{"id":"acd7d2ba-caec-4d55-9b55-29a5b2e1cd5b","arxiv_id":"2501.06849","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Isospin-dependent higher-twist corrections remove a systematic bias in large-x DIS global fits that otherwise distorts the n/p ratio and the deuteron off-shell function.","lead":"This paper shows that a seemingly minor modeling choice in global QCD fits, whether higher-twist corrections are added or multiplied and whether they are shared between proton and neutron, changes the extracted neutron-to-proton ratio at large x by about 25%. The authors recommend fitting isospin-dependent higher-twist corrections to remove this systematic bias.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central 'bias' claim is inferred from refitting the same data with different HT ansatze, not from recovery of a known truth; a mock-data closure test is needed to confirm that isospin-dependent HT removes a true bias rather than a basis artifact.","rationale":"The reader's weakest assumption is the Q2-independence of delta_f; I agree this is a real soft spot, but I see an even more load-bearing issue: the 'bias' label itself is not validated against any ground truth. The paper compares two refits of the same experimental data; the difference in n/p and delta_f is consistent with the analytic argument but does not prove which (if either) is biased. This is why the missing mock-data test is the decisive check. The Q2 concern is a specific instance of this: a Q2-dependent delta_f could absorb the 1/Q2 HT difference, changing the inferred bias. The numerical demonstrations are otherwise internally consistent, the framework is the established CJ22 setup, and the Sec. IV cross-checks add value. The recommendation to use isospin-dependent HT is reasonable and likely correct, but as a field-standard prescription it should be backed by a closure test and functional-form/code release. Therefore the reader's CONDITIONAL verdict is appropriate; my read does not move it.","tokens_in":22783,"tokens_out":10982,"duration_ms":116792,"concrete_test":"Closure test in the CJ22 framework: choose a benchmark truth from the isospin-dependent fit of Fig. 4 (PDFs, delta_f_truth(x), Hp(x), Hn(x) or Cp, Cn, at Q2=10 GeV2) and generate pseudo-data for F2p and F2D at the SLAC/NMC/HERMES/BONuS kinematics with the real experimental uncertainties and correlations. Refit with the four prescriptions (iso-indep add/mult, iso-dep add/mult) and compare extracted n/p and delta_f to the benchmark. The central claim predicts iso-indep fits miss the benchmark n/p by more than the tolerance bands at x~0.6-0.8 while iso-dep fits recover it; if this fails, the 'bias' is a basis artifact. Repeat with a benchmark delta_f containing a small 1/Q2 term to test whether the Q2-independence assumption is load-bearing.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim is that isospin-independent additive or multiplicative HT implementations bias the large-x n/p ratio and that the fitted off-shell function 'artificially compensates' this bias. The analytic argument in Sec. IID (Eqs. 13-15) establishes only that for a fixed, identical H or C coefficient the additive and multiplicative forms give different x->1 limits of n/p. But in the fits H and C are free parameters and are fitted to different values; the observed 25% difference in n/p at x=0.8 is therefore a property of the chosen ansatze, not by itself evidence that one of them is biased. The paper's own Sec. IV shows that when W-asymmetry data are excluded the compensation moves partly into the d/u ratio, so the compensation channel is not unique. The clearest explicit vulnerability is the Q2-independence of delta_f (Sec. II.B.2), which the authors themselves call 'somewhat peculiar' in Sec. III.C: over the limited SLAC Q2 range a constant delta_f can mimic a 1/Q2 HT term. If delta_f were allowed a 1/Q2 or logarithmic Q2 dependence, the additive and multiplicative isospin-independent fits might be reconciled without isospin-dependent HT, weakening the claim that the off-shell compensation is artificial. Because the 'bias' is an interpretive label attached to refits of the same experimental data, the decisive missing element is a mock-data closure test with a known input model.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper investigates how the phenomenological implementation of higher-twist (HT) corrections in global QCD analyses of deep-inelastic scattering affects extracted large-x quantities. Working in the CTEQ-JLab (CJ22) framework, the authors compare multiplicative (F2[1+C(x)/Q2]) and additive (F2+H(x)/Q2) HT forms, each with the HT function either constrained to be identical for proton and neutron (isospin-independent) or fitted separately (isospin-dependent). They find that the isospin-independent additive and multiplicative implementations describe the deuteron data with comparable quality but imply substantially different n/p structure function ratios at large x (about 25% at x=0.8, Fig. 1) and oppositely signed off-shell deformation functions δf; the near-identical D/p predictions (Fig. 2) are interpreted as the fitted off-shell function compensating for the HT implementation choice. When isospin dependence is allowed (Fig. 4), the additive and multiplicative fits agree in n/p, d/u and δf, the n/p tail is closer to the multiplicative result as anticipated by the analytic argument of Sec. IID (Eqs. 13-15), and the neutron HT function is found to be roughly half the proton one. The paper further shows (Sec. IV) that without Tevatron W-asymmetry data the compensation can partly shift into the d/u ratio, but the isospin-dependent conclusion persists.","tokens_in":23135,"tokens_out":17630,"duration_ms":164290,"significance":"If the claims hold, the paper has immediate practical consequences for large-x PDF analyses: global fits that silently assume isospin-independent higher-twist corrections carry a large, previously underestimated systematic uncertainty in the n/p ratio and in the extracted deuteron off-shell deformation. The paper's strengths are substantial: a transparent analytic demonstration of the implementation ambiguity (Eqs. 13-15), a controlled comparison in which only the HT implementation is varied within the same framework, explicit treatment of the WBA smearing and off-shell formalism, a concrete recommendation (isospin-dependent HT with the additive/multiplicative spread quoted as uncertainty), and unusually candid statements of the model limitations (Q2-independence of δf, the catch-all nature of the fitted HT term, and the WBA validity range). Inclusion of the full CJ22 data set and the BONuS cross-checks (Figs. 3 and 6) lends weight to the numerical results.","major_comments":[{"comment":"The central claim — that isospin-independent HT implementations are biased and that separate proton and neutron HT functions remove the bias — is inferred from refitting the same experimental data with different ansatze, not from recovery of a known input. The analytic argument in Sec. IID (Eqs. 13-15) fixes the asymptotic n/p behavior for fixed coefficients H and C, and Eqs. (11)-(12) correctly show that an isospin-independent multiplicative term is isospin-dependent when rewritten additively. In the fits, however, C(x) and H(x) are free parameters, so the roughly 25% difference at x=0.8 (Sec. IIIA, Fig. 1) is a property of the fitted ansatze under the data constraints; it does not by itself establish which scenario is closer to the truth. A mock-data closure test — generating pseudodata from a known model with specified input HT functions and δf, then refitting under all four scenarios — would directly establish whether the isospin-dependent fits recover the input n/p, d/u, and δf, and whether the isospin-independent additive and multiplicative fits deviate in the claimed direction. I recommend adding such a test, or explicitly reframing the conclusions as demonstrating a systematic ambiguity with a well-motivated preferred resolution rather than a demonstrated bias.","section":"Secs. II.D, III, and V"},{"comment":"The compensation mechanism underlying the paper assumes δf is independent of Q2 while the HT terms fall as 1/Q2. The assumption is stated in Sec. II.B.2, and Sec. III.C correctly flags it as 'somewhat peculiar' given that a Q2-independent term is used to absorb a power-suppressed bias. Over the limited Q2 range of the SLAC deuteron data, a constant shift is partially degenerate with a 1/Q2 term, so the partition of the correction between δf and the HT functions is ansatz-dependent. If δf were allowed a mild Q2 dependence (e.g., a logarithmic or 1/Q2 term), the additive and multiplicative isospin-independent fits might be reconciled differently, which could weaken the conclusion that isospin-dependent HT is required. The concluding paragraph of Sec. V anticipates this caveat qualitatively, but the central message depends on the point; I would like to see a robustness test in which a Q2-dependent parameter is added to δf and the four scenarios of Secs. IIIA-IIIB are rerun, to determine whether the main conclusions survive.","section":"Sec. II.B.2 and Sec. III.C"},{"comment":"Section IV demonstrates that the compensation channel is not unique. With W-asymmetry data excluded and b fixed to 0.06 (Fig. 8), the multiplicative fit agrees with the additive fit in n/p and δf, but the implementation difference reappears in d/u, at a cost of Δχ2=10 and χ2_W/npt=6; the text further states that a similar reconciliation can be obtained by increasing the multiplicative off-shell function, again pushing the difference into d/u. This shows that the allocation of the HT implementation difference among n/p, δf, and d/u depends on the data set and on the parametrization rather than being determined uniquely by the physics. The paper draws the right conclusion regarding the value of the W-asymmetry data, but the non-uniqueness should be acknowledged when the 'artificial compensation by the off-shell function' is stated in Secs. IIIA and V, and ideally the effect should be quantified as a range over the allowed compensation channels.","section":"Sec. IV, Figs. 7-9"}],"minor_comments":[{"comment":"'in teh global fit' is a typo for 'in the global fit'.","section":"Fig. 2 caption"},{"comment":"'aknowledge' is a typo for 'acknowledge'.","section":"Acknowledgments"},{"comment":"The clause defining δF2D is garbled ('dwefined as i.e.'); it should read 'is defined as'.","section":"Appendix C"},{"comment":"'there there is no need' contains a duplicated word.","section":"Sec. III.B"},{"comment":"For the four main scenarios, the paper quotes no total or per-dataset χ2 values; reporting them (as it does for the Δχ2=10 and χ2_W/npt values in Sec. IV) would let the reader see whether the isospin-dependent fits are preferred by the data or merely equally acceptable.","section":"Secs. III.A-III.B"},{"comment":"The appearance of the factor 9 in the numerator and denominator would be clearer if the text noted that the charge factors (4/9, 1/9) have been cleared by a common factor 9; without this remark, the reader may wonder about the normalization convention for H(x).","section":"Eq. (14)"}],"recommendation":"major_revision","confidential_remarks":"To the editor: this is a well-executed phenomenological study from the CJ collaboration, honest about its limitations and fair in its comparison with the AKP and JAM analyses. The main question is whether the 'bias' claim requires a mock-data closure test to meet the evidentiary standard for publication; in my view the claims modestly outrun the evidence, and the requested test (or a softening of the language) is well within the authors' reach. The paper is squarely within the journal's scope. There is partial overlap with the authors' own conference proceedings (Ref. [69]), but the journal version contains the full analysis and is sufficiently distinct."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear [Colleague],\n\nThe thing to know about this paper is that it makes a persuasive case that the standard assumption of isospin-independent higher-twist corrections biases the large-x n/p ratio and the fitted off-shell function in global DIS fits. Within the CJ22 framework, the additive and multiplicative implementations give n/p ratios that differ by about 25% at x=0.8, and the off-shell function flips sign to compensate. Allowing proton and neutron HT functions to differ brings the two implementations into agreement. That is a real result, and as a controlled comparison inside one fitting framework it is new.\n\nThe paper does several things well. The analytic inequalities in Sec. II.D are simple but they explain why the two implementations behave differently. The numerical work is coherent: the fits describe the D/p and n/D data equally well in all scenarios, which is exactly what makes the point about compensation. The authors are honest about the caveats: they state that the off-shell function is assumed Q2-independent, that the WBA is used only at x<0.8, and that the fitted HT term is a catch-all. The relevant prior literature, including AKP and JAM, is cited and discussed rather than ignored.\n\nThe soft spots are real but not fatal. The word 'bias' is an interpretation, not a demonstrated fact about nature. What the paper actually shows is that the extracted n/p ratio and off-shell function depend strongly on the HT implementation unless one allows isospin-dependent HT terms. That is a solid systematic-uncertainty statement even if you do not call the isospin-independent result biased. The stress-test raises the lack of a mock-data closure test; I think that would strengthen the claim, but its absence does not undermine the main message. The more concrete problems are reproducibility: no code or parameter files are provided, and the functional forms of the fitted HT functions are not given. Those should be fixed. The Q2-independence of delta_f is a genuine vulnerability, and the authors acknowledge it; over the limited SLAC Q2 range a constant off-shell deformation can mimic a 1/Q2 effect. Still, the conclusion that isospin-dependent HT corrections reduce implementation sensitivity is robust enough to stand.\n\nI would send this to peer review. The paper is useful for anyone doing global fits or extracting neutron structure at large x, and the recommendation to report isospin-dependent HT systematics is worth taking seriously. With code and a mock-data test added in revision, it would be a strong reference.","headline":"A careful systematic study showing that isospin-independent higher-twist corrections bias the large-x n/p ratio and off-shell function in global fits; the recommendation to allow isospin-dependent HT is sound, though reproducibility and a mock-data test would firm it up.","tokens_in":23695,"tokens_out":4007,"would_cite":true,"duration_ms":37247,"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":"This paper shows that in global QCD fits the choice of how higher-twist corrections enter proton and neutron structure functions — additive versus multiplicative, and isospin-independent versus isospin-dependent — shifts the extracted…","keywords":["higher-twist corrections","deep inelastic scattering","neutron-to-proton structure function ratio","parton distribution functions","off-shell nucleon deformation","deuteron structure function","global QCD analysis","large-x physics"],"falsifier":"Re-run the isospin-independent additive and multiplicative fits on deuteron data with substantially wider $Q^2$ coverage at large $x$ — the paper identifies such data as the experimental route forward. Since the higher-twist terms fall as $1/Q^2$ while the off-shell deformation is assumed constant, a dataset in which $1/Q^2$ changes by a factor of two or more within the same $x$ bins should either make the fitted off-shell function visibly $Q^2$-dependent, or force the additive and multiplicative fits to disagree with the data; either outcome would refute the compensation mechanism as described.","tokens_in":22547,"feed_emoji":"⚛️","tokens_out":20316,"duration_ms":162980,"temperature":0.7,"pith_summary":"The paper sets out to show that one of the silent choices in global QCD analyses — how the $1/Q^2$ \"higher-twist\" corrections to deep-inelastic structure functions are parametrized — is not neutral at large Bjorken $x$. Working in the CJ22 global-fit framework used and developed by the authors, it demonstrates that an isospin-independent additive correction inflates the neutron-to-proton structure function ratio by about 25% at $x = 0.8$ compared with an isospin-independent multiplicative correction, purely as an artefact of the functional form. The fits then disguise the discrepancy by deforming a second fitted ingredient, the off-shell function describing nucleons bound inside the deuteron, so that both implementations reproduce the data equally well. When the proton and neutron higher-twist corrections are fitted independently, the additive and multiplicative choices converge to compatible $n/p$ ratios and off-shell functions, and the residual spread provides a usable systematic uncertainty. The result matters because the large-$x$ $n/p$ ratio is a window onto quark confinement and the $d/u$ ratio, and because deuteron data are the main route to the neutron structure function.","feed_headline":"Higher-twist choice biases the neutron/proton ratio by 25 percent","feed_subtitle":"Global fits that force proton and neutron corrections to match let fitted nuclear effects hide a 25 percent shift in n/p.","key_machinery":"The argument runs on three pieces. First, the $x \\to 1$ asymptotics of the $n/p$ ratio: an isospin-independent multiplicative higher-twist term $C(x)$ cancels in the ratio, while an isospin-independent additive term $H(x)$ leaves a $1/Q^2$ tail proportional to $H/(u Q^2)$, which is what inflates the ratio. Second, the equivalence formula $\\tilde H(x,Q^2) = F_2(x,Q^2)\\,C(x)$: an isospin-independent multiplicative correction becomes isospin-dependent when recast additively, and vice versa, so the common \"isospin-independent\" implementation is not self-consistent. Third, the off-shell expansion $\\delta_f(x)$ of nucleons bound in the deuteron, fitted to data through the weak-binding-approximation smearing formula: because the off-shell function is constrained in practice only for $x \\lesssim 0.7$, it has the freedom to compensate the higher-twist implementation bias in the deuteron and hide the discrepancy from the data.","core_discovery":"Stated as the authors would state it: the large-$x$ behavior of the neutron-to-proton structure function ratio is partly an artefact of how higher-twist corrections are implemented, not a pure measurement. For an isospin-independent multiplicative correction, the factor $(1 + C/Q^2)$ cancels in the ratio and $n/p$ approaches the leading-twist value $1/4$ as $x \\to 1$. For an isospin-independent additive correction, the ratio picks up a term proportional to $H/(u Q^2)$, and the fitted $n/p$ rises noticeably — about 25% higher at $x = 0.8$. Because the neutron structure function reaches global fits almost entirely through deuteron data, the fit compensates for this artificial rise by shifting the fitted off-shell deformation $\\delta_f$: positive in the additive case, negative in the multiplicative case, with both fits describing the deuteron-to-proton ratio equally well. Relaxing the isospin-independence constraint lets the proton and neutron higher-twist functions absorb the isospin-dependent $1/Q^2$ effects; the additive and multiplicative fits then yield compatible $n/p$ ratios and compatible off-shell functions, and the difference between them becomes a bounded, quantifiable systematic uncertainty.","pith_inferences":["The bias mechanism is generic rather than framework-specific: any fit that forces proton and neutron $1/Q^2$ corrections to be identical and chooses the additive form inherits the same artificial $n/p$ tail, because the argument turns only on the structure-function ratio.","If the off-shell deformation carries intrinsic $Q^2$ dependence — which the paper itself notes is peculiar to exclude — then both the extracted $\\delta_f$ and the higher-twist functions would need revisiting; a joint fit with $Q^2$-dependent off-shell terms is a direct testable extension.","The paper's effective off-shell function for the deuteron structure function, defined in its Appendix C, is a portable quantity that other fitting groups could compute, making it a natural cross-framework benchmark for nuclear corrections.","The same isospin-blind treatment could bias flavor-dependent off-shell extractions from tritium and helium-3 data, so future analyses of those targets should adopt isospin-dependent higher-twist terms from the outset."],"forward_implications":["Extracting the neutron-to-proton ratio without reporting the higher-twist implementation is incomplete: isospin-independent fits can shift the ratio by about 25% at $x = 0.8$ depending on the additive or multiplicative form.","Fitting proton and neutron higher-twist functions separately removes the compensation mechanism, and the residual additive-versus-multiplicative spread becomes the systematic uncertainty to quote.","The fitted off-shell deformation of deuteron-bound nucleons is currently correlated with the higher-twist implementation, so its large-$x$ values should not be read as a physical nucleon deformation until the isospin-dependent treatment is standard.","The $d/u$ PDF ratio stays protected from the bias only when free-proton data such as the $W$-boson rapidity asymmetry from proton-antiproton collisions pin it down; without them the bias migrates into $d/u$.","Deuteron data with wider $Q^2$ reach or higher statistical power at large $x$ are required to separate the $Q^2$-independent off-shell effect from the $1/Q^2$ higher-twist terms."],"supporting_citations":[{"why":"Supplies the CJ22 global-fit framework, PDF parametrization, datasets, and kinematic cuts on which all fits in this paper are built.","marker":"[20]"},{"why":"Established the earlier CJ framework's treatment of target-mass corrections and fitted power-correction terms that the present study varies.","marker":"[17]"},{"why":"Provides the weak-binding-approximation smearing functions and the off-shell nucleon expansion used to model the deuteron target.","marker":"[46]"},{"why":"The fixed-target deuteron deep-inelastic data whose large-x, narrow-energy-range coverage lets the off-shell function absorb the implementation bias.","marker":"[74]"},{"why":"W-boson rapidity asymmetry data from proton-antiproton collisions that pin down the free-nucleon d/u ratio and push the bias into the off-shell function.","marker":"[70]"},{"why":"An earlier additive-fit study of nucleon off-shell corrections whose isospin-independent results the paper contrasts with its own to isolate the implementation effect.","marker":"[42]"},{"why":"Companion study showing d/u sensitivity to the power-correction choice, cited to argue that without W-asymmetry constraints the bias migrates into d/u.","marker":"[43]"},{"why":"A flavor-dependent decomposition of the off-shell deformation, cited to explain why a non-zero u-quark deformation could cancel against the d-quark one.","marker":"[39]"}],"fun_headline_variants":["Neutron/proton ratio bias traced to higher-twist treatment","Additive vs multiplicative twist: 25% shift in n/p","Isospin-dependent twist corrects n/p extraction bias","Twist handling, not data, drives large-x n/p ratio","Systematic twist uncertainty in neutron structure function"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that the off-shell deformation $\\delta_f$ of nucleons bound in the deuteron is independent of $Q^2$, even though the higher-twist corrections it is found to compensate fall as $1/Q^2$; the paper states this in Section II.B.2 and flags it as \"somewhat peculiar\" in Section III.C, so if $\\delta_f$ actually varies with $Q^2$, the compensation pattern and the inferred bias magnitude could change.","fun_headline_variants_meta":{"raw":{"variants":["Neutron/proton ratio bias traced to higher-twist treatment","Additive vs multiplicative twist: 25% shift in n/p","Isospin-dependent twist corrects n/p extraction bias","Twist handling, not data, drives large-x n/p ratio","Systematic twist uncertainty in neutron structure function"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000181,"raw_usage":{"total_tokens":1291,"prompt_tokens":912,"completion_tokens":379,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":528,"completion_tokens_details":{"reasoning_tokens":295}},"tokens_in":528,"tokens_out":379,"duration_ms":3348,"temperature":1.0,"reasoning_tokens":295,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T20:50:17.122947+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-run the isospin-independent additive and multiplicative fits on deuteron data with substantially wider $Q^2$ coverage at large $x$ — the paper identifies such data as the experimental route forward. Since the higher-twist terms fall as $1/Q^2$ while the off-shell deformation is assumed constant, a dataset in which $1/Q^2$ changes by a factor of two or more within the same $x$ bins should either make the fitted off-shell function visibly $Q^2$-dependent, or force the additive and multiplicative fits to disagree with the data; either outcome would refute the compensation mechanism as described.","supporting_citations":[{"cited_title":"Bacchetta, M","cited_arxiv_id":null,"evidence_quote":"Supplies the CJ22 global-fit framework, PDF parametrization, datasets, and kinematic cuts on which all fits in this paper are built."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the weak-binding-approximation smearing functions and the off-shell nucleon expansion used to model the deuteron target."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The fixed-target deuteron deep-inelastic data whose large-x, narrow-energy-range coverage lets the off-shell function absorb the implementation bias."},{"cited_title":"Alekhin, J","cited_arxiv_id":null,"evidence_quote":"An earlier additive-fit study of nucleon off-shell corrections whose isospin-independent results the paper contrasts with its own to isolate the implementation effect."},{"cited_title":"Cocuzza, C","cited_arxiv_id":null,"evidence_quote":"Companion study showing d/u sensitivity to the power-correction choice, cited to argue that without W-asymmetry constraints the bias migrates into d/u."},{"cited_title":"Virchaux and A","cited_arxiv_id":null,"evidence_quote":"A flavor-dependent decomposition of the off-shell deformation, cited to explain why a non-zero u-quark deformation could cancel against the d-quark one."}],"review_version":1}