manuscript.","major_comments":[{"comment":"The central result (2.58) relies on the assumption that all UV singularities of the composite operators O_{\\vec n} can be absorbed into the single coupling renormalization Z_{g*}. Footnote 10 asserts this because O_{\\vec n} carries an overall factor g_B^n, but that factor only changes the power of the bare coupling; it does not by itself imply the absence of an operator wave-function renormalization or of mixing with same-charge multi-trace operators. The two-loop calculation in Section 2 verifies the pole structure, but it does not determine the finite part of any operator counterterm, which is precisely what would enter C^{(2)} in (2.59). The matrix-model computation in (3.18)-(3.32) uses the same no-mixing assumption, so the two computations could in principle agree because both omit the same physics. I ask the authors to either prove that the chiral/anti-chiral operators in these non-conformal N=2 theories are protected in this normalization (e.g., from the N=2 chiral-ring structure) and do not mix at this order, or compute the relevant operator renormalization matrix and show that its effect cancels in the two-point functions at order g^{2n+4}.","section":"\\S2.5, Eq. (2.58), and footnote 10"},{"comment":"The identification between the normal-ordered matrix-model operators O_{\\vec n}(a) in (3.18) and the bare field-theory operators (2.5) is asserted rather than demonstrated. In the field theory O_{\\vec n}(x) is a composite operator built from elementary fields at the same point, while in the matrix model normal ordering removes all self-contractions. This equivalence may be standard in conformal cases, but in the non-conformal setting the interplay between normal ordering, the explicit g_B^n factor, and the renormalization prescription should be spelled out. If the normal-ordered basis is not an eigenbasis of the renormalization group, the comparison leading to (3.32) is incomplete. This concern is closely tied to the previous one, and a unified treatment would be welcome.","section":"\\S3.1, Eq. (3.18)"}],"minor_comments":[{"comment":"Footnote 10 is the only place where the operator-renormalization issue is discussed; given that it is the main potential gap, the discussion should be expanded and moved into the main text.","section":"\\S2.5"},{"comment":"The sentence \"all explicit dependence on R has disappeared\" is immediately followed by the important qualification that R enters through the running coupling. Please rephrase to avoid the apparent contradiction, for example by saying that the explicit sphere-radius dependence disappears except through the running coupling evaluated at |x|=2R.","section":"\\S3, Eq. (3.13)"},{"comment":"The identification of the matrix-model coupling g_* with the MS-bar coupling of Section 2 is essential to the match, but the scheme relation is only quoted from [53,54]. Please state the precise scheme and explain why the log 4 shift is negligible within the stated regime (1.1).","section":"\\S3, Eq. (3.2)"},{"comment":"The prefactor in G^*_{5,(3,2)} is written as (N^2-2)(N^2-2); this should presumably be (N^2-2)^2. Please check and correct the typo.","section":"Appendix B, Eq. (B.5)"}],"recommendation":"major_revision","confidential_remarks":"The paper is a solid and well-presented two-loop check, but the missing operator-mixing analysis is load-bearing. If the authors can supply a rigorous argument or a reference establishing the absence of operator renormalization/mixing in this normalization, the manuscript would be suitable for publication. I recommend major revision rather than rejection because the gap is fixable within the scope of the paper: it requires either a proof or a clear statement of the assumptions under which the result holds."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Hi, short take on 2505.03940. This is a careful two-loop calculation, and the match with the matrix model is real. It extends the localization/flat-space dictionary from Wilson loops to two-point correlators of chiral/anti-chiral operators in non-conformal N=2 theories, for any matter representation R. The new bits are the evanescent term e_{2,1} for correlators and a diagrammatic matrix-model method that makes the comparison clean. The final formula (2.58)/(3.32) comes out the same from two independent calculations, with the N beta_0 terms cancelling via the color identities (3.31). No fitted constants. That is solid work and I trust the algebra.\n\nThe weak point is footnote 10. The authors say the g_B^n factor in the operator definition lets you absorb all UV poles into the coupling renormalization Z_{g*}, so no separate operator renormalization or mixing is needed. In a non-conformal theory the U(1)_R is anomalous, so the usual chiral-ring protection is gone; mixing between same-charge operators like O_4 and O_{2,2} is not obviously absent. The matrix model is built with the same normal-ordering/no-mixing prescription, so the two sides could in principle be wrong in the same way. Having said that, the Feynman-diagram computation is a genuine direct calculation: the one- and two-loop poles are exactly the coupling-renormalization ones, and the finite parts match the matrix model. That is evidence, not bare assumption. But the authors should either prove the absence of mixing or clearly state that they are assuming it; right now the assertion in footnote 10 is weaker than the rest of the paper.\n\nThe other thing to know: the regularization embedding from the companion papers [53-55] is imported. It is well-motivated, but anyone citing this result should check that the identification (3.2) between the matrix-model coupling and the running coupling is on solid ground.\n\nWho is this for? The localization/N=2 exact-results crowd. It advances a known program by one more observable class and provides a useful diagrammatic tool. It deserves a serious referee. Send it to review, but ask for a direct treatment of the mixing question.","headline":"Solid two-loop match between S^4 localization and flat-space perturbation theory in non-conformal N=2 theories; the unproved no-mixing assumption is the one real caveat.","tokens_in":25021,"tokens_out":13961,"would_cite":true,"duration_ms":144780,"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":"The two-point correlator of chiral/anti-chiral operators in non-conformal N=2 gauge theories, computed by Feynman diagrams in flat space, agrees through two loops with the prediction of supersymmetric localization on a four-sphere.","keywords":["N=2 SYM theory","supersymmetric localization","matrix models","chiral correlators","non-conformal gauge theories","evanescent terms","two-point functions","dimensional regularization"],"falsifier":"Compute the flat-space three-loop two-point function of $O_2$ in SU(N) SQCD with $N_f$ fundamental flavours and compare it with the matrix-model prediction in eq. (4.4): a mismatch in the $\\zeta(5)$ coefficient would show the localization correspondence does not extend to the conjectured next order. Alternatively, compute the renormalization mixing matrix for the operators $O_2$ and $O_{(2,2)}$; any non-diagonal entry at one or two loops would invalidate the single-$Z_{g_*}$ assumption underlying the two-loop match.","tokens_in":24070,"feed_emoji":"🧮","tokens_out":11280,"duration_ms":100325,"temperature":0.7,"pith_summary":"This paper aims to establish that supersymmetric localization on a four-sphere can be trusted for two-point correlation functions even in N=2 gauge theories whose conformal symmetry is broken by a non-zero $\\beta$ function. In the scale regime where the dynamically generated scale $\\Lambda$ is much smaller than the inverse sphere radius $1/R$, which is in turn much smaller than the renormalization scale $\\mu$, the flat-space Feynman-diagram computation and the $S^4$ localization matrix model give the same renormalized correlator through two loops. The agreement relies on a mechanism the paper makes explicit: evanescent contributions, which vanish in exactly four dimensions, interfere with the ultraviolet poles of the bare coupling and produce finite terms at the next perturbative order. A successful match means localization can serve as a shortcut for protected observables in asymptotically free N=2 theories, where direct diagrammatic calculations become increasingly expensive.","feed_headline":"Localization matches flat-space two-loop N=2 correlators","feed_subtitle":"The 4-sphere matrix model could take over the hardest perturbative calculations in asymptotically free N=2 theories.","key_machinery":"The load-bearing mechanism is a difference-theory decomposition of the perturbative expansion. Every vector-multiplet loop correction is shared with N=4 SYM, where all correlators vanish in four dimensions, so it can be rewritten as minus the adjoint-hypermultiplet contribution plus an evanescent remainder; only diagrams with matter in representation $\\mathcal R$ minus adjoint matter need be evaluated. In a conformal theory the evanescent remainder is harmless, but here it multiplies the $1/\\epsilon$ poles of the bare coupling during renormalization and yields finite two-loop corrections, and the paper computes these interference terms explicitly. On the localization side, the matrix model has interaction action $\\mathrm{Tr}'\\log H(aR)$ with the difference-theory trace $\\mathrm{Tr}_{\\mathcal R}-\\mathrm{Tr}_{\\mathrm{adj}}$, whose leading expansion is the quartic vertex $\\frac{\\zeta(3)}{2}\\left(\\frac{g^2}{8\\pi^2}\\right)^2 \\mathrm{Tr}'a^4$. The dictionary is completed by identifying the antipodal distance $2R$ on the sphere with the flat-space separation $|x|$ and the matrix-model coupling with the running coupling at that scale.","core_discovery":"The paper's central claim is that in the regime $\\Lambda \\ll 1/R \\ll \\mu$, the renormalized two-point correlator of chiral/anti-chiral operators $O_{\\vec n}=\\prod_i g^{n_i}\\,\\mathrm{tr}\\,\\phi^{n_i}$ (multi-trace, gauge-invariant operators built from the complex scalar $\\phi$ of the vector multiplet) in $\\mathrm{SU}(N)$ N=2 gauge theory with massless hypermultiplets in a generic representation $\\mathcal R$ is \\begin{equation*} G^*_{\\vec n,\\vec m}=$g^{{2n}}$$G^{{(0)}}$_{\\vec n,\\vec m}\\left[1+\\left(\\frac{$g^{2}$}{8\\$pi^{2}$}\\right)^2 3\\zeta(3)\\,$C^{{(2)}}$_{\\vec n,\\vec m}+O($g^{6}$)\\right], \\end{equation*} with $C^{(2)}_{\\vec n,\\vec m}=2n(C_{\\mathcal R}i_{\\mathcal R}-N^2)+\\hat G_{\\vec n,\\vec m}/G^{(0)}_{\\vec n,\\vec m}$, where $g$ is the running coupling at the separation scale and $C_{\\mathcal R}$, $i_{\\mathcal R}$ are the Casimir and Dynkin index of the matter representation. The same expression is obtained independently from flat-space diagrams and from the $S^4$ matrix model, and the equality holds only after the evanescent one-loop vector-exchange term is combined with the $1/\\epsilon$ poles of the bare coupling. In the matrix model the ubiquitous coefficient $3\\zeta(3)$ comes from the symmetry factors of a single quartic vertex rather than from loop integration, which the paper presents as strong evidence that the two descriptions are computing the same object.","pith_inferences":["If the two-loop equality is the first sign of a broader equivalence, the matrix model could resum these correlators in the large-N limit of asymptotically free N=2 theories, potentially connecting to a holographic description; the paper names this as work in progress, so this is an inference about where the result leads.","The evanescent-pole interference implies that any dimensional-regularization computation in non-conformal theories must postpone setting $\\epsilon=0$ until after coupling renormalization; computations that discard evanescent terms too early will miss finite contributions.","The explicit $\\zeta(5)$ predictions for $O_2$ and $O_3$ in SQCD provide a ready-made falsification target: a direct three-loop flat-space calculation for any one value of $N_f$ would test whether the correspondence extends beyond two loops."],"forward_implications":["Within the stated scale hierarchy, localization on $S^4$ gives reliable perturbative results for chiral/anti-chiral two-point functions in asymptotically free N=2 theories, not only in superconformal ones.","The evanescent-plus-pole interference identified for Wilson loops in earlier work is shown to operate for local operators as well, suggesting it is the general rule for protected observables in non-conformal N=2 theories.","The matrix model produces explicit higher-order formulas, such as the $\\zeta(5)$ terms in eqs. (4.4) and (4.5), that would be very costly to obtain by Feynman diagrams and can be used as concrete predictions.","Outside the perturbative regime, when $\\Lambda R \\sim 1$, the correspondence is expected to break down through power-like infrared corrections, so the statement of agreement is tied to the hierarchy $\\Lambda \\ll 1/R \\ll \\mu$."],"supporting_citations":[{"why":"Establishes the localization matrix model on the four-sphere that the paper uses for the sphere-side computation.","marker":"[2]"},{"why":"Shows chiral/anti-chiral correlators in conformal N=2 SCFTs can be computed as matrix correlators, the starting point for the operator representation.","marker":"[3]"},{"why":"Provides an exact correlation-function computation in SU(2) N=2 superconformal QCD that validates the matrix-model approach for local operators.","marker":"[4]"},{"why":"Gives the general treatment of Coulomb-branch operator correlators from localization, supporting the normal-ordered operator map.","marker":"[5]"},{"why":"Supplies the super-Feynman rules, fusion/fission color identities, and the diagrammatic two-point-function formalism used in both calculations.","marker":"[8]"},{"why":"The predecessor flat-space two-loop computation in non-conformal N=2 SQCD whose diagram functions are generalized here to arbitrary representation R.","marker":"[52]"},{"why":"Introduces the scale regime and the evanescent/UV-pole interference for BPS Wilson loops in non-conformal theories.","marker":"[53]"},{"why":"Extends the Wilson-loop localization match to three loops, providing the direct template for the interference mechanism used here.","marker":"[54]"},{"why":"Argues for the general validity of localization predictions for protected observables in the perturbative regime, the statement this paper transfers to chiral correlators.","marker":"[55]"}],"fun_headline_variants":["Localization confirms flat-space N=2 correlators at two loops","S^4 matrix model matches Feynman diagrams for N=2 correlators","Two-loop N=2 correlators: localization meets perturbation theory","N=2 localization reproduces perturbative correlators","Localization and Feynman agree on two-loop N=2 correlators"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"Everything rests on the assumption that the composite operators require no independent renormalization or mixing beyond the overall power of the bare coupling in their definition; if chiral/anti-chiral operators mix with same-charge multi-trace operators or acquire their own anomalous dimensions, the single-coupling renormalization used here would not remove all ultraviolet divergences and the match would fail.","fun_headline_variants_meta":{"raw":{"variants":["Localization confirms flat-space N=2 correlators at two loops","S^4 matrix model matches Feynman diagrams for N=2 correlators","Two-loop N=2 correlators: localization meets perturbation theory","N=2 localization reproduces perturbative correlators","Localization and Feynman agree on two-loop N=2 correlators"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000767,"raw_usage":{"total_tokens":3472,"prompt_tokens":1086,"completion_tokens":2386,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":702,"completion_tokens_details":{"reasoning_tokens":2294}},"tokens_in":702,"tokens_out":2386,"duration_ms":18261,"temperature":1.0,"reasoning_tokens":2294,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T23:42:17.461576+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the flat-space three-loop two-point function of $O_2$ in SU(N) SQCD with $N_f$ fundamental flavours and compare it with the matrix-model prediction in eq. (4.4): a mismatch in the $\\zeta(5)$ coefficient would show the localization correspondence does not extend to the conjectured next order. Alternatively, compute the renormalization mixing matrix for the operators $O_2$ and $O_{(2,2)}$; any non-diagonal entry at one or two loops would invalidate the single-$Z_{g_*}$ assumption underlying the two-loop match.","supporting_citations":[],"review_version":1}