{"id":"db7a2851-13ff-47fc-aeae-f98a8dca1b93","arxiv_id":"1909.02760","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"The NNLO QCD corrections to DIS event shape distributions and their mean values are computed for the first time, reducing scale uncertainties and improving the description of H1 and ZEUS data.","lead":"This paper computes the first next-to-next-to-leading-order (NNLO) quantum chromodynamics corrections to event shape distributions in deep inelastic scattering. The new corrections shrink theory uncertainties and improve agreement with HERA data, enabling more precise studies of the strong nuclear force and the proton's quark-gluon structure.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Agreement with data depends on a strongly Q-dependent shift from the dispersive model; the fixed-order NNLO result is well supported, but the 'improved description' claim is most entangled with the constant-shift approximation at low Q2.","rationale":"The reader's weakest assumption is exactly the constant-shift hadronization approximation. I agree that this is the main weak point, but the correct stress-test response is not to lower the verdict: the fixed-order NNLO result is the paper's primary deliverable and is well supported by independent ingredients. The data-comparison claim is explicitly worded as 'in general' and is corroborated by multiple figures at medium/high Q2. The paper openly states its limitation, which is a sign of honesty, not a hidden flaw. No internal inconsistency exists. The reproducible-code absence limits precision but not correctness of the perturbation theory. If exactly one check had to settle the concern, the F-dependent shift test would be the decisive one. Since the concern is real but does not overturn the central claim, UNCHANGED is appropriate.","tokens_in":21854,"tokens_out":1615,"duration_ms":15805,"concrete_test":"For a representative low-Q2 bin (e.g., H1 Q = 14-16 GeV) and a representative medium-Q2 bin (e.g., ZEUS Q2 = 320-640 GeV), compute the data-theory residuals for tau_gamma and C. Use two alternative treatments: (a) the constant shift of Eq. 15, and (b) the same shift with F-dependent power correction, e.g., P(F) obtained from Monte Carlo hadronization or from a dispersion relation with running-coupling moment, or a two-parameter shift. If the shape of the residuals changes by more than the NNLO scale uncertainty in either bin, the improvement claim is not robust to the constant-shift approximation. If the residuals are stable, the claim survives.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim has two parts: (1) NNLO fixed-order corrections to DIS event shapes are computed, and (2) these corrections improve the description of H1/ZEUS data once dispersive-model power corrections are included. Part (1) is structurally sound: it reuses the NNLOJET antenna-subtraction machinery already validated for DIS dijets [26,27] and N3LO single jet [45], and the paper openly exposes instabilities at Sudakov shoulders and kinematical ridges. The load-bearing weakness is in part (2). The constant-shift approximation, applied as dσ_hadron(F) = dσ_parton(F - a_F P)/dF (Eq. 15), is used to compare with hadron-level data, and the authors themselves state (Section 4) that a constant shift is only an approximation. In the C-parameter and tau_T distributions, the shift is more than two bins large at low Q2, moving predictions across Sudakov shoulders and ridges (Section 5.1). There is no validation that a single a_F P reproduces the true hadronization for all F. However, the paper's strongest-claim wording, 'inclusion of NNLO corrections leads in general to an improved description,' is not singularly dependent on the model: the improvement is most clearly supported at medium/high Q2, where P is small and the constant-shift assumption is least stressed, and the parton-level scale uncertainties shrink regardless of hadronization. Still, for the distributions, the claim of 'improved description' is entangled with the model through a cancellation between positive parton-level NNLO contributions and reduced NNLO power corrections; without releasing grids or defining a quantitative fit criterion (e.g., chi2), the claim is qualitative and cannot be isolated from model error.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents the first next-to-next-to-leading-order (NNLO) QCD calculation of event shape distributions and their mean values in deep inelastic scattering. The calculation is implemented in NNLOJET by adapting the existing NNLO dijet-in-DIS calculation; the paper documents the phase-space cuts, scale choice, PDF set, and the technical low-F cut-offs. The fixed-order results are compared with H1 and ZEUS data after applying dispersive-model hadronization corrections as a constant shift (Eq. 15). The authors find that NNLO corrections reduce scale uncertainties and generally improve the description of the data, while explicitly identifying regions where fixed-order predictions are unreliable (small F, Sudakov shoulders, kinematical ridges).","tokens_in":22154,"tokens_out":5967,"duration_ms":64598,"significance":"This is a substantial step forward: DIS event shapes were previously known only to NLO, and the scale uncertainty on NLO predictions limited precision studies; the present work makes NNLO predictions available for all five standard DIS event shapes and their mean values. The fixed-order part is parameter-free once PDFs and alpha_s are fixed, and it reuses machinery already validated in DIS dijet and single-jet calculations. The authors are commendably explicit about the known limitations of the constant-shift dispersive model and about the instability regions. If the results hold, they also open the door to a consistent NNLO extraction of alpha_s and alpha0 from HERA data.","major_comments":[],"minor_comments":[{"comment":"The sentence 'inclusion of the NNLO corrections leads in general to an improved description' should be qualified to specify the regions where the statement is robust (medium/high Q2, away from the first bin and from Sudakov shoulders), because at low Q2 the comparison relies on the constant-shift approximation and the authors themselves note large shifts for the C-parameter.","section":"Abstract and Section 6"},{"comment":"The text states that the normalization cross sections are computed to NNLO for all predictions; this means the NLO and NNLO distribution predictions are normalized by the same NNLO total cross section. This choice should be stated explicitly in Section 3 so that the NLO curves are not misinterpreted as normalized to an NLO total cross section.","section":"Section 3"},{"comment":"The statement that the smallness of the NNLO correction to tau_T is due to a cancellation between positive parton-level corrections and decreased power corrections would be more compelling with a quantitative illustration; consider giving P_NNLO/P_NLO values for a few Q bins or plotting the separate contributions.","section":"Section 5.1 and Eq. (15)"},{"comment":"Axis labels in Figures 2 and 3 are inconsistent or missing for some rows (for example, the label 'T' leaves ambiguity between tau_T and B_T); please label each panel explicitly with the event shape variable.","section":"Figures 2 and 3"},{"comment":"The definition of the 'left-most non-vanishing bin' after the shift depends on the cut values in Eq. (13); please specify exactly which experimental bins are excluded in the comparison, for example by listing the bin indices, to make the comparisons reproducible.","section":"Section 3.1 and Section 5.1"}],"recommendation":"minor_revision","confidential_remarks":"This is a solid fixed-order calculation with honest caveats. The only point I would keep in mind for editorial purposes is that the 'improved description of data' is not a sharp test of the NNLO corrections because the dispersive-model power correction is not fitted in this paper but taken from previous analyses including HERA data; I do not consider this a blocker, but the authors should not present the data comparison as a precision test of the NNLO terms alone. The manuscript is a good fit for the journal's scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis paper does something genuinely new: it computes NNLO QCD corrections to the full set of DIS event shape distributions and their mean values. e+e- event shapes have had NNLO for over a decade; DIS was stuck at NLO. The result closes that gap, and for the first time brings HERA event shape data into the regime where scale uncertainties are below experimental errors at moderate and high Q2. The technical setup is an extension of the group's NNLOJET dijet calculation, so it is not a new framework, but the adaptation is non-trivial: the authors document the phase-space cuts, the fine-binned histograms, the handling of the F to 0 region, and the seven-point scale variation. The fixed-order predictions show reasonable convergence, and the instabilities near Sudakov shoulders and kinematical ridges are flagged explicitly rather than hidden.\n\nThe soft spots are real but not fatal. First, the hadron-level comparison relies on the dispersive model with a constant shift in F. The authors themselves call this an approximation, and at low Q2 the shift is more than two bins wide in C and tau_T—big enough to move predictions across the Sudakov shoulder. The 'improved description' claim is therefore most reliable at medium to high Q2, and should be read as qualitative, since no chi2 is given. I don't think this undermines the main result: the NNLO fixed-order corrections are parameter-free given PDFs and alpha_s, and the parton-level scale uncertainty shrinks regardless of the hadronization model. The comparison also uses alpha0 = 0.5 taken from fits to the same class of data, so the agreement with data is not an independent test of the model—but the paper does not oversell it.\n\nSecond, no grids or code are released, so other groups cannot reproduce the numerical results easily. That is a minor point for a paper of this type; the method is established and the authors state they plan convolution grids for a future alpha_s extraction.\n\nA note on the citation pattern: heavy self-citation, but the cited results are the actual machinery being used. That is appropriate.\n\nOverall: technically sound, clearly written, and honest about its limitations. The main value is the NNLO fixed-order phenomenology, not the hadronization comparison.\n\nI would send this to a serious referee. The report should push for a quantitative goodness-of-fit statement and a clearer separation of the perturbative improvement from the model-dependent shift, but the calculation itself is exactly what the field needs.","headline":"First NNLO DIS event shapes: solid fixed-order result, model-dependent data comparison, worth refereeing.","tokens_in":22769,"tokens_out":1945,"would_cite":true,"duration_ms":18740,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Second-order QCD corrections now cover DIS event shapes","keywords":["QCD","NNLO corrections","event shapes","deep inelastic scattering","power corrections","dispersive model","HERA","jet physics"],"falsifier":"Refit $\\alpha_0$ in each $Q^2$ bin, or in separate ranges of $F$, using the NNLO predictions; the constant-shift dispersive model predicts a single universal $\\alpha_0$ with no systematic trend, so a trend of the fitted $\\alpha_0$ with $Q^2$ or with the event shape variable that exceeds the NNLO scale uncertainty would rule out the hadronization treatment.","tokens_in":21627,"feed_emoji":"⚛️","tokens_out":5996,"duration_ms":60179,"temperature":0.7,"pith_summary":"This paper establishes that event shape distributions and their mean values in deep inelastic lepton-proton scattering can now be predicted at next-to-next-to-leading order (NNLO) in perturbative QCD. The corrections differ in size and shape among the thrust, jet mass, broadening, and C-parameter variables, and they typically shrink the renormalization and factorization scale uncertainty from about ten percent at NLO to about five percent, or a few percent at high $Q^2$. Supplemented with dispersive-model power corrections to account for hadronization, the NNLO predictions describe the HERA data better than NLO, especially the shape of the distributions. This matters because it removes the main theory limitation on extracting the strong coupling constant and the non-perturbative parameter $\\alpha_0$ from HERA event shape data.","feed_headline":"Second-order QCD corrections now cover DIS event shapes","feed_subtitle":"Scale uncertainty drops to a few percent, and HERA data are better described than at NLO.","key_machinery":"The calculation is built by extending the NNLO corrections to di-jet production in deep inelastic scattering, replacing the jet algorithm with computations of the event shape variables. It combines four-parton tree amplitudes, three-parton one-loop amplitudes, and two-parton two-loop amplitudes using the antenna subtraction method, which isolates and cancels infrared singularities numerically, inside a parton-level Monte Carlo program. The hadronization treatment is the dispersive model, which represents the leading power correction as a universal quantity $P$ times a variable-dependent coefficient $a_F$, shifting the distribution $d\\sigma/dF$ to $d\\sigma/d(F-a_F P)$ and adding $a_F P$ to mean values; $P$ is expanded to $\\alpha_s^3$ at NNLO. The paper also identifies the kinematical ridges and Sudakov shoulders in $C$ and $\\tau_T$ that destabilize fixed-order predictions at exceptional values.","core_discovery":"The paper's central claim is that the complete NNLO QCD corrections to the current-hemisphere event shape distributions ($\\tau_\\gamma$, $\\tau_T$, $\\rho$, $B_\\gamma$, $C$) and their mean values in deep inelastic scattering have been computed and are phenomenologically relevant. In the bulk of the distributions the NNLO corrections are positive, up to about 20% at low and moderate $Q^2$, while they become small or negative at high $Q^2$ and near the upper kinematical boundaries; the NNLO/NLO ratio has a non-trivial shape for every variable. The scale uncertainty is reduced from roughly 10% at NLO to roughly 5% at NNLO, and to below 4% at high $Q^2$, so that the theory uncertainty falls below the experimental errors for moderate and high $Q^2$. When hadronization is modelled by a dispersive power correction that shifts each distribution, the NNLO predictions improve the description of the data. For the mean values, the positive NNLO corrections to the fixed-order result are largely compensated by negative NNLO contributions to the power correction, leaving a small net shift but a substantially smaller scale uncertainty.","pith_inferences":["If the constant-shift hadronization approximation is inadequate, the NNLO improvement may be specific to the bulk of each distribution; fits of $\\alpha_0$ in separate $Q^2$ bins or $F$ regions would reveal whether the effective shift varies, which the paper leaves open.","The compensation between fixed-order and power-correction NNLO terms in the mean values implies that the value of $\\alpha_0$ extracted from data may shift when the fit is upgraded from NLO to NNLO, as was seen in $e^+e^-$ event shape moments; this shift can be tested by repeating the combined fit to the existing HERA data.","The same calculation can be recast into convolution grids for fast re-evaluation, which would enable a practical $\\alpha_s(M_Z)$ extraction from the existing HERA data at NNLO."],"forward_implications":["DIS event shape distributions and mean values are now available at NNLO, with scale uncertainties of a few percent at moderate and high $Q^2$, typically below the experimental errors of the HERA data.","The improved theory description of the shape of the distributions, particularly at large jet mass and in the broadening variable, strengthens the case for using these observables in precision QCD studies.","A fully consistent NNLO-based combined fit of $\\alpha_s(M_Z)$ and $\\alpha_0$ to event shapes is now possible; the theory uncertainty on $\\alpha_s$, previously about 5% from NLO scale variation, should be substantially reduced.","The first bin near $F \\to 0$ remains unreliable without resummation, and Sudakov shoulders and kinematical ridges in $C$ and $\\tau_T$ require new resummation approaches before fixed-order predictions can be used there.","High-resolution measurements at future lepton-hadron colliders would be able to resolve the ridge and shoulder structures and test the need for the new resummations."],"supporting_citations":[{"why":"Supplies the H1 event shape distribution and mean-value data and the $Q$ bins used for comparison.","marker":"[23]"},{"why":"Supplies the ZEUS event shape data and the $(Q^2,x)$ bins used for comparison.","marker":"[24]"},{"why":"Describes the parton-level Monte Carlo program in which the NNLO calculation is implemented.","marker":"[25]"},{"why":"Provides the NNLO corrections to di-jet production in DIS that are extended to event shapes.","marker":"[26, 27]"},{"why":"Defines the dispersive model that supplies the non-perturbative power corrections.","marker":"[28]"},{"why":"Works out the dispersive model for DIS event shapes and provides the coefficients $a_F$.","marker":"[29]"},{"why":"Supplies the Milan factor normalization used in the universal power correction $P$.","marker":"[57]"},{"why":"Provides earlier DIS event shape moment fits that contribute to the adopted value $\\alpha_0=0.5$.","marker":"[58]"},{"why":"Provides the NLL resummation for DIS event shapes, defining the low-$F$ region where fixed-order predictions break down.","marker":"[22]"},{"why":"Identifies the kinematical ridges in DIS event shape phase space that destabilize fixed-order predictions.","marker":"[21]"}],"fun_headline_variants":["NNLO QCD corrections halve scale uncertainty in DIS event shapes","Complete NNLO QCD corrections improve DIS event shape predictions","DIS event shapes: NNLO scale error drops to 5% and below","NNLO QCD improves DIS event shape agreement with HERA data"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The hadron-level comparison assumes that hadronization can be represented by one constant shift per event shape distribution, with $\\alpha_0=0.5$ taken from earlier fits; if the true shift depends on the value of the event shape variable, the claimed improvement in the data description does not follow.","fun_headline_variants_meta":{"raw":{"variants":["NNLO QCD corrections halve scale uncertainty in DIS event shapes","Complete NNLO QCD corrections improve DIS event shape predictions","DIS event shapes: NNLO scale error drops to 5% and below","NNLO QCD improves DIS event shape agreement with HERA data"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000745,"raw_usage":{"total_tokens":3286,"prompt_tokens":873,"completion_tokens":2413,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":489,"completion_tokens_details":{"reasoning_tokens":2338}},"tokens_in":489,"tokens_out":2413,"duration_ms":18553,"temperature":1.0,"reasoning_tokens":2338,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T04:39:16.168084+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Refit $\\alpha_0$ in each $Q^2$ bin, or in separate ranges of $F$, using the NNLO predictions; the constant-shift dispersive model predicts a single universal $\\alpha_0$ with no systematic trend, so a trend of the fitted $\\alpha_0$ with $Q^2$ or with the event shape variable that exceeds the NNLO scale uncertainty would rule out the hadronization treatment.","supporting_citations":[{"cited_title":"Measurement of Event Shape Variables in Deep-Inelastic Scattering at HERA","cited_arxiv_id":"hep-ex/0512014","evidence_quote":"Supplies the H1 event shape distribution and mean-value data and the $Q$ bins used for comparison."},{"cited_title":"Event shapes in deep inelastic scattering at HERA","cited_arxiv_id":"hep-ex/0604032","evidence_quote":"Supplies the ZEUS event shape data and the $(Q^2,x)$ bins used for comparison."},{"cited_title":"Power Corrections to Event Shapes in Deep Inelastic Scattering","cited_arxiv_id":"hep-ph/9704297","evidence_quote":"Works out the dispersive model for DIS event shapes and provides the coefficients $a_F$."},{"cited_title":"Investigation of Power Corrections to Event Shape Variables measured in Deep-Inelastic Scattering","cited_arxiv_id":"hep-ex/9912052","evidence_quote":"Provides earlier DIS event shape moment fits that contribute to the adopted value $\\alpha_0=0.5$."}],"review_version":1}