{"id":"e251da3d-8b81-4015-9a06-03769e0bc38e","arxiv_id":"2411.18976","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"New STAR double-spin asymmetry data for dijets at intermediate pseudorapidity favor positive gluon polarization and disfavor the JAM22 negative solution by 3.5 sigma.","lead":"STAR measures the proton spin asymmetry for dijet production in polarized proton collisions at 200 GeV, focusing on jets at intermediate forward angles. The new data agree with earlier results, improve precision, and rule out a contested negative gluon polarization solution at the 3.5 sigma level.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 3.5σ disfavor of the JAM22 negative gluon polarization solution rests on an unjustified chi²-to-significance conversion; the quoted chi² values alone support only about 2.1σ if evaluated with 20 degrees of freedom.","rationale":"The paper's central quantitative claim is the 3.5σ disfavor of the JAM22 negative gluon polarization solution. The reader's weakest assumption concerned simulation and MLP correction fidelity, but the most load-bearing issue is the statistical conversion: the quoted chi² values can be checked from the text alone, and the apparent conversion sqrt(32.7 − 20.4) = 3.5 treats the difference of two average chi² values as a one-parameter delta-chi². This is not automatically wrong, but it requires justification that the comparison is equivalent to a one-parameter nested test, and the paper provides none. The alternative goodness-of-fit interpretation, chi² = 32.7 for 20 dof, yields a much weaker significance of about 2.1σ, so the headline claim is highly sensitive to the unstated procedure. This is a correctness risk rather than a claim about the experiment's conduct. A revised significance with explicit per-replica distributions, effective dof, and a stated p-value conversion would settle the issue and would not require new experimental data. Therefore the appropriate verdict is CONDITIONAL: accept the measurement as a valuable data release, but require the authors to justify or correct the 3.5σ claim.","tokens_in":30517,"tokens_out":4877,"duration_ms":49413,"concrete_test":"Reproduce the comparison from the published data points and correlation matrices: compute chi² for every JAM22 positive and negative replica using Eq. (7), then report (i) the full distribution of chi²_pos and chi²_neg, (ii) the distribution of Δchi² = chi²_pos − chi²_neg, and (iii) the goodness-of-fit p-value for chi²_neg = 32.7 with 20 dof. Also run a toy Monte Carlo with 20 bins and the stated uncertainties to obtain the null distribution of Δchi². If p(chi²_20 > 32.7) ≈ 0.038, or if the per-replica Δchi² distribution overlaps zero substantially, the 3.5σ statement must be revised downward or replaced by an explicit likelihood ratio.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Section VII.A reports average chi² values of 20.4 (101 positive replicas) and 32.7 (91 negative replicas) for 20 degrees of freedom, and then states that the negative gluon polarization fit is disfavored at the 3.5σ level. No conversion is given. The numbers are numerically consistent with sqrt(32.7 − 20.4) = 3.5, i.e., treating the difference of the average chi² values as a one-degree-of-freedom delta-chi². That is not a valid hypothesis test for two non-nested global-fit solutions with fixed predictions. If each solution is instead assessed by its goodness of fit to 20 data points, chi² = 32.7 gives a p-value of about 0.038 (roughly 2.1σ), while chi² = 20.4 is perfectly acceptable. The difference-of-averages also ignores the width of the replica ensembles: a subset of negative replicas may fit the data as well as the positive ensemble, and averaging chi² over replicas does not by itself define a significance. The chi² formula in Eq. (7) includes three nuisance parameters that reduce the effective number of degrees of freedom, but the paper does not state the effective dof or the resulting p-value. Until the per-replica chi² distributions and the likelihood-ratio or p-value derivation are provided, the headline 3.5σ claim is unsupported.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper reports STAR measurements of the longitudinal double-spin asymmetry A_LL for dijet production with at least one jet at intermediate pseudorapidity (0.8 < eta_jet < 1.8) in polarized proton-proton collisions at sqrt(s) = 200 GeV using the 2015 data set (52 pb^-1). Dijets are reconstructed with the anti-kT algorithm (R = 0.6), and dedicated corrections are applied for tracking inefficiencies (ML-based pT and mass regressions), underlying event, trigger bias, and reconstruction bias. The final A_LL values are presented as a function of parton-level dijet invariant mass for three topologies and are compared with the earlier 2009 STAR measurement and with NLO pQCD predictions using DSSV2014, NNPDFpol1.1, and JAM22 positive/negative gluon polarization sets. The data agree with the 2009 results and with positive-gluon predictions, while the JAM22 negative-gluon solution is reported to be disfavored at the 3.5 sigma level.","tokens_in":30802,"tokens_out":4590,"duration_ms":44828,"significance":"If the measurement and its interpretation are sound, this result provides a new, more precise constraint on the gluon helicity distribution in the x range from about 0.01 to 0.5, specifically through dijet topologies that are sensitive to low-x gluons. The analysis is thorough in its treatment of detector effects: the trigger thresholds, machine-learning jet corrections, underlying event subtraction, and reconstruction bias are each addressed with dedicated systematic studies, and the agreement with the earlier 2009 data supports the consistency of the experimental procedure. The paper also provides correlation matrices that will be useful for future global QCD analyses. However, the headline claim that the JAM22 negative-gluon solution is disfavored at 3.5 sigma is not supported by the statistical evidence presented in the manuscript, and this claim is central to the abstract and conclusions.","major_comments":[{"comment":"The comparison with the JAM22 solutions is performed with the data and theory both at parton level, but the particle-to-parton mass shift used to place the data points relies on the PYTHIA Perugia 2012 tune and the same simulation used to train the ML corrections (Sec. VI.A). The systematic studies in Sec. VI.C.2 cover a range of alternative tunes, which is good, but the manuscript does not state whether the spread of the particle-to-parton mass shift across the alternative tunes is included in the x-axis systematic uncertainties used in the chi2 test of Eq. (7). If this shift is correlated with the theory curve, the beta2 term in Eq. (7) may not capture the full uncertainty. Please clarify that the tune-dependent mass shifts are indeed propagated into the beta2 (dijet energy scale) uncertainty, or provide the corresponding additional uncertainty.","section":"VII.A, Eq. (7)"}],"minor_comments":[{"comment":"The printed formula for A_LL in Eq. (3) appears to be missing the division sign: as typeset, \"ALL = P(PY PB)(N ++ - rN +-)P(PY PB)2(N ++ + rN +-)\" is not a valid expression. The correct form should be the ratio (N++ - rN+-)/(N++ + rN+-) multiplied by 1/(P_Y P_B).","section":"Eq. (3)"},{"comment":"The abstract and summary state that the negative gluon polarization solution is \"strongly disfavored,\" while Sec. VII.A reports a 3.5 sigma disfavor. Once the significance is properly computed, the wording should be aligned with the quantitative result.","section":"Abstract and Sec. VIII"},{"comment":"The pseudorapidity bounds for endcap jets are given inconsistently: the text uses 0.8 < eta_jet < 1.8 in some places and 0.8 <= eta_jet <= 1.8 in others. Please use a consistent convention.","section":"Sec. IV.B"},{"comment":"The figure labels \"Jet + Jet + X -> p + p\" should read \"p + p -> Jet + Jet + X\" to reflect the physical process.","section":"Fig. 2 and Fig. 3 labels"},{"comment":"The statement that the 2009 and 2015 data are \"combined by average weighting\" is vague. Please specify the weights and how the correlated systematic uncertainties (including the relative luminosity and polarization scale uncertainties) are handled in the combination, since these enter the chi2 test of Eq. (7).","section":"Sec. VII.A"}],"recommendation":"major_revision","confidential_remarks":"The experimental measurement and its systematic treatment appear careful and credible, and the agreement with the earlier 2009 dataset is reassuring. The main reason for major revision is the unsupported 3.5 sigma claim; this is a load-bearing statement in the abstract and conclusions. I would encourage the authors to either provide a rigorous statistical derivation of the significance (e.g., per-replica likelihood ratio or a properly defined p-value with effective dof) or to soften the claim to a level consistent with the quoted chi2 values. The remaining technical details (ML corrections, trigger bias, tune variations) are addressed with appropriate systematic studies."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: the measurement is real and worth engaging with; the statistical claim attached to it is not yet supported.\n\nWhat's new: 2015 data at 52 pb^-1, twice the 2009 luminosity; extends dijets with at least one endcap jet to intermediate pseudorapidity; first use of MLP corrections for jet pT and mass in this analysis; direct quantitative comparison to JAM22 positive and negative gluon solutions. The analysis is careful: trigger, underlying event, tracking, ML corrections, and mass shifts each get dedicated systematic studies, and the 2015 points agree with the 2009 data.\n\nThe soft spots are concentrated in the JAM22 comparison. The paper quotes average chi2 = 20.4 for 101 positive replicas and 32.7 for 91 negative replicas with 20 degrees of freedom, then says the negative solution is disfavored at 3.5 sigma. No conversion is shown. The only way I can reproduce 3.5 is sqrt(32.7 - 20.4), which treats the difference of average chi2 values as a chi2 difference at one degree of freedom. That is not a valid test between two non-nested global fits. If you instead take chi2 = 32.7 with 20 dof as a goodness-of-fit, the p-value is about 0.038, roughly 2.1 sigma; chi2 = 20.4 is perfectly acceptable. The per-replica chi2 distributions and the actual likelihood-ratio or p-value derivation are missing. The effective number of degrees of freedom after the three nuisance parameters in Eq. 7 is not stated. So the 3.5 sigma headline is unsupported as written. This is a load-bearing flaw in the paper's central claim, though not in the measurement itself.\n\nAlso worth noting: comparing against a positivity-violating solution is inherently awkward. The JAM22 negative-gluon solution gives negative cross sections for gluon-gluon scattering at high dijet mass, so the \"disfavored\" statement is partly about an unphysical branch of the fit. That does not make the measurement less useful, but it tempers the physics conclusion. The trigger and reconstruction bias correction uses NNPDFpol1.1 as input, introducing a mild model dependence; the correction is roughly 10% of A_LL with a small systematic, so this is not a major worry. Tabulated data and the ML architecture are not released, which would help independent scrutiny.\n\nWho this is for: the spin-physics community and global fitters. It deserves review, but the significance claim needs to be redone and the numbers released. I would send it to a serious referee, with the clear instruction that Section VII.A must be fixed before acceptance.","headline":"A careful 2015 STAR dijet A_LL measurement, but the headline 3.5 sigma disfavor of the JAM22 negative-gluon solution is built on a chi-square conversion that does not hold up.","tokens_in":31343,"tokens_out":2043,"would_cite":true,"duration_ms":19834,"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":"The STAR Collaboration finds that the JAM22 negative-gluon-polarization solution is disfavored at the 3.5 sigma level by a new precision measurement of the dijet double-spin asymmetry at intermediate pseudorapidity in 200 GeV polarized…","keywords":["longitudinal double-spin asymmetry","dijet production","gluon helicity","polarized proton-proton collisions","STAR experiment","RHIC","JAM22","parton distribution functions"],"falsifier":"Rerun the JAM22 global fit including the STAR 2015 intermediate-pseudorapidity dijet data while keeping the negative-gluon solution: if the fit still yields an acceptable $\\chi^2$ under the same correlated-systematics treatment, the 3.5-$\\sigma$ exclusion is not robust. A second, more direct check is to recompute the parton-level dijet mass using a data-driven endcap tracking-efficiency correction instead of the PYTHIA-trained neural network; if the resulting $A_{LL}$ points move by more than the quoted systematic uncertainty, the comparison against JAM22 would need revision.","tokens_in":30271,"feed_emoji":"⚛️","tokens_out":8754,"duration_ms":71695,"temperature":0.7,"pith_summary":"This paper reports a new measurement of the longitudinal double-spin asymmetry $A_{LL}$ for dijet production in which at least one jet lies at intermediate pseudorapidity ($0.8 < \\eta_{\\rm jet} < 1.8$), using 52 pb$^{-1}$ of longitudinally polarized proton-proton collisions at $\\sqrt{s}=200$ GeV recorded by the STAR experiment in 2015. The measurement probes partonic momentum fractions $x$ from 0.01 to 0.5, where the scattering is dominated by high-$x$ valence quarks hitting low-$x$ gluons, so $A_{LL}$ is a direct handle on the gluon helicity distribution $\\Delta g(x)$. The data agree with the earlier 2009 STAR measurement and with the predictions of global QCD analyses that find $\\Delta g(x)>0$, and they disfavor the negative-gluon-polarization solution of the JAM22 global fit at the 3.5 $\\sigma$ level. If the result holds, it settles a genuine ambiguity: two fits of nearly equal quality to earlier inclusive-jet data had opposite signs for the gluon spin, and this dijet measurement breaks the degeneracy toward a positive gluon polarization.","feed_headline":"STAR dijet data reject negative gluon spin at 3.5 sigma","feed_subtitle":"New 200 GeV asymmetry measurement with an endcap jet breaks the sign ambiguity in the gluon helicity.","key_machinery":"The carrying object is the longitudinal double-spin asymmetry $A_{LL} = (\\sigma^{++} - \\sigma^{+-})/(\\sigma^{++} + \\sigma^{+-})$, measured as a function of the dijet invariant mass. At leading order the dijet mass and the pseudorapidity sum of the two jets give $M = \\sqrt{s x_1 x_2}$ and $\\eta_3 + \\eta_4 = \\ln(x_1/x_2)$, so the endcap topology selects collisions in which a high-$x$ valence quark meets a low-$x$ gluon. Because the TPC tracking efficiency falls sharply in the endcap, the analysis uses a Multilayer Perceptron (a neural-network regression) trained on PYTHIA Perugia 2012 plus GEANT3 simulations to correct each jet's $p_T$ and mass, and applies a particle-to-parton mass shift from the same simulation to place the measured $A_{LL}$ at the parton-level dijet mass. A trigger and reconstruction bias correction computed from 100 NNPDFpol1.1 replicas is subtracted from the raw asymmetry, and the JAM22 comparison is made with a $\\chi^2$ that folds in the three dominant correlated systematics.","core_discovery":"The central claim is that the gluon helicity distribution $\\Delta g(x)$ is positive in the $x$ range 0.01 to 0.5, and that the alternative negative solution found in the JAM22 global analysis is excluded. The paper establishes this by measuring $A_{LL}$ as a function of parton-level dijet invariant mass for three dijet topologies (east Barrel--Endcap, west Barrel--Endcap, and Endcap--Endcap), combining the 2009 and 2015 STAR data. A $\\chi^2$ test that accounts for correlated uncertainties from relative luminosity, dijet energy scale, and beam polarization gives an average $\\chi^2$ of 20.4 for 20 degrees of freedom for the positive JAM22 solution and 32.7 for the negative one, corresponding to a 3.5-$\\sigma$ disfavoring of the negative solution. The paper further notes that in the negative-gluon scenario the gluon-gluon subprocess would need a negative opposite-helicity cross section at high dijet mass, which is the feature the data reject.","pith_inferences":["If the exclusion is robust, one can translate the measured $A_{LL}$ into a quantitative lower bound on the truncated moment of $\\Delta g(x)$ over $0.01 < x < 0.5$; the paper does not quote such a bound directly from this dataset.","A future measurement at $\\sqrt{s}=510$ GeV with forward endcap coverage could push the same technique to $x$ below 0.01, testing whether the positive sign persists where the gluon density is large.","The strength of the 3.5-sigma exclusion depends on the JAM22 fit's scale choices and on the NLO dijet calculation; repeating the comparison with a different NLO framework or with the DSSV14 uncertainty band would reveal how much of the significance is tied to the JAM22 model.","The positivity-violating feature of the negative solution could be probed directly: a dedicated measurement of the gluon-gluon subprocess fraction in the high-mass endcap bins would provide an independent check of whether $\\sigma^{+-}$ for gluon-gluon scattering is truly negative there."],"forward_implications":["The 2015 data, when combined with the 2009 points, give the most precise $A_{LL}$ measurement to date for dijets with an endcap jet at 200 GeV, with sensitivity extending down to $x \\sim 0.01$.","Incorporating these results into global QCD analyses should reduce the uncertainty on the integrated gluon helicity contribution without significantly shifting its central value.","The JAM22 negative-gluon solution, which violates the positivity bound $|\\Delta f_i| \\le f_i$, is now excluded at 3.5 sigma by direct dijet data, strengthening the case that positivity-breaking negative solutions cannot describe RHIC spin data.","The result closes a loop opened by the 2022 JAM22 analysis, which had shown that inclusive-jet $A_{LL}$ data alone could not distinguish opposite-sign gluon solutions."],"supporting_citations":[{"why":"The JAM22 global QCD analysis that produced the positive and negative gluon polarization solutions; the negative solution is the target of the comparison.","marker":"[25]"},{"why":"Previous STAR measurement of intermediate-pseudorapidity dijet $A_{LL}$ from 2009 data, which this paper extends and compares against.","marker":"[16]"},{"why":"The DSSV14 polarized PDF set used as the positive-gluon theoretical prediction for $A_{LL}$.","marker":"[3]"},{"why":"The NNPDFpol1.1 polarized PDF set used for theoretical predictions and as input to the trigger and reconstruction bias correction through its replica ensemble.","marker":"[2]"},{"why":"The NLO dijet production code used to compute the theoretical $A_{LL}$ expectations for the positive and negative gluon scenarios.","marker":"[53]"},{"why":"The Perugia 2012 tune of PYTHIA used to generate the simulation that trains the machine-learning jet corrections and provides the particle-to-parton mass shift.","marker":"[38]"},{"why":"The RHIC polarimetry report that supplies the beam polarization values and the 6.1% scale uncertainty used in the $\\chi^2$ test.","marker":"[36]"},{"why":"The STAR mid-rapidity inclusive jet and dijet $A_{LL}$ results used for consistency checks and correlation matrices.","marker":"[24]"}],"fun_headline_variants":["STAR: negative gluon spin ruled out at 3.5 sigma","Dijet data pin gluon helicity to positive, 3.5 sigma","Gluon polarization: STAR confirms positive sign","STAR dijets reject negative gluon spin decisively","3.5 sigma: gluon helicity positive, not negative"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The result assumes that the PYTHIA Perugia 2012 tune (with energy-scale parameter PARP(90)=0.213) combined with the GEANT3 detector simulation faithfully reproduces the reduced tracking efficiency in the endcap and the particle-to-parton relation; if that simulation is biased, the parton-level $A_{LL}$ values could shift.","fun_headline_variants_meta":{"raw":{"variants":["STAR: negative gluon spin ruled out at 3.5 sigma","Dijet data pin gluon helicity to positive, 3.5 sigma","Gluon polarization: STAR confirms positive sign","STAR dijets reject negative gluon spin decisively","3.5 sigma: gluon helicity positive, not negative"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000236,"raw_usage":{"total_tokens":1504,"prompt_tokens":947,"completion_tokens":557,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":563,"completion_tokens_details":{"reasoning_tokens":466}},"tokens_in":563,"tokens_out":557,"duration_ms":5917,"temperature":1.0,"reasoning_tokens":466,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T10:40:56.456875+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Rerun the JAM22 global fit including the STAR 2015 intermediate-pseudorapidity dijet data while keeping the negative-gluon solution: if the fit still yields an acceptable $\\chi^2$ under the same correlated-systematics treatment, the 3.5-$\\sigma$ exclusion is not robust. A second, more direct check is to recompute the parton-level dijet mass using a data-driven endcap tracking-efficiency correction instead of the PYTHIA-trained neural network; if the resulting $A_{LL}$ points move by more than the quoted systematic uncertainty, the comparison against JAM22 would need revision.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The JAM22 global QCD analysis that produced the positive and negative gluon polarization solutions; the negative solution is the target of the comparison."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The DSSV14 polarized PDF set used as the positive-gluon theoretical prediction for $A_{LL}$."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The NLO dijet production code used to compute the theoretical $A_{LL}$ expectations for the positive and negative gluon scenarios."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The RHIC polarimetry report that supplies the beam polarization values and the 6.1% scale uncertainty used in the $\\chi^2$ test."},{"cited_title":"Adareet al.(PHENIX), Phys","cited_arxiv_id":null,"evidence_quote":"The STAR mid-rapidity inclusive jet and dijet $A_{LL}$ results used for consistency checks and correlation matrices."}],"review_version":1}