{"id":"ba1f83a3-3d34-476b-a125-a6e21492ef6d","arxiv_id":"2501.14487","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"A public extension of MadGraph5_aMC@NLO now computes automated NLO QCD cross sections for asymmetric hadron-hadron collisions, validated against MCFM and JAM.","lead":"The authors extended the MadGraph5_aMC@NLO event generator to handle collisions between two different hadron species, such as proton-lead or pion-nucleus, at next-to-leading order in QCD. The new tool lets heavy-ion and hadron-structure physicists compute cross sections and nuclear modification factors with automatic scale and PDF uncertainties from a single run.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The alpha_s-from-second-PDF convention (Eq. 7) makes automatically computed h1h2 and h2h1 cross sections inconsistent when the two PDF sets carry different alpha_s, undermining the both-ordering claim.","rationale":"The paper is a solid tool paper: it validates the asymmetric implementation against MCFM for pPb W/Z production and against the JAM prediction for pi-W Drell-Yan, the code is publicly available, and the main limitations (unproven factorization for pA/AB, fixed-order-only operation, approximate NMF uncertainties) are explicitly stated. The reader's weakest assumption was collinear factorization for different hadron species, but that is a physics framework assumption that the authors do not claim to prove; the tool is explicitly presented as working within collinear factorization. A more concrete, internal load-bearing concern is the alpha_s-from-second-PDF convention in Eq. (7), which the reader noted only as a secondary practical consequence. Because the code advertises automatic computation of both h1h2 and h2h1 cross sections, the convention can produce two numerically different results for the same physical process when the two PDF sets are not alpha_s-consistent. This is testable and would be a genuine bug if confirmed. However, the paper already explicitly warns users about the practical consequences of the alpha_s choice, and for the validated cases (compatible proton/nPDF sets) the issue does not affect the demonstrated agreement. Therefore the reader's ACCEPT verdict stands; I would not change it without first seeing the proposed consistency test fail on the public code.","tokens_in":24795,"tokens_out":11823,"duration_ms":116903,"concrete_test":"Using the public code, run pi- p -> mu+ mu- X at sqrt(s) = 21.7 GeV twice: once with a pion PDF (e.g., JAM21) as the first LHAID and a proton PDF second, and once with the order reversed, keeping beam energies and kinematical cuts fixed. Compare dsigma/dxF in the two runs under xF -> -xF; if the two results differ by more than the Monte Carlo integration error, the h2h1 output is not the physical mirror image, and the alpha_s choice must be made user-selectable or the both-ordering claim must be restricted to PDF sets with a common alpha_s.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim promises automated NLO cross sections for asymmetric systems, and Sec. 3.1 states that the code computes 'both h1h2 and h2h1 collisions' in a single run. However, Eq. (7) fixes alpha_s from LHAID_h2, i.e. from the PDF set of the second hadron. When the h2h1 ordering is computed, the roles of the two PDF sets are swapped, so the code evaluates the hard-scattering matrix element using the alpha_s of the other PDF set. If the two chosen LHAPDF sets do not share the same alpha_s value or running (a situation the paper explicitly anticipates for pion-induced reactions with older or inconsistent PDF sets), then dsigma(h1h2) and dsigma(h2h1) will not be mirror images under y -> -y, even though they describe the same physical process. The paper warns that the alpha_s choice 'can have practical consequences' (Sec. 3.1), but the code still silently outputs both orderings, so a user can obtain two different results for one physical observable depending on the arbitrary ordering in the LHAID list. This is an internal consistency issue in an advertised capability, and it is distinct from the broader factorization caveat for pA/AB collisions.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper reports an extension of MadGraph5_aMC@NLO in which the two incoming hadrons can be assigned distinct LHAPDF sets, enabling NLO QCD fixed-order cross sections for asymmetric collisions such as pPb, pi-p, pi-A, and AB. The authors adapt the existing reweighting machinery so that scale and PDF uncertainties are computed automatically for both symmetric and asymmetric settings, add a lab-frame/c.m.s.-frame option, and provide automatic computation of the two orderings h1h2 and h2h1. The implementation is validated by comparing W and Z production in pPb at sqrt(s_NN)=5.02 TeV against MCFM and Drell-Yan pair production in pi-W collisions at sqrt(s)=21.7 GeV against the JAM collaboration's NLO calculation, with good agreement. Illustrative pPb results for W/Z, c and b quark production, and H+b bbar production, including nuclear modification factors, are also presented, and the code is publicly available at nloaccess.in2p3.fr.","tokens_in":25069,"tokens_out":7676,"duration_ms":69063,"significance":"If the advertised capabilities hold, this fills a concrete gap: there is no publicly available, automated NLO QCD code for asymmetric hadronic collisions. The validations in Figs. 1 and 2 are genuine benchmarks against independent physics targets, and the automatic treatment of scale and PDF uncertainties is a practical strength. The pPb W/Z comparison is particularly valuable because it includes nPDF uncertainties, and the pi-W Drell-Yan validation against JAM is an important complement. The main caveats are that the MCFM comparison relies on an APPLgrid setup previously developed by the same group (refs. [110,111]), that the heavy-flavour predictions in Fig. 6 are not independently cross-checked, and that the collinear-factorisation assumption for pA/AB is not formally proven (acknowledged in Sec. 2.1). These caveats do not invalidate the central claim, but they should be visible to users. The code is downloadable from a URL, though no versioned archive is provided.","major_comments":[{"comment":"Equation (7) fixes the strong coupling used in the matrix element from LHAID_h2, i.e. from the PDF set of the second hadron. When the code computes 'both h1h2 and h2h1 collisions', as stated in item 2 of Sec. 3.1, the roles of the two PDF sets are exchanged and the alpha_s source is therefore swapped. If the two LHAPDF sets do not share the same alpha_s value and running, a case the paper explicitly anticipates in Sec. 3.1 for pion PDFs, then dsigma(h1h2) and dsigma(h2h1) will not be mirror images under y -> -y even though they describe the same physical process. The paper warns that the choice 'can have practical consequences' but the code still silently outputs both orderings, so a user can obtain two different results for one physical observable depending on the arbitrary ordering of the LHAID list. This inconsistency is load-bearing for the advertised both-ordering capability. Please either enforce alpha_s consistency when both orderings are requested, or compute both orderings with alpha_s taken from a single designated PDF set, and document the residual ambiguity in the run_card documentation.","section":"Sec. 3.1, Eq. (7); item 2 of the default computations"},{"comment":"All validation plots test only one ordering: h1h2 for pPb and for pi-W. The h2h1 ordering is advertised as an automatic output but is never validated, and because of the alpha_s asymmetry in Eq. (7) it is not guaranteed to equal the appropriately reflected h1h2 result. I request at least one explicit numerical check, e.g. a process with consistent PDF sets where the h2h1 output is compared with the reflected h1h2 output, or a clear statement limiting the h2h1 output to PDF sets with identical alpha_s.","section":"Sec. 3.2"}],"minor_comments":[{"comment":"There is a duplicated word in the sentence 'We compare our MG5aMC-based computation for d2sigma/dsqrt(tau)dx_F with with that of the JAM Collaboration'; 'with with' should be 'with'.","section":"Sec. 3.2"},{"comment":"The run_card template in the appendix contains the placeholder 'For details see: arXiv:XXXX'; this should be replaced with the actual paper identifier.","section":"Appendix"},{"comment":"The heavy-flavour and H+b bbar predictions in Fig. 6 are not cross-checked against any independent calculation; the text should state explicitly that these are illustrative and not validated, rather than leaving this to be inferred from the general disclaimer at the start of Sec. 4.","section":"Sec. 4.2, Fig. 6"},{"comment":"The LO MG5aMC band is significantly below the data and the NLO result; a brief comment on whether this is the expected K-factor for this kinematics would help the reader interpret the figure.","section":"Sec. 3.2, Fig. 2"},{"comment":"The code availability statement gives a URL but no version or checksum; for reproducibility, consider referencing a versioned archive (e.g. Zenodo) in addition to the live service.","section":"Sec. 5"},{"comment":"The notation for the hadron is 'h' in the symmetric formulas of Sec. 2.2 and 'h1/h2' in the asymmetric formulas of Sec. 3.1; using h1/h2 consistently throughout would avoid confusion.","section":"Sec. 2.2 and Sec. 3.1"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of EPJC and the code will likely be useful to the heavy-ion and pion-structure communities. My main concern is the alpha_s-ordering inconsistency in Eq. (7), which affects an explicitly advertised capability (both h1h2 and h2h1 outputs); this is fixable but should be addressed before acceptance. The reliance on the same group's APPLgrid setup for the MCFM benchmark is acceptable given the independent JAM comparison. I do not see grounds for rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is a solid tooling paper. The genuinely new thing is a public, automated NLO generator for asymmetric hadron-hadron collisions: two distinct LHAPDF sets, lab-frame bookkeeping, HwU extensions, and automatic scale/PDF uncertainties. That fills a real gap, since the existing asymmetric codes are mostly private or LO-only. The paper is honest about this gap, and the code is publicly available.\n\nThe validation is the strength. pPb W/Z production matches MCFM, and pi-W Drell-Yan matches JAM. Those are two independent physics targets with shared external PDFs, so the central claim, that the extension reproduces known NLO results, is well supported. The code appears careful and the documentation of the HwU format and run_card usage is clear enough for a user to reproduce the results.\n\nSoft spots, in proportion: the alpha_s-from-second-PDF convention in Eq. (7) is the one thing I would want fixed before publication. The paper explicitly warns that the choice can have practical consequences, and the appendix reminds users that alpha_s comes from the second LHAPDF set. But the code still computes both h1h2 and h2h1 in a single run, and if the two PDF sets carry different alpha_s values, those two results will not be mirror images under y -> -y even though they describe the same physical process. A user with an inconsistent pion/proton pair could silently get two different numbers. This is not fatal: the validations use consistent sets, and the physical content is otherwise standard collinear factorization. The fix is simple, either enforce equal alpha_s for the two orderings or emit a warning when the sets differ. I would treat this as a minor-to-moderate robustness issue, not a load-bearing flaw.\n\nThe other caveats are stated by the authors and are fine: factorization for pA/AB is not formally proven and power corrections may be enhanced; parton showering is untested; the heavy-flavour predictions in Fig. 6 are not independently cross-checked; and the NMF uncertainty in Eq. (11) ignores proton-PDF uncertainty in the denominator. The MCFM comparison uses an APPLgrid setup previously developed by the same group, so the independence of that benchmark is slightly weaker than it first appears, but the JAM comparison compensates.\n\nWho is this for? Heavy-ion and hadron-structure phenomenologists, and experimentalists who need NLO predictions for pPb, pion-induced, or nucleus-nucleus systems. It is a significant capability improvement, not a conceptual shift. The paper deserves a serious referee; I would recommend acceptance after a minor revision addressing the both-orderings alpha_s consistency.\n\nYes, bring it to reading group and cite it if you do phenomenology in this area.","headline":"Useful, well-validated tooling extension for asymmetric NLO collisions; the alpha_s-from-second-PDF bookkeeping is a real but documented caveat, not a reason to hold it up.","tokens_in":25612,"tokens_out":2662,"would_cite":true,"duration_ms":25979,"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 authors present an extension of MadGraph5_aMC@NLO that automates NLO QCD predictions for asymmetric hadron-hadron collisions (pA, AB, pion-hadron), validated against MCFM and JAM, with automatic scale and PDF uncertainties.","keywords":["asymmetric hadron-hadron collisions","next-to-leading order QCD","collinear factorization","nuclear parton distribution functions","MadGraph5_aMC@NLO","Drell-Yan production","nuclear modification factor","pion PDFs"],"falsifier":"A decisive check would be to run the code for a process in pPb or PbPb collisions where power corrections are expected to be large—for example forward $J/\\psi$ production, which the paper lists among processes sensitive to non-factorising effects—and compare the NLO prediction against LHC data; a systematic discrepancy growing toward forward rapidity or with nuclear size would signal that the factorisation assumption underlying the extension is breaking down. A second, more direct test of the implementation choices is to recompute the same asymmetric observable with the LHAID order reversed, so that $\\alpha_S$ is taken from the other PDF set, and check that the shift stays within the quoted uncertainties.","tokens_in":24610,"feed_emoji":"⚛️","tokens_out":8790,"duration_ms":71463,"temperature":0.7,"pith_summary":"This paper reports an extension of the MadGraph5_aMC@NLO event generator that enables automated next-to-leading-order (NLO) QCD calculations for hadronic collisions of two different species: proton–nucleus, nucleus–nucleus, pion–hadron, and pion–nucleus. The authors' aim is to give the heavy-ion and pion-beam communities a public tool that returns NLO cross sections, scale uncertainties, and PDF uncertainties for essentially any Standard Model or beyond-Standard-Model process, without the need for private modified codes. They validate the implementation against established references for W/Z production in proton–lead collisions and for Drell-Yan production in pion–tungsten collisions, and they illustrate it with nuclear modification factors for W, Z, charm, bottom, and associated Higgs production. If the extension is correct, it closes a gap that has forced asymmetric-collision predictions to rely on non-public or leading-order-only programs.","feed_headline":"One code now automates NLO QCD for proton-nucleus collisions","feed_subtitle":"Extension of MadGraph5_aMC@NLO validates on pPb W/Z and pion-tungsten Drell-Yan, with automatic uncertainties.","key_machinery":"The key object is the factorisation formula for asymmetric collisions (Eq. (7)), in which the two incoming hadrons $h_1$ and $h_2$ carry independent parton distributions. The extension leans on the existing NLO weight decomposition of MadGraph5_aMC@NLO, where each contribution to the cross section is written as a product of PDFs and a weight $W^{(\\alpha)}$ whose scale dependence is factored into coefficients; this lets the code reweight for different scales and PDFs without recomputing matrix elements. The practical modifications are a list of two or more LHAIDs in the run_card (the first is hadron $h_1$, the rest are $h_2$ variants), an `asymm_choice` flag to activate the mode, and a `cms_frame` flag that selects whether the cuts and output are in the laboratory or hadronic centre-of-mass frame.","core_discovery":"The central claim is that the collinear-factorisation machinery of MadGraph5_aMC@NLO can accommodate asymmetric hadron collisions with a minimal modification: the run_card accepts two distinct LHAPDF sets, one per incoming hadron, and the cross section is computed from the factorisation formula with $f_{a/h_1}$ and $f_{b/h_2}$, with the strong coupling $\\alpha_S$ taken from the second PDF set. The code then automatically produces both symmetric and asymmetric cross sections, with scale and PDF uncertainties, in the laboratory frame by default and optionally in the hadronic centre-of-mass frame. The validations—$W^+$ and $Z$ production in $p$Pb at 5.02 TeV against MCFM, and $\\pi^-W$ Drell-Yan at 21.7 GeV against the JAM prediction and E615 data—are reported as agreeing within uncertainties, which the paper takes as establishing the numerical correctness of the asymmetric implementation.","pith_inferences":["A natural test the authors did not report is to quantify how much the cross section changes when $\\alpha_S$ is taken from the first PDF set instead of the second; if that shift is sizable relative to the quoted uncertainties, the $\\alpha_S$-compatibility assumption would need a more careful treatment.","The automatic NMF computation from stored histograms could be extended to ratios between different nuclear species in a single run, which would make system-size scans (for example $p$Ne versus $p$Pb) straightforward for the LHC fixed-target programme.","If collinear factorization proves insufficient in the nuclear case, the two-PDF input interface built here is a natural platform for generalising the framework toward $k_T$-factorisation or other schemes, though that would require new subtraction machinery.","The validations cover electroweak-boson and Drell-Yan processes; applying the tool to heavy-flavour production at forward rapidities—where the gluon nPDF is poorly constrained—would be a stronger test of the implementation's numerical accuracy."],"forward_implications":["Heavy-ion analyses can obtain NLO QCD predictions for pA and AB reactions from a single public tool, with automatic scale and nuclear-PDF uncertainties, instead of maintaining private modified codes.","Nuclear modification factors for any process—including W, Z, charm, bottom, and associated $H+b\\bar b$ production—can be produced in one run, with the nPDF uncertainty propagated as recommended by the PDF sets.","Pion-beam phenomenology, such as Drell-Yan and heavy-flavour production used to constrain pion PDFs, becomes accessible at NLO in an automated generator.","Because the extension inherits the process automation of MadGraph5_aMC@NLO, it applies to any Standard Model or BSM process, not only the illustrative examples in the paper.","The laboratory-frame default with an optional centre-of-mass frame accommodates fixed-target and other asymmetric-energy configurations."],"supporting_citations":[{"why":"Supplies the base automated NLO framework that the extension modifies.","marker":"[70]"},{"why":"Provides the LHAPDF interface through which two different PDF sets and alpha_s are accessed.","marker":"[101]"},{"why":"Gives the NLO weight decomposition and reweighting scheme that enables automatic scale and PDF uncertainties.","marker":"[102]"},{"why":"Provides the MCFM reference results against which the pPb W/Z implementation is validated.","marker":"[75]"},{"why":"Supplies the nCTEQ15 nuclear PDFs used for the pPb validation and illustrations.","marker":"[3]"},{"why":"Supplies the EPPS16 nuclear PDFs used for the pPb validation and illustrations.","marker":"[5]"},{"why":"Supplies the JAM pion PDFs and the NLO Drell-Yan prediction used to validate pion-induced collisions.","marker":"[69]"},{"why":"Supplies the E615 pion-tungsten Drell-Yan data used as the experimental benchmark in the pion validation.","marker":"[113]"}],"fun_headline_variants":["MadGraph5 automates NLO QCD for asymmetric hadron collisions","Proton-nucleus and pion-tungsten NLO now in MadGraph5","NLO cross sections for any hadron pair, automated in MadGraph","Asymmetric hadron collisions get automated NLO in MadGraph5"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that collinear factorization remains valid when the two incoming hadrons are of different species, an assumption the paper itself flags as not formally proven for most asymmetric systems, with power-suppressed corrections believed to be enhanced.","fun_headline_variants_meta":{"raw":{"variants":["MadGraph5 automates NLO QCD for asymmetric hadron collisions","Proton-nucleus and pion-tungsten NLO now in MadGraph5","NLO cross sections for any hadron pair, automated in MadGraph","Asymmetric hadron collisions get automated NLO in MadGraph5"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000616,"raw_usage":{"total_tokens":2882,"prompt_tokens":991,"completion_tokens":1891,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":607,"completion_tokens_details":{"reasoning_tokens":1810}},"tokens_in":607,"tokens_out":1891,"duration_ms":15634,"temperature":1.0,"reasoning_tokens":1810,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T15:05:51.339953+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive check would be to run the code for a process in pPb or PbPb collisions where power corrections are expected to be large—for example forward $J/\\psi$ production, which the paper lists among processes sensitive to non-factorising effects—and compare the NLO prediction against LHC data; a systematic discrepancy growing toward forward rapidity or with nuclear size would signal that the factorisation assumption underlying the extension is breaking down. A second, more direct test of the implementation choices is to recompute the same asymmetric observable with the LHAID order reversed, so that $\\alpha_S$ is taken from the other PDF set, and check that the shift stays within the quoted uncertainties.","supporting_citations":[],"review_version":1}