{"id":"9fba32b5-2ce1-4351-97d9-9256204db981","arxiv_id":"2611.05042","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"A four-parameter parton-pair correlation folded over hydrodynamic fireball evolution improves direct-photon yields at RHIC and LHC, but raises elliptic flow mainly at the cost of violating local positivity.","lead":"This paper asks whether correlations between pairs of quarks and gluons in the hot fireball created by heavy-ion collisions can explain the measured direct-photon yields and their elliptic flow together. It finds that a four-parameter correlation model improves the description of photon yields at RHIC and LHC, while improving the flow mainly only when the model is allowed to dip into unphysical negative emission rates.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek/deepseek-v4-flash","headline":"Central claim lacks independent support: C uses VISHydro profiles while baseline B uses a different IP-Glasma+KøMPøST+music evolution (b88, b96), so fitted coefficients may absorb profile differences; a same-evolution refit is needed before 'correlation raises yield/v2' is physical.","rationale":"The paper is honest and technically careful: it reports the positivity trade-off (b167, Fig. 4g,h), the missing bremsstrahlung and inelastic-annihilation responses (b176, b178), and the different hydrodynamic evolutions (b88, b96). I therefore treat none of these as concealment. The most load-bearing issue for the abstract's causal claim is the footing mismatch, because it precedes the others: every fitted coefficient, including the yield improvement that survives the hard constraints, is defined relative to a baseline from one evolution and a response from another. Since the four coefficients are selected and fitted to exactly the yield and v2 residuals, they can in principle compensate for differences in temperature history, flow profile, and switching time between the two calculations. Fig. 3's attribution of most of the v2 increase to the q_T columns is especially exposed, because ln[T/(0.2 GeV)] is the feature most sensitive to which hydrodynamic temperature history is used. Mitigating factors are real: the fits are heavily overdetermined, and the within-energy centrality transfer and shared-fit tests (Figs. 5-6) are genuine out-of-sample checks. The missing-channel and positivity concerns are secondary for this verdict: the yield improvement survives hard constraints, and the missing channels are explicitly acknowledged, but neither can be interpreted cleanly until B and C live on the same evolution. The proposed check uses the authors' own two evolutions and does not require rederiving the published companion calculation, so it directly settles whether the fitted four columns are physical or compensatory.","tokens_in":19942,"tokens_out":8218,"duration_ms":60787,"concrete_test":"Recompute the baseline B on the same VISHydro profiles used for C, or equivalently fold C over the IP-Glasma+KøMPøST+music profiles of Ref. [39], and refit the four coefficients under the same hard/soft constraints. Then compare the resulting chi-squared, the fitted coefficients, and the Fig. 3 decomposition. If the yield and v2 improvements and the dominance of the q_T columns survive, the different-footing issue is not responsible for the result; if the fit degrades or the coefficients shift substantially, the four columns were absorbing hydro-profile differences rather than a physical pair correlation.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing condition for the headline claim is that B and C describe the same evolving medium. Equation (13) adds B_n = P_n + E_n + H_n from the published source files of Ref. [39] to C_n folded over VISHydro/MCGlb profiles (Sec. II B). The paper itself states that these profiles are 'similar but not identical' (b88) and lists the mismatch as a limitation of interpretation (b95, b96). Because the four coefficients are fitted to the joint yield and v2 data, any systematic difference in temperature history, flow profile, or switching-time sampling between the two evolutions enters C_n directly. The temperature feature q_T = ln[T/(0.2 GeV)] is exactly the feature that Fig. 3 identifies as supplying most of the intermediate-pT v2 gain, and it is the feature most sensitive to which hydrodynamic temperature history is used. The cross-energy transfer asymmetry (Sec. III, Figs. 5-6) is consistent with this concern: a single coefficient vector does not preserve fit quality and positivity across systems. As it stands, the improvement could be a four-parameter compensation of two different hydro calculations rather than evidence for a physical parton-pair correlation. The authors acknowledge this limitation and promise a same-evolution calculation (b227, b292), but the central claim is advanced before that check is done. This is an external-support deficit, not an internal inconsistency.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek/deepseek-v4-flash","summary":"The paper explores whether a connected two-parton correlation, added to the local 2-to-2 direct-photon rate, can simultaneously improve the description of direct-photon yields and elliptic flow in Au+Au (200 GeV) and Pb+Pb (2.76 TeV) collisions. The authors construct a correlation response library from zero-marginal modes (Eq. 5), fold the response over VISHydro profiles, add it to a fixed baseline P+E+H from Ref. [39], and fit four coefficients per centrality class to the joint yield and v2 data (Eq. 14). They report that the correlation raises the yield toward the data and increases v2 mainly at intermediate pT, with the temperature-dependent coefficient columns supplying most of the net flow increase (Figs. 2 and 3). They also test positivity constraints on the local rate and two-particle distribution, and find a trade-off: better in-sample v2 fits are associated with more frequent negativity (Fig. 4). Transfer tests across centralities and collision energies are examined (Figs. 5-7). The authors explicitly acknowledge that the baseline and correlation response use similar but not identical hydrodynamic evolutions, and that the response uses a leading-logarithmic 2-to-2 classical-statistics approximation.","tokens_in":22241,"tokens_out":1846,"duration_ms":32139,"significance":"The work is a serious exploratory study of a mechanism that could address the long-standing direct-photon puzzle: a pair correlation can change emission rates without altering single-particle distributions. The technical machinery is largely sound and transparent: the rate convolution, moment folding, the exact decomposition of the v2 change in Eq. (15), and the positivity diagnostics are useful contributions. The paper is also commendably honest: it labels the fits as in-sample, reports the limitations of the mixed hydrodynamic profiles and the truncated response, and includes transfer tests. However, the central claim as stated in the abstract is only fully realized in the soft- and unconstrained fits; the hard-constraint fit improves v2 less than the yield, and the improvement is partly a consequence of the in-sample fitting of four coefficients per class against the very data being described. As an exploratory mechanism study, the paper is valuable; as an independent confirmation of a physical pair-correlation mechanism, it is not yet conclusive.","major_comments":[{"comment":"The central claim that the correlation 'increases v2 mainly at intermediate pT' is not supported by the hard-constraint fit, which is the physical fit; the paper itself states that most measured v2 points remain above the baseline and the hard-constraint curve (b163), and that the three reconstructions separate most clearly around the shoulder. The claim is fully realized only in the soft- and unconstrained fits, where positivity is violated. The authors should either make the claim explicitly conditional on the constraint setting, or move the hard-constraint fit to the center of the presentation.","section":"Abstract and Sec. IIB, IIIC (b4, b88, b96)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The baseline B = P+E+H is taken from published source files of Ref. [39] (IP-Glasma + KøMPøST + music), while the correlation response C is folded over VISHydro profiles that the authors themselves describe as 'similar but not identical' (b88). Because the four coefficients are fitted to the joint yield and v2 data, any systematic difference in temperature history, flow profile, or switching-time sampling between the two evolutions enters C_n directly. The temperature feature q_T = ln[T/(0.2 GeV)] is exactly the feature that Fig. 3 identifies as supplying most of the intermediate-pT v2 gain, and it is the feature most sensitive to which hydrodynamic temperature history is used. As the authors acknowledge (b292), this is an external-support deficit. The current results therefore cannot distinguish a physical pair-correlation effect from a four-parameter compensation of two different hydro calculations. A same-evolution refit (which the authors promise in b227) is needed before the central claim can stand; at minimum, the abstract and conclusions should state that this test has not been done.","section":"Sec. IIB, IIC and Fig. 2 (b88, b95-b97)","authors_acknowledge":"The improved agreement may reflect a physical pair- correlation mechanism, although the use of different hy- drodynamic evolutions for the baseline and response and the leading-logarithmic approximation for the response limit the conclusions that can be drawn.","authors_acknowledge_p":0.92},{"comment":"The correlation response includes only 2-to-2 Compton and annihilation processes in the classical-statistics, leading-logarithmic limit (b176), while the AMY rate used for the positivity condition includes near-collinear bremsstrahlung and inelastic annihilation (b178). The missing channels could change Y0 and Y2 differently after integration over the medium, and thus affect v2 = Y2/Y0. Since the coefficients are fitted, the fit can partially compensate for the missing physics; this is an acknowledged limitation (b181), but it is load-bearing for the physical interpretation of the fitted amplitudes. The authors should note that the fitted coefficients are effective parameters that absorb missing response channels, not direct measurements of a physical correlation amplitude.","section":"Sec. IIA and IIB, Eqs. (8)-(9), b176-b178","authors_acknowledge":"These channels enter the AMY rate used for the local positiv- ity condition, but their response to the pair correlation is absent from the fitted correction.","authors_acknowledge_p":0.91},{"comment":"The paper reports that the v2 fit quality improves as the constraints are relaxed, with negative local-rate points rising to 2%--15% in the unconstrained fits and negative pair distributions appearing. This means the improvement in v2 over the baseline at intermediate pT is largely carried by fits that violate the physical positivity conditions. The hard-constraint fit, which enforces both conditions, shows a smaller improvement. The paper should be explicit that the mechanism's ability to resolve the direct-photon puzzle under the physical constraints is, at present, only modest; the strong intermediate-pT v2 increase is a property of the unconstrained fit.","section":"Fig. 4(g,h) and Sec. III, b167","authors_acknowledge":null,"authors_acknowledge_p":null}],"minor_comments":[{"comment":"There are several typos and formatting issues: 'rspectively' in the Fig. 3 caption, 'contraint' and 'experiemental' in the Fig. 2 caption, 'tranferred' in Sec. III, and 'hotizontally' in the Fig. 7 caption. These should be corrected in a final pass.","section":"Throughout","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The panels in Fig. 4(a)-(d) use a symmetric logarithmic scale that is linear for small absolute values. A short explanation in the caption of how to read negative and near-zero values would help the reader.","section":"Sec. III, Fig. 4","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The notation B_n = P_n + E_n + H_n is introduced without a table of symbols. Since B_n appears throughout the results and figures, a concise definition of each term in a table or near Eq. (13) would improve readability.","section":"Sec. IIC, Eq. (13)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The selection of the four columns is described by forward selection, but the paper does not report the values of the four fitted coefficients beyond a qualitative plot in Fig. 7. Reporting the actual values in a table would strengthen reproducibility and allow readers to assess the physical size of the correlation.","section":"Sec. IIA and Appendix A","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The conclusion that the two centralities at each collision energy are consistent with a shared coefficient vector (Sec. III and Fig. 5) is interesting, but the paper does not quantify the significance of the small increase in squared pulls (64.7 to 66.0, 31.9 to 33.4). A brief comment on whether these differences are within the fit noise would be useful.","section":"Sec. IV","authors_acknowledge":null,"authors_acknowledge_p":null}],"recommendation":"major_revision","confidential_remarks":"The paper is well-written and technically careful, but the central claim is advanced before the acknowledged same-evolution test is done. I would urge the editor to require that the abstract and conclusions explicitly distinguish the exploratory fit improvement from a confirmed physical mechanism. If the authors can provide a same-evolution demonstration (even in one centrality class) or substantially weaken the claim, the paper would be acceptable. The paper's strength is its transparent methodology and honest reporting of limitations; its weakness is that the headline claim overstates the evidential support."},"author_rebuttal":null,"desk_editor":{"model":"deepseek/deepseek-v4-flash","letter":"You should know two things before reading: this is a serious, unusually honest phenomenological paper, and the abstract is one notch stronger than what the fits actually show. The robust result is the yield improvement, which survives hard positivity constraints. The headline v2 gain is fully realized only in the soft and unconstrained fits, which admit negative local rates and negative pair distributions; the paper says so itself (sentence b167) and plots the trade-off openly in Fig. 4g,h.\n\nWhat is actually new: the authors take the local pair-correlation response from their companion paper (Ref [42], posted on arXiv but not yet published) and fold it over realistic hydrodynamic profiles and published PHENIX and ALICE bins. The machinery is a 73-mode response library, two features, four fitted coefficients per class, and a joint yield-plus-v2 fit per centrality. The yield improvement over the Ref [39] baseline persists even when both positivity conditions are enforced, which is a genuine and non-obvious result. The exact decomposition with K = ΔY2 − vB2 ΔY0 (Eq. 15) is clean, and splitting K into temperature-column and constant-column parts is a sensible way to localize where the gain comes from. The fits are heavily overdetermined, and the cross-centrality transfer is a real out-of-sample check.\n\nThe soft spots, in proportion. First and load-bearing: B and C come from different hydrodynamic evolutions, VISHydro for the correlation response and IP-Glasma plus KøMPøST plus music for the baseline. The authors acknowledge this (b88, b96) and promise the same-evolution refit (b227, b292), but it is not a minor technicality. Four fitted coefficients per class are exactly enough to compensate systematic differences between two hydro calculations, and the temperature feature q_T = ln(T/0.2 GeV), which supplies most of the net v2 gain, is the feature most sensitive to which temperature history you use. The central claim is therefore conditional until that refit is done, and the asymmetric cross-energy transfer is consistent with the worry. Second, only 2-to-2 Compton and annihilation in a classical-statistics, leading-log limit enter the response; bremsstrahlung and inelastic annihilation are missing (b176, b178). The authors flag this as well. Third, the fitted coefficients are shown without uncertainties (Fig. 7), and the rate calculation lives in the companion, so the response library is not independently verifiable from this paper alone.\n\nNone of this kills the paper. It is honestly labeled throughout: fits are called fits, failures are reported, trade-offs are quantified. It is for heavy-ion phenomenologists working on electromagnetic probes, and it deserves a serious referee; I would expect a conditional accept. The decisive test is the promised same-evolution, complete-leading-order calculation, and that version, if it holds up, will be the citable one. The cs.LG category is simply wrong; this is hep-ph.","headline":"A serious, honest attempt on the direct-photon problem whose yield improvement survives positivity constraints, but whose headline v2 gain rests on fits with unphysical negative rates and on two different hydro evolutions.","tokens_in":20998,"tokens_out":5185,"would_cite":false,"duration_ms":37282,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"pith_extraction":{"msc":[],"pacs":["25.75.-q","12.38.Mh"],"model":"deepseek/deepseek-v4-flash","headline":"A connected parton-pair correlation that leaves single-particle distributions unchanged raises the direct-photon yield toward the PHENIX and ALICE data and increases elliptic flow $v_2$ at intermediate $p_T$ once folded over the evolving…","keywords":["direct photons","heavy-ion collisions","elliptic flow","parton pair correlations","quark-gluon plasma","thermal photon emission","connected two-particle correlation","direct-photon puzzle"],"falsifier":"Recompute the baseline and the pair-correlation correction on one common hydrodynamic evolution with the response extended to complete leading order, so that the bremsstrahlung and inelastic-annihilation channels enter on the same footing as in the AMY rate; if the joint yield and $v_2$ improvement then disappears, or survives only while a non-negligible fraction of cells has negative local rates or negative pair distributions, the mechanism is not supported. A second, already available check is the unfitted STAR virtual-photon yield, which the fitted correlation worsens; a measurement with finer $p_T$ coverage would sharpen that discrimination.","tokens_in":21247,"feed_emoji":"🔆","tokens_out":12951,"duration_ms":81618,"temperature":0.7,"pith_summary":"Direct photons escape the collision fireball almost without rescattering, so they carry the full emission history, and the long-standing problem is that one calculation must reproduce both the measured photon yield and the sizable elliptic flow $v_2$. This paper proposes a connected correlation between the two partons entering a $2\\to 2$ photon-producing collision: because such a correlation enters the local rate linearly without changing any one-particle distribution, it can modify emission without disturbing the medium itself. Fitting just four correlation coefficients per collision class and folding them over the expanding medium, the paper moves a fixed multistage baseline's yield toward the fitted PHENIX and ALICE data and raises $v_2$ mainly at intermediate $p_T$. The temperature-dependent part of the correlation supplies most of the net flow increase, which the authors read as the correlation reweighting photon emission between hotter and cooler fluid cells. The authors state plainly that the evidence is suggestive rather than conclusive, because the baseline and the correlation response come from different hydrodynamic evolutions and the response covers only a subset of photon-production channels.","feed_headline":"Pair correlations raise photon yield and v2 in one fit","feed_subtitle":"Folded over the expanding medium, four fitted coefficients move yield and elliptic flow toward the measured data.","key_machinery":"The load-bearing object is the connected two-parton correlation $g_{ab}$, defined by the requirement that its integrals over either momentum vanish, so it changes the photon rate without disturbing the one-particle distributions $f_a f_b$. It is expanded in a momentum-space basis of Laguerre polynomials in the temperature-scaled energies $\\xi_i=p_i\\cdot u/T$ and Legendre polynomials in the relative angle $z_{ab}=\\hat p_a\\cdot\\hat p_b$, multiplied by two spatial features — a constant and $\\ln[T/(0.2\\,\\mathrm{GeV})]$ — for 146 candidate coefficients, of which four are kept: modes $(0,0,3)$ and $(1,0,2)$ with both features. Each mode carries a precomputed local spectral response $R_{nm\\ell}(x,T)$ with $x=k\\cdot u/T$; because the correction is linear in $g_{ab}$, the azimuthal moments $Y_0$ and $Y_2$ are affine in the coefficients and $v_2=Y_2/Y_0$ is their ratio. The argument that the correlation can raise flow runs through the identity $K=\\Delta Y_2-v_2^B\\Delta Y_0$: $v_2$ rises if and only if $K>0$, so the added second harmonic must outweigh the anisotropy-weighted yield increase. The emission model is the complete leading-order AMY rate plus the correlation correction, with the AMY rate serving as the reference for the local-positivity constraint.","core_discovery":"The central claim is that a connected two-parton correlation $g_{ab}(X;p_a,p_b)$, defined by vanishing one-particle marginals, enters the local $2\\to2$ photon rate linearly and, once folded over the hydrodynamic medium, improves both direct-photon observables at once in Au+Au at 200 GeV and Pb+Pb at 2.76 TeV. Keeping the prompt, pre-equilibrium, and hydrodynamic baseline fixed, four correlation coefficients per centrality class raise the yield toward the fitted PHENIX and ALICE bins — even when local nonnegativity of the rate and of the pair distribution is enforced — and increase $v_2$ mainly at intermediate $p_T$. The decomposition $K=\\Delta Y_2 - v_2^B\\Delta Y_0$ shows that the correlation adds a second harmonic while the accompanying yield increase partly cancels it, and that the temperature-dependent columns $q_T=\\ln[T/(0.2\\,\\mathrm{GeV})]$ provide most of the positive $K$ in all four classes. The $v_2$ gain grows as the positivity constraints are relaxed, at the cost of negative local rates at 0.4–15% of tested points and negative pair-distribution values in up to roughly 1.3% of the weighted domain.","pith_inferences":["If the mechanism is physical, the four fitted coefficients constitute a first measurement of the medium's connected two-parton correlation, which could be checked against correlations predicted by solving the transport equation for correlation functions — a test the paper itself does not perform.","Because the $v_2$ gain grows precisely as local positivity fails, an alternative reading is that the omitted bremsstrahlung and inelastic-annihilation responses, or the mismatch between the two hydrodynamic evolutions, carry part of the signal; only the same-profile, complete-leading-order calculation the authors propose can separate the readings.","The same zero-marginal construction applies to dilepton emission, where the final state is cleaner; a dilepton version of the fit would cross-check the direction and temperature dependence of the fitted correlation independently of photon data.","The asymmetric transfer across energies — RHIC-fitted coefficients applied to LHC profiles improve $v_2$ but worsen positivity — suggests the fitted temperature dependence is steeper than the hotter LHC medium can accommodate, which additional high-$p_T$ photon-flow data could test."],"forward_implications":["Within each collision energy, the two centrality classes are consistent with one shared coefficient vector, so a single correlation prescription per system describes both 0–20% and 20–40% collisions without refitting.","The yield improvement survives hard local-positivity constraints, so raising the yield toward the PHENIX and ALICE data does not depend on unphysical negative emission rates.","The largest $v_2$ gains appear when the positivity constraints are relaxed, so the strongest flow signal coincides with negative local rates and negative pair-distribution values in a small fraction of the domain.","The temperature-dependent part of the correlation, not the constant part, drives most of the net $v_2$ increase at intermediate $p_T$, meaning the data primarily constrain how correlations reweight emission from hotter versus cooler cells.","The same mechanism increases the tension with the unfitted STAR virtual-photon yields, so the improvement is specific to the fitted PHENIX and ALICE data products."],"supporting_citations":[{"why":"Supplies the fixed baseline $B=P+E+H$ of prompt, pre-equilibrium, and hydrodynamic thermal photons whose published results the correlation correction is added to.","marker":"[39]"},{"why":"Develops the local pair-correlation photon-rate calculation that this paper folds over the medium; the earlier demonstration that left the data comparison open.","marker":"[42]"},{"why":"Provides the complete leading-order equilibrium photon rate $\\Gamma^{\\mathrm{AMY}}_\\gamma$ used as the reference in the additive emission model and in the local-positivity constraint.","marker":"[40]"},{"why":"Supplies the VISHydro hydrodynamic profiles over which the correlation response is folded into laboratory observables.","marker":"[43]"},{"why":"PHENIX conversion-method direct-photon $v_2$ data in Au+Au used as fitted targets.","marker":"[35]"},{"why":"PHENIX direct-photon yield spectrum in Au+Au used as fitted data.","marker":"[34]"},{"why":"ALICE direct-photon $v_2$ data in Pb+Pb used as fitted targets.","marker":"[14]"},{"why":"ALICE direct-photon yield in Pb+Pb used as fitted data, including the two upper-limit bins.","marker":"[38]"},{"why":"STAR virtual-photon yield compared at published points but excluded from the fit; the comparison registers the added tension.","marker":"[36]"}],"fun_headline_variants":["Parton pair correlations fix photon yield and v2 together","Correlated partons lift direct-photon yield and elliptic flow","One correlation term moves photon yield and v2 at once","Parton correlations narrow the direct-photon v2 puzzle","Two-parton correlations nudge photon data into one fit"],"cache_read_input_tokens":0,"weakest_assumption_plain":"The load-bearing premise is that the four fitted coefficients describe a real connected pair correlation of the medium, rather than absorbing the differences between the baseline and response hydrodynamic evolutions and the photon channels that the response omits.","fun_headline_variants_meta":{"raw":{"variants":["Parton pair correlations fix photon yield and v2 together","Correlated partons lift direct-photon yield and elliptic flow","One correlation term moves photon yield and v2 at once","Parton correlations narrow the direct-photon v2 puzzle","Two-parton correlations nudge photon data into one fit"]},"model":"deepseek/deepseek-v4-flash","effort":"low","cost_usd":7.9e-05,"raw_usage":{"cost":7.8897e-05,"is_byok":false,"cost_details":{"upstream_inference_cost":7.8897e-05,"upstream_inference_prompt_cost":2.6663e-05,"upstream_inference_completions_cost":5.2234e-05},"total_tokens":1332,"prompt_tokens":1045,"completion_tokens":287,"prompt_tokens_details":{"audio_tokens":0,"video_tokens":0,"cached_tokens":940,"cache_write_tokens":0},"completion_tokens_details":{"audio_tokens":0,"image_tokens":0,"reasoning_tokens":218}},"tokens_in":1045,"tokens_out":287,"cost_source":"openrouter_usage","duration_ms":4750,"image_pages":[],"temperature":1.0,"billed_attempts":1,"reasoning_tokens":218,"transport_provider":"openrouter","cache_read_input_tokens":0,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-11-04T06:00:50.291835+00:00","model_set":{"reader":"deepseek/deepseek-v4-flash"},"falsifier":"Recompute the baseline and the pair-correlation correction on one common hydrodynamic evolution with the response extended to complete leading order, so that the bremsstrahlung and inelastic-annihilation channels enter on the same footing as in the AMY rate; if the joint yield and $v_2$ improvement then disappears, or survives only while a non-negligible fraction of cells has negative local rates or negative pair distributions, the mechanism is not supported. A second, already available check is the unfitted STAR virtual-photon yield, which the fitted correlation worsens; a measurement with finer $p_T$ coverage would sharpen that discrimination.","supporting_citations":[{"cited_title":"Multi-messenger heavy-ion collision physics","cited_arxiv_id":"2106.11216","evidence_quote":"Supplies the fixed baseline $B=P+E+H$ of prompt, pre-equilibrium, and hydrodynamic thermal photons whose published results the correlation correction is added to."},{"cited_title":"Probing Pair Correlations in QCD Matter with Photon Spectra","cited_arxiv_id":"2606.02414","evidence_quote":"Develops the local pair-correlation photon-rate calculation that this paper folds over the medium; the earlier demonstration that left the data comparison open."},{"cited_title":"Photon Emission from Quark-Gluon Plasma: Complete Leading Order Results","cited_arxiv_id":"hep-ph/0111107","evidence_quote":"Provides the complete leading-order equilibrium photon rate $\\Gamma^{\\mathrm{AMY}}_\\gamma$ used as the reference in the additive emission model and in the local-positivity constraint."},{"cited_title":"The iEBE-VISHNU code package for relativistic heavy-ion collisions","cited_arxiv_id":"1409.8164","evidence_quote":"Supplies the VISHydro hydrodynamic profiles over which the correlation response is folded into laboratory observables."},{"cited_title":"Adareet al.(PHENIX), Azimuthally anisotropic emis- sion of low-momentum direct photons in Au+Au colli- sions at √sN N = 200 GeV, Phys","cited_arxiv_id":null,"evidence_quote":"PHENIX conversion-method direct-photon $v_2$ data in Au+Au used as fitted targets."},{"cited_title":"Centrality dependence of low-momentum direct-photon production in Au+Au col- lisions at √sN N = 200 GeV,","cited_arxiv_id":null,"evidence_quote":"PHENIX direct-photon yield spectrum in Au+Au used as fitted data."},{"cited_title":"Direct photon elliptic flow in Pb-Pb collisions at $\\sqrt{s_{\\rm NN}}$ = 2.76 TeV","cited_arxiv_id":"1805.04403","evidence_quote":"ALICE direct-photon $v_2$ data in Pb+Pb used as fitted targets."},{"cited_title":"Direct photon production in Pb-Pb collisions at $\\sqrt{s_\\rm{NN}}$ = 2.76 TeV","cited_arxiv_id":"1509.07324","evidence_quote":"ALICE direct-photon yield in Pb+Pb used as fitted data, including the two upper-limit bins."},{"cited_title":"Direct virtual photon production in Au+Au collisions at $\\sqrt{s_{NN}}$ = 200 GeV","cited_arxiv_id":"1607.01447","evidence_quote":"STAR virtual-photon yield compared at published points but excluded from the fit; the comparison registers the added tension."}],"review_version":1}