{"id":"d8e13928-e3fb-4e04-99e8-656a6c7db9c3","arxiv_id":"1908.03779","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"The authors recompute the T_F-proportional pieces of the polarized three-loop QCD anomalous dimensions using massive operator matrix elements and find full agreement with the published 2014 results.","lead":"An independent QCD calculation reproduces the known three-loop polarized splitting functions for the parts proportional to the color factor T_F, confirming the 2014 result from Moch, Vermaseren and Vogt. The work supplies technical ingredients for next-to-next-to-leading-order predictions of polarized deep-inelastic scattering at colliders such as the EIC and RHIC.","discovery_kind":"replication","skeptic_critique":{"model":"deepseek-v4-flash","headline":"gamma_qg rests on an unproved guessed recurrence from 2640 Mellin moments; agreement with Ref. [6] is strong but does not remove the need for a proof or a documented higher-moment validation.","rationale":"The paper is a technically demanding independent recalculation of the T_F-proportional polarized three-loop anomalous dimensions, and it provides multiple genuine consistency checks: exact agreement with Ref. [6], independent two-loop results, the N=1 checks, the beta-function check, and the large-N_F comparisons. These checks make the central claim highly plausible, and I do not see evidence of circularity, data manipulation, or overclaim. The weakest point is indeed the gamma_qg^(2) derivation, exactly as the reader identified: a finite-moment guessing step produces the recurrence from which the all-N result is obtained, with no formal proof that the recurrence is valid for every N. The external agreement with Ref. [6] is strong practical evidence, but the paper's own stated goal is an independent calculation, and an independent derivation should either prove the guessed recurrence or clearly document validation on held-out moments. The paper mentions that 4000 moments were generated and 2640 were sufficient, which suggests such a validation may exist, but it is not described, and the ancillary Mathematica files that would allow inspection are not present in the submission. This warrants the CONDITIONAL verdict already issued, but it does not by itself justify rejection or a stronger verdict. My concern does not move the reader's verdict, so I recommend UNCHANGED.","tokens_in":39192,"tokens_out":5028,"duration_ms":63017,"concrete_test":"Generate the 1/epsilon-pole moments of A^(3)_Qg for odd N = 2641, 2643, ..., 3999 directly from the IBP recurrence system used to produce the first 2640 moments, apply the Larin-to-M finite renormalization of Eq. (40), and compare each value with Eq. (79) evaluated at the same N. If all 1360 additional moments agree, the guessed recurrence is validated well beyond the fitted range; if any disagree, the all-N expression for gamma_qg^(2) is not correct. As a stronger formal check, use Sigma or certified creative telescoping to derive the same recurrence from the IBP system rather than by guessing, and verify that the recurrence in Table 1 matches for all N.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim includes the complete three-loop anomalous dimension gamma_qg^(2). In Section 3 and Table 1, the all-N result for gamma_qg^(2) is obtained by generating Mellin moments of the 1/epsilon pole of the massive OME A^(3)_Qg, guessing a difference equation from finitely many moments using the method of Ref. [20], and solving that recurrence with Sigma. The paper states that 4000 moments were generated and that 2640 turned out to be sufficient, but it does not prove that the guessed recurrence is the true recurrence of the moment sequence for all odd N, nor does it document that the additional 1360 moments were used as an independent validation. In principle a different recurrence could reproduce the first 2640 moments and disagree later; if that happened, the resulting closed form for gamma_qg^(2) would be an artifact of the guessing step rather than the true anomalous dimension. Because gamma_qg^(2) is one of the two complete three-loop quantities claimed, this is the most load-bearing assumption in the argument. The agreement with Ref. [6] is a strong external check, and it substantially reduces the practical risk, but it does not formally close the gap in the independent derivation. The absence of the announced Mathematica ancillary files means the reader cannot inspect the guessing/validation protocol directly.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper computes all contributions proportional to T_F to the polarized three-loop anomalous dimensions in the M-scheme, using on-shell massive operator matrix elements (OMEs) in the Larin scheme followed by a finite renormalization to the M-scheme. The complete anomalous dimensions γ_qq^{(2),PS} and γ_qg^{(2)} are obtained, together with the T_F parts of γ_gq^{(2)} and γ_gg^{(2)}, and the complete two-loop polarized anomalous dimensions are reproduced independently. Most master integrals are evaluated with the standard toolkit (hypergeometric representations, differential equations, Almkvist–Zeilberger, Sigma, HarmonicSums); for the OME A_Qg^{(3)} the method of arbitrarily high Mellin moments is used, with a difference equation for the moment sequence obtained by guessing and then solved with Sigma. The paper reports full agreement with the earlier massless computation of Ref. [6] and provides cross-checks against MATAD moments for N=1,3,5,7,9, the N=1 limits, the three-loop β-function, and large-N_F predictions. Splitting functions in z-space and corrected operator Feynman rules are given in the appendices.","tokens_in":39399,"tokens_out":6217,"duration_ms":65305,"significance":"If the results stand, this is a valuable independent confirmation of the polarized NNLO anomalous dimensions from a physically different calculational setup, which is important for polarized deep-inelastic phenomenology and for the EIC program. The paper's strengths are the multiple internal cross-checks (low Mellin moments from MATAD, N=1 sum rules, β-function consistency, large-N_F comparisons) and the final all-N agreement with Ref. [6]. No parameters are fitted to the target result; the comparison with Ref. [6] is an external benchmark. The main caveat is that the γ_qg^{(2)} result relies on an unproved guessed recurrence, so the strength of the claim of a 'fully independent' derivation is limited unless that recurrence is certified or validated on held-out moments.","major_comments":[{"comment":"The all-N expression for γ_qg^{(2)} is obtained by guessing a difference equation from 2640 Mellin moments and then solving that recurrence with Sigma. The manuscript does not prove that the guessed recurrence is the true recurrence of the moment sequence for all odd N. Since γ_qg^{(2)} is one of the two complete three-loop quantities claimed, this is a load-bearing step in the argument. Please either (i) supply a proof of the recurrence, or (ii) document that the remaining 1360 of the 4000 generated moments were held out and used as an independent validation, and state explicitly that the recurrence reproduces them; ideally both. The agreement with Ref. [6] is a strong external check of the final result, but it does not by itself close the formal gap in the claimed independent derivation.","section":"Section 3, Table 1; Section 6, Eq. (79)"}],"minor_comments":[{"comment":"The abstract states that 4000 moments were generated and 2640 turned out to be sufficient, while Section 3 says 2000 moments were generated for most projections and 4000 for the C_F C_A T_F and C_A^2 T_F projections. Please reconcile these statements and clarify whether the 1360 unused moments were employed as a validation set or only as part of the guessing input.","section":"Abstract and Section 3"},{"comment":"The conclusions promise that the results are provided in Mathematica ancillary files, but the manuscript version made available for review does not include these files. Since they are the only way to inspect the guessing and validation protocol, please ensure the ancillary files are submitted with the revised version.","section":"Section 8"},{"comment":"In the last term of the four-gluon operator vertex, the exponent 'm−j−4' contains an undefined symbol m; this should presumably be 'N−j−4' to match the other sums in the equation.","section":"Appendix A, Eq. (144)"},{"comment":"The statement that the next-to-leading 1/N^3 terms 'more than cancel' the leading 1/N^5 terms is qualitative. Consider displaying the subleading coefficients or giving a short numerical illustration to substantiate the claim.","section":"Section 7, Eqs. (132)-(136)"}],"recommendation":"major_revision","confidential_remarks":"The core physics result is very likely correct given the agreement with Ref. [6] and the number of internal consistency checks. My concern is not the numerical result but the epistemic status of the 'fully independent' claim for γ_qg^{(2)}: an unproved guessed recurrence is acceptable as a practical method only if the validation protocol is documented. If the authors either certify the recurrence or demonstrate a held-out validation on the remaining 1360 moments, I would support acceptance. Please also verify that the promised ancillary files are actually included in the submission."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Candid take: this is a confirmation paper with real technical substance, and the confirmation is worth having. The authors recompute the T_F pieces of the polarized three-loop anomalous dimensions through massive operator matrix elements, in QCD only, and reproduce the earlier Moch-Vermaseren-Vogt result exactly. The technical route is genuinely new: the modified quarkonic projector in the Larin scheme, the correction of two operator Feynman rules from Mertig-van Neerven, and the use of arbitrarily high Mellin moments to sidestep elliptic sectors for gamma_qg. Those are concrete contributions, not packaging.\n\nThe internal checks are solid: fixed moments from MATAD, the N=1 limits, the beta-function check, and the exact agreement with the external benchmark. For a calculation this size, that is strong evidence the final expressions are right.\n\nThe soft spots. The gamma_qg derivation relies on guessing a recurrence from 2640 computed Mellin moments and solving it with Sigma. The paper does not prove that the guessed recurrence is the true recurrence for all odd N, and it does not say whether the extra 1360 moments were reserved as a validation set. The stress-test note is correct that a different recurrence could in principle agree on the first 2640 moments and diverge later. I would not call this fatal: the final answer agrees with Ref. [6], so the physical claim does not hang on the guessing step alone. But as an independent derivation it is a gap, and the authors should either prove the recurrence or document a held-out validation. The announced Mathematica ancillary files are not actually in the submission; for a paper whose value is reproducibility, that is a fixable but real deficiency. The citation pattern is normal for this group: self-citations point mostly to their own summation and OME methods, and the external benchmarks are properly credited.\n\nThe reader's conditional verdict seems right to me. This is not a new-physics paper; it is a high-cost independent confirmation with a couple of methodological improvements. But for anyone working on polarized NNLO evolution or heavy-flavor DIS, it is a useful benchmark and a warning that the earlier Feynman rules needed correcting.\n\nMy recommendation: send it to a specialist referee. The main things I would ask the referee to check are the guessing/validation protocol for gamma_qg and the finite renormalization to the M-scheme. Those are exactly the places where a paper like this can hide a subtle error. I would not cite it in my own next twelve months, but that is because I do not work in this subfield.","headline":"A technically heavy independent recalculation that confirms the known polarized three-loop anomalous dimensions; the gamma_qg recurrence step is a real but bounded gap, and the missing ancillary files should be fixed.","tokens_in":40007,"tokens_out":2744,"would_cite":false,"duration_ms":30370,"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":"Polarized three-loop QCD splitting functions, computed by an independent massive-operator method, agree with the earlier result in every channel.","keywords":["polarized deep-inelastic scattering","three-loop anomalous dimensions","massive operator matrix elements","splitting functions","Mellin moments","Larin scheme","NNLO QCD","heavy flavor"],"falsifier":"Generate one additional Mellin moment of the $1/\\varepsilon$ pole of the unrenormalized massive OME $A_{Qg}^{(3)}$ at an $N$ beyond the 4000 moments used in the paper, by a direct fixed-$N$ evaluation of the master integrals that does not rely on the guessed recurrence, and test whether the published closed form for $\\gamma_{qg}^{(2)}$ matches it; a single mismatch would disprove the all-$N$ claim, while agreement would add evidence but not proof.","tokens_in":38949,"feed_emoji":"⚛️","tokens_out":12748,"duration_ms":126442,"temperature":0.7,"pith_summary":"This paper tries to establish, by an independent route, the complete $T_F$-dependent part of the polarized three-loop QCD anomalous dimensions, the kernels that control how spin-dependent quark and gluon densities evolve with energy scale; $T_F$ is the color factor of a closed quark loop. Working from massive on-shell operator matrix elements rather than from the approach behind the earlier computation, the paper obtains closed forms in Mellin-$N$ space for the full $\\gamma_{qq}^{(2),\\mathrm{PS}}$ and $\\gamma_{qg}^{(2)}$, plus the $T_F$ parts of $\\gamma_{gq}^{(2)}$ and $\\gamma_{gg}^{(2)}$, and reports complete agreement with the previous result. This matters because these kernels are the last ingredient needed to promote polarized deep-inelastic-scattering and collider spin analyses from next-to-leading to next-to-next-to-leading order, and because an independent confirmation protects precision spin and $\\alpha_s$ measurements from a possible systematic error in the only earlier calculation. As byproducts, the paper independently recovers the two-loop polarized anomalous dimensions and the $T_F$ part of the three-loop QCD $\\beta$-function.","feed_headline":"Spin-dependent QCD evolution verified at three loops","feed_subtitle":"The kernels that evolve quark and gluon spin distributions at NNLO now rest on two independent calculations.","key_machinery":"The carrying objects are the massive on-shell operator matrix elements (OMEs): Green functions with local operator insertions whose ultraviolet poles in $\\varepsilon = D-4$ contain the anomalous dimensions, at order $1/\\varepsilon^3$ for one loop, $1/\\varepsilon^2$ for two loops, and $1/\\varepsilon$ for the $T_F$ parts of the three-loop result. For most channels the master integrals are expanded by standard differential-equation and summation methods. The exception is $A_{Qg}^{(3)}$, where the needed deeper $\\varepsilon$-expansion touches elliptic sectors; there the paper uses the method of arbitrarily high Mellin moments. The integration-by-parts relations are converted into recurrences for the master integrals, thousands of moments are generated exactly, a difference equation for the moments of the OME's $1/\\varepsilon$ coefficient is obtained by a guessing algorithm, and the recurrence, which is first-order and factorizable, is solved in terms of nested harmonic sums. The Larin scheme fixes the treatment of $\\gamma_5$, and a finite renormalization converts the Larin-scheme anomalous dimensions to the M-scheme where the comparison with the earlier result is made.","core_discovery":"The central claim is that the contributions proportional to $T_F$ to the polarized three-loop anomalous dimensions in the M-scheme can be computed from massive on-shell operator matrix elements, and that the resulting closed expressions agree with the earlier computation in every channel. For $\\gamma_{qq}^{(2),\\mathrm{PS}}$ and $\\gamma_{qg}^{(2)}$ the paper gives complete results valid for all Mellin moments; for $\\gamma_{gq}^{(2)}$ and $\\gamma_{gg}^{(2)}$ it gives the full $T_F$-dependent parts. The derivation is not a rerun of standard methods: for $\\gamma_{qg}^{(2)}$, deeper expansions of the master integrals in the dimensional parameter $\\varepsilon = D-4$ would have introduced elliptic contributions, so the paper generates up to 4000 Mellin moments of the relevant operator matrix element, guesses a first-order factorizable recurrence from 2640 of them, solves that recurrence in closed form, and then transforms from the Larin scheme to the M-scheme by a finite renormalization. All obtained anomalous dimensions are stated to agree with the previous computation.","pith_inferences":["The unproven step is the guessed recurrence for $\\gamma_{qg}^{(2)}$; if a future evaluation of a moment beyond the fitted range disagreed with it, only that anomalous dimension would be affected, since the other channels use direct methods.","The same high-moment pipeline is a natural tool for the next stage of the project, the $O(\\varepsilon)$ terms of $A_{Qg}^{(3)}$ that enter the massive polarized Wilson coefficients at NNLO, because those terms are expected to carry the same elliptic complications.","The corrected quarkonic projector may have consequences beyond anomalous dimensions: earlier polarized massive OME results obtained with the alternative projector could deserve re-examination even where the final anomalous dimensions agree.","The small-$z$ analysis implies that the leading $1/N^5$ terms are not numerically dominant because the next-order terms largely cancel them, so using only the leading small-$x$ pole would misestimate NNLO spin evolution in phenomenological fits."],"forward_implications":["The polarized NNLO evolution of quark and gluon spin densities can now be implemented with a cross-checked set of splitting functions, allowing existing NLO analyses of polarized deep-inelastic-scattering data to be promoted to NNLO.","The first-moment identities are reproduced, including $\\gamma_{gg}^{(k)}(N=1)=-2\\beta_k$ and $\\gamma_{qg}^{(k)}(N=1)=0$, so the new expressions respect the axial anomaly and fermion-number conservation.","The large-$N_F$ predictions from the literature are recovered after a small identified finite-renormalization adjustment, with one term in the earlier pure-singlet prediction traced to a missing M-scheme conversion.","The two corrected operator Feynman rules from the two-loop literature do not change the earlier two-loop anomalous dimensions, which are confirmed as correct.","Because the Mellin-space results invert to $z$-space splitting functions given in the appendix, the expressions are in the form needed for convolution codes and global spin fits."],"supporting_citations":[{"why":"the earlier computation of the polarized three-loop splitting functions in the M-scheme that the present massive calculation independently reproduces.","marker":"[6]"},{"why":"supplies the method of arbitrarily high Mellin moments used to obtain the moments of $A_{Qg}^{(3)}$ when elliptic contributions block standard techniques.","marker":"[19]"},{"why":"supplies the guessing method that produces the difference equation whose solution gives $\\gamma_{qg}^{(2)}$.","marker":"[20]"},{"why":"defines the Larin scheme for $\\gamma_5$ in dimensional regularization, the starting scheme from which the results are finite-renormalized.","marker":"[23]"},{"why":"gives the finite renormalization connecting Larin-scheme and M-scheme anomalous dimensions at the required orders.","marker":"[24]"},{"why":"provides the two-loop polarized splitting functions and the operator Feynman rules that the paper corrects; its two-loop results remain valid.","marker":"[7]"},{"why":"the unpolarized three-loop massive OME calculation whose master integrals and techniques are reused for most of the channels treated here.","marker":"[16]"}],"fun_headline_variants":["Polarized NNLO anomalous dimensions: two independent ways agree","Massive operator matrix elements nail three-loop spin evolution","Closed-form three-loop spin kernels from 2640 Mellin moments","Independent check of polarized three-loop QCD evolution","Elliptic-free path to polarized three-loop anomalous dimensions"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"That the recurrence for $\\gamma_{qg}^{(2)}$ guessed from 2640 computed Mellin moments is the true recurrence for every $N$; the paper validates it by agreement with the earlier result rather than by proof.","fun_headline_variants_meta":{"raw":{"variants":["Polarized NNLO anomalous dimensions: two independent ways agree","Massive operator matrix elements nail three-loop spin evolution","Closed-form three-loop spin kernels from 2640 Mellin moments","Independent check of polarized three-loop QCD evolution","Elliptic-free path to polarized three-loop anomalous dimensions"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000256,"raw_usage":{"total_tokens":1582,"prompt_tokens":958,"completion_tokens":624,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":574,"completion_tokens_details":{"reasoning_tokens":544}},"tokens_in":574,"tokens_out":624,"duration_ms":7400,"temperature":1.0,"reasoning_tokens":544,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:01:59.265426+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Generate one additional Mellin moment of the $1/\\varepsilon$ pole of the unrenormalized massive OME $A_{Qg}^{(3)}$ at an $N$ beyond the 4000 moments used in the paper, by a direct fixed-$N$ evaluation of the master integrals that does not rely on the guessed recurrence, and test whether the published closed form for $\\gamma_{qg}^{(2)}$ matches it; a single mismatch would disprove the all-$N$ claim, while agreement would add evidence but not proof.","supporting_citations":[{"cited_title":"The Method of Arbitrarily Large Moments to Calculate Single Scale Processes in Quantum Field Theory","cited_arxiv_id":"1701.04614","evidence_quote":"supplies the method of arbitrarily high Mellin moments used to obtain the moments of $A_{Qg}^{(3)}$ when elliptic contributions block standard techniques."},{"cited_title":"Matiounine, J","cited_arxiv_id":null,"evidence_quote":"gives the finite renormalization connecting Larin-scheme and M-scheme anomalous dimensions at the required orders."}],"review_version":1}