{"id":"4cc77f5c-e94b-44f4-8d65-1e7995a720b8","arxiv_id":"2603.29803","paper_version":5,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Fits of combined KEDR and truncated BESIII data yield α_s(M_Z)=0.1179 (NLO), 0.1221 (NNLO), 0.1313 (N3LO), with the rise attributed to π² analytic-continuation effects.","lead":"This paper fits electron-positron collision data to measure the strong-force coupling, α_s, at low energies. The result depends on how many QCD correction terms are kept, drifting from 0.118 to 0.131 and exposing a known tension with part of the BESIII data.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Massless f=3 pQCD without non-perturbative power corrections is the load-bearing assumption: the NLO-to-N3LO drift is KEDR-driven (Table 2), while the BESIII covariance reconstruction is a real but secondary issue.","rationale":"The paper is an honest, internally consistent fit study of published data. The KEDR covariance is taken from the experiment, the chi^2_1 model is explicitly described, and the order-by-order drift is clearly shown. The reader's primary concern about the reconstructed BESIII covariance is real, but it is not the most load-bearing point for the headline alpha_s values: the KEDR-only fits in Table 2 give NLO 0.1182, NNLO 0.1222, N3LO 0.1333, and the combined Table 5 values are 0.1179, 0.1221, 0.1313. Thus even a very different BESIII covariance, or dropping BESIII entirely, leaves the central trend intact. The deeper issue is that the theoretical prediction is a massless, purely perturbative R(s) with no power corrections and no assigned theoretical uncertainty, in an energy region where 1/s^2 corrections and charm-mass effects are not negligible. Because the N3LO term is large and negative, missing non-perturbative contributions could be absorbed by changing alpha_s by an amount comparable to the quoted experimental errors. A concrete power-correction refit would settle whether the pi^2 interpretation is actually isolated. I also note the abstract/body numeric discrepancy (v1 abstract gives 0.1151/0.1190/0.1283 vs body Eqs. 31-33 giving 0.1179/0.1221/0.1313); this is a consistency flaw that should be fixed, but it does not change the scientific assessment. The appropriate verdict remains conditional acceptance, with the added condition that the fit be tested against non-perturbative power corrections and that the numerical inconsistency be resolved.","tokens_in":21724,"tokens_out":21378,"duration_ms":209031,"concrete_test":"Refit the KEDR-only and combined datasets with an added free power-correction term R_th(s) -> R_th(s) + C/s^2 at NLO, NNLO, and N3LO, and report how the central alpha_s(M_Z) values (Eqs. 31-33) change. If alpha_s(N3LO) moves by more than the quoted fit uncertainty (~0.007) or if C is preferred at >2 sigma, the massless-pQCD assumption is load-bearing and the central claim must be made conditional on the power-correction model. In the same numerical run, also replace the reconstructed BESIII covariance (Eq. 27) with alternative correlation assumptions (all systematic sources uncorrelated; all fully correlated) to check whether the full-dataset incompatibility and the 6-point truncation survive.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing assumption is the applicability of the massless f=3 fixed-order expansion (Eq. 20) to the whole fit range sqrt(s)=1.84-3.72 GeV. At these energies a_s=alpha_s/pi ~ 0.08-0.10, so the O(a_s^3) coefficient -10.377 in Eq. (20) changes R by about -0.008 to -0.010. That is the same size as the expected non-perturbative power corrections (gluon condensate / duality violations, ~O(Lambda^2/s)) and as unsubtracted charm-mass effects near the open-charm threshold. Since the KEDR fits have chi^2/ndf ~ 4-6/21, a shift of the theoretical curve by ~1% can be absorbed by a much larger change in Lambda_MS^(3); the analysis includes no such term and assigns no theoretical uncertainty. Thus the claimed association of the alpha_s increase with 'not totally controlled pi^2 contributions' is not uniquely supported: a 1/s^2 power correction would move N3LO alpha_s downward, and the paper does not test this. The BESIII covariance reconstruction (Eq. 27) could alter the chi^2 ~ 54/13 incompatibility, but the central combined values differ from KEDR-only values by <=0.002, so it is not the deciding input for the headline numbers.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper compares the KEDR (1.84–3.72 GeV) and BESIII (2.23–3.67 GeV) measurements of the e+e− hadronic R-ratio with fixed-order massless f=3 perturbative QCD expressions, using the MS-bar scheme and the FOPT running coupling. By minimizing χ²0 and a modified χ²1 with an extra normalization parameter ν, the authors extract Λ_MS^(3) and then α_s(M_Z) at NLO, NNLO, N3LO, and an estimated N4LO. For the combined KEDR + truncated BESIII data set (the six BESIII points below the J/ψ mass), the headline results are α_s(M_Z)=0.1179^{+0.0051}_{−0.0069} (NLO), 0.1221^{+0.0063}_{−0.0080} (NNLO), and 0.1313^{+0.0027}_{−0.0067} (N3LO). The authors attribute the order-by-order rise to analytically continued π² contributions that change the signs of the higher-order coefficients, and they comment on the disagreement with other α_s determinations. The full 14-point BESIII sample is found to be incompatible with pQCD (χ²0/ndf ≈ 54/13), which motivates the truncation.","tokens_in":22125,"tokens_out":6153,"duration_ms":68812,"significance":"If the result were fully robust, the paper would provide a new low-energy α_s determination from recent R-ratio data and a concrete illustration of FOPT π² effects below the charm threshold. The manuscript is transparent in stating the fit function, the order-by-order evolution, and the ad hoc elements of the analysis; the tables are informative and the KEDR-only fits are internally consistent. The historical summary and the explicit separation of correlated/uncorrelated BESIII systematics are also useful. However, the significance is currently conditional: the headline combined fits rest on a post-hoc BESIII data selection, on a covariance matrix reconstructed from published error summaries rather than the true correlation matrix, and on the omission of any non-perturbative power-correction term. These issues directly affect the central numerical claims, especially the N3LO value and its interpretation.","major_comments":[{"comment":"The BESIII covariance matrix is reconstructed by treating the 'last four sources' of systematic uncertainty as fully correlated across energy points and the remaining three as fully uncorrelated, based only on the qualitative caption of Table I of Ref. [34]. This is a strong assumption that is not validated against the experiment's internal correlations. Since the conclusion that the full 14-point BESIII sample is incompatible with pQCD (χ²0/ndf ≈ 54/13, Table 3) and the subsequent decision to truncate the sample both depend on this matrix, the paper should provide a sensitivity study (e.g., varying the correlation coefficients between 0 and 1 for the four correlated sources) and show how the χ² values and the Table 5 α_s values change. Without such a test, the statistical rejection of BESIII data is partly an artifact of an assumption.","section":"§5, Eq. (27)"},{"comment":"The exclusion of the eight BESIII points above 3.4 GeV is decided after observing the large χ²0/ndf ≈ 54/13 for the full sample, and is motivated by the known tension of those points with KEDR and with pQCD. Using the outcome of the fit to select the data then biases the combined results in Table 5: the 'truncated BESIII' agreement is not an independent confirmation. The paper needs an a priori, physically motivated selection rule — for example, a threshold justified independently of the χ² comparison, with results also shown for the full 14-point sample under a robust treatment of the discrepancy (e.g., an additional correlated systematic parameter). Without this, the combined NLO, NNLO, and N3LO values in Eqs. (31)–(33) are not supported as a single coherent data set.","section":"§5, Tables 3 and 4"},{"comment":"The fit function is the massless f=3 fixed-order expansion with no non-perturbative terms. In the fit range √s = 1.84–3.72 GeV, a_s ≈ 0.08–0.10, so the O(a_s³) coefficient −10.377 changes R by roughly −0.008 to −0.010, which is comparable to expected Λ²/s power corrections and to unsubtracted charm-mass effects near the open-charm threshold. Since the KEDR fits have χ²0/ndf ≈ 4–6/21, a 1% shift in the theoretical curve can be absorbed into a substantially different Λ_MS^(3). The conclusion that the NLO→N3LO rise is due to 'not totally controlled π² contributions' is therefore not unique: adding a 1/s² power correction or an OPE parameter would likely move the N3LO α_s downward. The paper should test this by including a power-correction term in a simultaneous fit, or else quote a theoretical uncertainty that covers such effects. As written, the N3LO value of Eq. (33) is over-interpreted.","section":"§2 Eq. (20), §6 Eqs. (31)–(33)"},{"comment":"The quoted uncertainties on α_s(M_Z) are the experimental fit errors only. The central message of the paper is the order-by-order drift: 0.1179 (NLO), 0.1221 (NNLO), 0.1313 (N3LO), and 0.1268 (N4LO estimate). If truncation effects are the subject, the comparison should include a truncation/scale uncertainty, for example by varying the renormalization scale or by treating the order-by-order spread as a systematic error. Without such an error, the apparent agreement of the NLO/NNLO values with the world average and the disagreement of the N3LO value are not quantified on the same footing as the experimental errors, and the 'tendency of growth' is presented with overprecise error bars.","section":"§6, Table 5"}],"minor_comments":[{"comment":"There are several typos and stylistic issues: 'Bejing' should be 'Beijing', 'coordially' should be 'ordinarily' or similar, 'massles' appears in the text, and the abstract contains duplicated punctuation ('BESIII data ,').","section":"Abstract and text"},{"comment":"The informal notation 'χ²0 ≈ 50/13 (?!)' and 'χ²1 ≈ 51/12 (?!)' is not appropriate for a journal report; the question marks and exclamation marks should be removed and replaced by a clear statement of the statistical significance.","section":"§5, Table 3"},{"comment":"The figures contain 'PSfrag replacements' and Cyrillic text ('R по результатам КМД-3...'), which appear to be LaTeX/artifact labels. The figures should be regenerated with clean, readable axis labels and legends.","section":"Figure 1 and Figure 2"},{"comment":"The underlining and double-underlining in Eq. (29) are not explained in the text. The reader cannot tell which terms are 'directly Euclidean' and which are 'analytical continuation contributions' without an explicit key or a more detailed explanation in the caption or text.","section":"§6, Eq. (29)"},{"comment":"The notation Δ_k(f) is introduced without a clear statement of its argument (the number of flavors f) and its relation to the coefficients d_k(f). This makes the reader reconstruct the sign convention; a one-sentence definition would help.","section":"§2, Eq. (17)"},{"comment":"The reference list contains entries that appear to be from the future relative to standard arXiv dates (e.g., Ref. [62], Ref. [40]). While this may be a preprint artifact, the authors should ensure all bibliographic entries are complete and correctly dated.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The paper is in scope for a QCD phenomenology journal and the authors have made a serious, transparent attempt to fit the new KEDR and BESIII data. My main concern is not the internal consistency of the fits but the fact that the central numerical claims depend on three fragile ingredients: a reconstructed BESIII covariance matrix, a post-hoc exclusion of eight BESIII points, and the absence of non-perturbative power corrections. The latter is the most important: the N3LO result can be shifted by a 1/s² term of the natural size, and the paper does not test this. I would require a sensitivity analysis and a proper theoretical uncertainty before this can be accepted as an α_s determination. The historical review and the explicit comparison of χ²0 and χ²1 are useful contributions."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Honest, internally consistent fit study, but the headline numbers should not be treated as final. The genuinely new piece is that they fit alpha_s directly from KEDR and truncated BESIII data instead of fixing it to the world average as Boito and Caram did. The NLO and NNLO combined fits give 0.1179 and 0.1221, consistent with the world average; the N3LO drifts to 0.1313, which they attribute to unsubtracted pi^2 effects. The tables and the order-by-order discussion are useful.\n\nThe soft spots are real. The exclusion of eight of fourteen BESIII points above the J/psi mass is post-hoc, and the BESIII covariance matrix is reconstructed from the published error table rather than the true correlations. That matters for the claim that the full BESIII dataset is incompatible with pQCD (chi^2 ~ 54/13), though it barely affects the combined alpha_s values, which are KEDR-dominated.\n\nMore load-bearing is the theory side: the fits assume massless f=3 pQCD with no non-perturbative power corrections in the 1.8-3.7 GeV region. The KEDR fits have chi^2/ndf around 4-6/21, so a 1% shift in the theory curve can be absorbed by a large change in Lambda_MS. A 1/s^2 power correction could move the N3LO alpha_s down and would weaken the claimed association of the rising alpha_s with pi^2 effects. The paper does not test this, and assigns no theory uncertainty for it.\n\nThere are also smaller issues: the abstract quotes different numbers (0.1151, 0.1190, 0.1283) than the body's final values (0.1179, 0.1221, 0.1313), and the N4LO estimate is a bare number without uncertainty. These are fixable but sloppy.\n\nBottom line: the paper deserves a serious referee. The method is standard chi^2 fitting but the extraction is new, the results are potentially useful as a low-energy alpha_s cross-check, and the N3LO drift is a genuine puzzle. The authors need to address the theory uncertainty, justify the BESIII truncation, and reconcile the abstract with the body. Send it to review, conditional on revision.","headline":"Genuinely new alpha_s extraction from KEDR+truncated BESIII, but the N3LO drift is not uniquely pinned to pi^2 effects and the abstract doesn't match the body.","tokens_in":22600,"tokens_out":3306,"would_cite":false,"duration_ms":30912,"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":"Fitting KEDR and truncated BESIII data to massless QCD yields alpha_s(M_Z) values that rise with each perturbative order, from 0.1179 at NLO to 0.1313 at N3LO.","keywords":["R-ratio","e+e- annihilation","alpha_s determination","perturbative QCD","analytic continuation","pi^2 contributions","KEDR","BESIII"],"falsifier":"If the BESIII collaboration released its full correlation matrix and a refit of all 14 points with that matrix gave chi-squared per degree of freedom near 1 and order-stable alpha_s values, the paper's central claims—that the full dataset is incompatible with pQCD and that the order-dependence is driven by pi^2 terms—would be falsified.","tokens_in":21606,"feed_emoji":"⚛️","tokens_out":6067,"duration_ms":56533,"temperature":0.7,"pith_summary":"The paper fits recent low-energy electron-positron annihilation data to the perturbative QCD prediction for the R-ratio, treating the strong coupling as a free parameter. Combining 22 KEDR points with the six BESIII points below the J/psi mass, the authors extract alpha_s(M_Z)=0.1179 at next-to-leading order, 0.1221 at next-to-next-to-leading order, and 0.1313 at next-to-next-to-next-to-leading order. The systematic rise is traced to the pi-squared terms that appear when the QCD series is analytically continued from the Euclidean to the time-like region, which change the sign pattern of the R-ratio coefficients. The full 14-point BESIII dataset is found to be incompatible with massless perturbative QCD, and the eight points above the J/psi mass are excluded; the remaining six agree with KEDR. The NLO and NNLO results agree with the world average coupling, while the N3LO result does not, indicating that the fixed-order time-like series is not yet under control.","feed_headline":"Alpha_s climbs from 0.1179 to 0.1313 as QCD orders stack up","feed_subtitle":"Order-dependent alpha_s values flag uncontained pi-squared terms in time-like QCD.","key_machinery":"The central object is the relation between the Euclidean Adler function D(Q^2) and the time-like R-ratio, R(s)=12 pi Im Pi(-s+i epsilon), implemented through the analytic-continuation shifts Delta_k(f) that convert the Euclidean coefficients d_k into the Minkowski coefficients r_k = d_k - Delta_k. The Delta_k contain powers of pi^2 (for example Delta_3 = pi^2 beta_0^2 d_1/3), and they change the sign structure of the series from sign-constant in the Euclidean region to sign-alternating in the time-like region. These shifts, together with the chi^2_1 minimization function that adds a free normalization nu, carry the argument that the extracted coupling's order-dependence is driven by uncontai","core_discovery":"The central claim is that the combined KEDR and truncated BESIII fits provide new, order-dependent determinations of alpha_s(M_Z) from the R-ratio below charm threshold, and that the order dependence is a physical signal rather than a statistical artifact. Using massless f=3 pQCD with fixed-order inverse-log running, the fits give alpha_s(M_Z)=0.1179^{+0.0051}_{-0.0069} (NLO), 0.1221^{+0.0063}_{-0.0080} (NNLO), 0.1313^{+0.0027}_{-0.0067} (N3LO), and about 0.1268 when an estimated N4LO term is included. The growth is attributed to large pi^2 analytic-continuation contributions that make the Minkowski-region coefficients r3=-10.377 and r4=-106.15 negative and sizable, so the fixed-order series","pith_inferences":["A natural extension would be to repeat the fits with contour-improved or analytic perturbation theory, which resum the pi^2 terms; if alpha_s then stabilizes across orders, that would confirm the paper's diagnosis that the growth is a fixed-order artifact.","The reconstructed BESIII covariance matrix is a single point of failure: if the true correlations differ, the chi^2_0 roughly 54/13 conclusion and the decision to drop eight points could both change, and the combined-fit alpha_s values in Table 5 would shift.","The same pi^2 mechanism should affect other time-like observables such as hadronic tau decays; the order-dependence seen here could be compared with tau-decay extractions to test whether the effect is universal.","A dedicated scan of R in the 3.4-3.6 GeV region with reduced systematic correlations would settle whether the eight BESIII points reflect new physics, underestimated uncertainties, or a breakdown of the massless approximation."],"forward_implications":["If the central claim is right, alpha_s(M_Z) determinations from low-energy R data agree with the world average only at NLO (0.1179) and NNLO (0.1221); the N3LO value 0.1313 overshoots it.","The eight BESIII points above the J/psi mass are not describable by massless three-flavor pQCD, so future extractions from that energy region must either exclude them or add missing physics.","The estimated N4LO correction (about 0.1268) partially reverses the N3LO increase, indicating the fixed-order series oscillates before settling.","Fits to the full 14-point BESIII dataset are statistically unacceptable, so combined analyses should use the truncated set.","The order-dependence pattern provides a concrete test for when fixed-order perturbation theory fails in the time-like region."],"fun_headline_variants":["QCD alpha_s rises from 0.118 to 0.131 with perturbative order","KEDR-BESIII fits show alpha_s climbs with QCD loop order","Alpha_s from R-ratio grows: pi^2 terms leak into fixed orders","Higher-order QCD fits push alpha_s(MZ) to 0.131","Order-dependent alpha_s hints at pi^2 term problem in QCD"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that the reconstructed BESIII covariance matrix, built by treating four systematic sources as fully correlated and three as uncorrelated, correctly represents the experiment's true correlations; if it does not, the decision to drop eight BESIII points and the resulting alpha_s values would change.","fun_headline_variants_meta":{"raw":{"variants":["QCD alpha_s rises from 0.118 to 0.131 with perturbative order","KEDR-BESIII fits show alpha_s climbs with QCD loop order","Alpha_s from R-ratio grows: pi^2 terms leak into fixed orders","Higher-order QCD fits push alpha_s(MZ) to 0.131","Order-dependent alpha_s hints at pi^2 term problem in QCD"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000381,"raw_usage":{"total_tokens":1913,"prompt_tokens":853,"completion_tokens":1060,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":597,"completion_tokens_details":{"reasoning_tokens":953}},"tokens_in":597,"tokens_out":1060,"duration_ms":7900,"temperature":1.0,"reasoning_tokens":953,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T16:59:26.893095+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"If the BESIII collaboration released its full correlation matrix and a refit of all 14 points with that matrix gave chi-squared per degree of freedom near 1 and order-stable alpha_s values, the paper's central claims—that the full dataset is incompatible with pQCD and that the order-dependence is driven by pi^2 terms—would be falsified.","supporting_citations":[],"review_version":3}