Notes on recasting the ATLAS-EXOT-2019-23 search for pairs of displaced hadronic jets in the ATLAS calorimeter

Andreas Goudelis; Louie Corpe; Thomas Chehab

Notes on recasting the ATLAS-EXOT-2019-23 search for pairs of displaced hadronic jets in the ATLAS calorimeter

T0 review · 0 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper validates the efficiency map published with the EXOT-2019-23 search and shows that it reproduces the original limits for high-transverse-energy long-lived-particle models with lifetimes above about 50 cm.

desk verdict Honest, useful validation of an ATLAS efficiency map for LLP recasting, with clear documentation of where it works and where it breaks. read the letter →

arxiv 2412.13976 v1 pith:EH5GIILW submitted 2024-12-18 hep-ph hep-ex

Louie Corpe , Thomas Chehab , Andreas Goudelis This is my paper · ORCID

classification hep-phhep-ex

keywords long-livedparticlesdisplacedjetsrecastingefficiencymapRun-2datahadronisationhiddenAbelianHiggsmodel

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This note asks whether the reinterpretation material released with the EXOT-2019-23 search can substitute for a full detector-level analysis when one wants to estimate a new model's signal — a procedure often called recasting. The material is a six-dimensional efficiency map that takes truth-level decay positions, transverse momenta, and decay-product identities for a pair of long-lived particles and returns the probability that the event would pass the search's signal selection. The authors regenerate the benchmark signal, push the generated events through the map, and compare the resulting selection efficiencies and cross-section limits with the published values. Their central finding is that the procedure recovers the published results reasonably well for high-transverse-energy benchmarks, is usable by an external analyst, but becomes less reliable for low-energy, low-mass benchmarks and for lifetimes below roughly 50 cm. The note also documents gaps in the public documentation that made the validation harder than it needed to be.

What carries the argument

The central object is the efficiency map: a binned lookup table, provided with the search, that maps truth-level long-lived-particle properties — decay position (transverse in the barrel, longitudinal in the endcap), transverse momentum, and decay-product type — to the probability that a pair of such decays would be selected in the signal region. It folds in trigger, reconstruction, machine-learning discriminants, and all analysis selections. The validation machinery works by generating benchmark events, computing each decay's bin index, reading per-event selection probabilities from the map, summing them into a sample efficiency, and converting that efficiency into a cross-section limit with a single-bin signal-plus-background fit using the published region-A yields.

What would settle it

Regenerate one of the high-ET benchmark points with a different shower tune or a different hadronisation model and push the events through the same map; if the resulting cross-section limits move by more than the quoted map uncertainties (about 25% for efficiencies above 0.5%), the validation's agreement is tune-dependent rather than robust. A second test: generate a model with a deliberately different pseudorapidity distribution and compare map-based limits with a full detector-level recast; if they disagree by more than a factor of a few, the map's folded acceptance assumption fails.

Watch

Extended reading notes

Core claim

The paper's claim is that the published efficiency map is a serviceable recasting tool within a documented range, not a replacement for the full analysis everywhere. Concretely, for the high-ET selection the map-derived efficiencies and limits agree with the originally published curves once the map's stated uncertainties (about 25% for efficiencies above 0.5%) and validity limits are respected, with agreement best for large lifetime values. For the low-ET selection the procedure degrades: agreement is only to order of magnitude, hadronisation is essential, and one of the six benchmark points (mediator mass 60 GeV, LLP mass 5 GeV) yields efficiencies too low to be useful. The authors therefore recommend using the map only for lifetimes above about 50 cm, while noting that the map implicitly assumes new models resemble the training model in pseudorapidity distribution, tracklessness, and missing-hadronic-energy fraction.

Load-bearing premise

The whole validation rests on the assumption that truth-level kinematics from an external generator, after showering and hadronisation, match the generator-level distributions the original analysis used closely enough that the published efficiency map assigns the same selection probabilities; if the hadronisation model shifts jet momenta or decay positions differently from the original simulation, the agreement with the published limits could be a coincidence.

Editorial extensions

If this is right

External users can reinterpret the search for new long-lived-particle models using only truth-level generator output, without rerunning detector simulation or the full analysis.
The recasting is reliable for high-transverse-energy benchmarks with mediator masses at or above a few hundred GeV, provided lifetimes exceed about 50 cm.
For low-mass, low-energy models, parton-shower and hadronisation effects are essential; omitting them pushes jet transverse momenta below threshold and badly distorts efficiencies.
The map silently assumes new models pass the trackless-jet and missing-hadronic-energy side requirements, so those assumptions must be checked before trusting a limit.
The public record currently lacks enough generator-level documentation to make independent validation straightforward, and improving that documentation is essential for recasting.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A reasonable rule of thumb extending the paper's findings: do not quote a recast limit from this map for any model with mean proper lifetime below 50 cm, regardless of how well the high-lifetime part of the curve agrees.
The map's folded-acceptance assumption could be tested directly by generating a model with a deliberately different pseudorapidity distribution (for example, production via a vector-boson-fusion-like topology) and comparing map-based limits with a full-simulation recast; the paper identifies the assumption but does not quantify its failure.
If the same map format and validation procedure were applied to other searches, recasting could become a semi-automated check rather than a custom per-analysis exercise; the paper's wish list of standard formats points in that direction.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

This is a solid, honest recasting note. The new content is the validation itself: six benchmark points, public code, and specific findings about the map's domain of validity. For high-pT benchmarks the map reproduces the ATLAS limits well; for low-pT and cτ below about 50 cm it doesn't, and hadronisation matters a lot at low mass. Those observations go beyond the original ATLAS paper and are genuinely useful for the LLP recasting community.

What the paper does well: it ships code and data, which is the norm for recasting work and is done properly here. It is candid about the map's coarseness, the limited validation, and the ad-hoc map uncertainties. The list of obstacles for external users—inadequate MC documentation, missing model files, undocumented parameter choices—is a real contribution to the broader reproducibility discussion. The efficiency comparison figures are informative, and the physical explanation for why hadronisation is essential at low mediator mass (the pT threshold cannot be reached without shower recoil) is clear.

Soft spots: the validation uses the same HAHM benchmark model that produced the map, with a single generator chain (MG+Pythia8 default). The high-pT agreement could partly reflect internal consistency of the map rather than robustness to generator choices. A cross-generator or cross-tune test would settle that, but the paper's central claim is carefully scoped: it says the procedure works 'in a satisfactory manner' for the setup used, and it flags limitations. I don't think this is fatal—it's a stated boundary, not a hidden one. The limit curves are shown without uncertainty propagation, which is a minor presentation issue. Six benchmark points are few, but they span a reasonable mass and lifetime range. The unchecked assumption that u,d,s,gluon decays behave like c-quarks is explicitly flagged, so readers are not misled.

Who is this for? Phenomenologists doing LLP recasting, and anyone concerned with how experimental collaborations package auxiliary material for reinterpretation. It deserves serious peer review, not desk rejection. I would accept it and send to a knowledgeable referee, ideally one who has used the ATLAS map or built similar recasting tools. Recommended minor revisions: add a cross-generator test if feasible, or at least a paragraph on why it was not done, and propagate the map uncertainties to the limit curves.

Referee Report

0 major / 6 minor

Summary. This note validates the ATLAS-EXOT-2019-23 reinterpretation efficiency map, which maps truth-level LLP kinematics (decay position, transverse momentum, decay-product PDGIDs) to a signal-region selection probability. The authors generate HAHM events with MadGraph5_aMC@NLO and Pythia8 for six benchmark points, apply the published map, and compare the resulting efficiency curves and cross-section limits with the ATLAS publication. They find good agreement for high-ET benchmarks, degraded agreement for low-ET benchmarks, and they catalogue practical obstacles facing external users of the ATLAS material. The paper also provides open-source code implementing the recasting procedure.

Significance. If the map is reliable, it is a valuable and computationally cheap tool for reinterpretation of an important displaced-jet search, and this note is a useful independent test of that tool. The authors are candid about the limitations they find, and the code release is a concrete contribution that others can reuse. The validation is, however, a closure test against the same ATLAS results that motivated the map, using a single generator setup and a small number of benchmark points; the agreement shown is therefore not a fully independent check of the map's generality. The paper's central claim, that the procedure approximates the published results satisfactorily for high-ET benchmarks, is nevertheless supported by the comparisons shown.

minor comments (6)

[Section IV-B, Figs. 7-8] The limit curves in Figs. 7 and 8 are presented without propagating the map uncertainties quoted in Section II-B (25% for High-ET, 33% for Low-ET). Since the map accuracy is limited, the visual agreement in Fig. 7 may be partly fortuitous. Please either add uncertainty bands to the recast limits or explicitly state in the text and captions that these are central values only, so that readers can judge the significance of the agreement.
[Section III-B and Section VI] The validation uses only MadGraph5_aMC@NLO with Pythia8 at default settings and only the HAHM benchmark used by ATLAS. The paper should state more explicitly that the demonstrated agreement is a closure test for this particular generator setup, and that robustness to alternative shower tunes or generators has not been tested. This would prevent readers from over-generalizing the 'satisfactory' verdict beyond the tested conditions.
[Section V-A] In the bullet about hadronisation effects, 'Fig, 3' should be 'Fig. 3'.
[Figure 8 caption] The caption reads 'm_s = 100, 55 GeV', but Table I lists m_s = 50 GeV for m_phi = 200 GeV and m_s = 55 GeV for m_phi = 125 GeV. Please correct the inconsistency or clarify the intended values.
[Section II-A] The sentence 'As we can see in Fig. 2, the map is symmetric between the two LLPs, The choice of LLP 1 and 2 is arbitrary' contains a grammatical error; consider rewording to 'As we can see in Fig. 2, the map is symmetric between the two LLPs, so the choice of LLP 1 and 2 is arbitrary.'
[Table I] 'Nbr events' should be spelled out as 'Number of events' in the caption.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the recasting validation compares an externally published ATLAS efficiency map against externally published ATLAS results using independently generated events.

full rationale

The paper's central claim is that the ATLAS-EXOT-2019-23 efficiency map can be used by external users to approximate the published ATLAS efficiencies and limits. The validation is an external benchmark test: the map is taken from the HEPData record, the target efficiencies and limits are the published ATLAS values, and the event samples are independently generated with MadGraph5_aMC@NLO and Pythia8. No parameter is fitted to the ATLAS results; the map outputs per-event selection probabilities from truth-level bins, and the validation sums these probabilities to compare with the published curves. This is a consistency check rather than a derivation, because the map was not constructed by fitting the lifetime-dependent efficiency curves it is being tested against. The agreement for high-ET benchmarks is therefore nontrivial and could in principle fail. The only self-citation is a non-load-bearing footnote pointer to related work by the first author (Ref. [1]); it does not support any step of the derivation. The choice to use the same HAHM benchmark model as ATLAS is a model-dependence limitation, explicitly acknowledged by the authors, not a circular step. The paper also identifies several validation caveats, such as the region-A-only approximation, the eta-acceptance assumption, and the importance of hadronisation effects, further showing that the results are not forced by construction. No circular step can be quoted from the paper.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The paper introduces no free parameters or invented entities. Its central claim depends on the public ATLAS efficiency map, the HAHM benchmark model, and several transferability assumptions listed above. These assumptions are stated in the text but some (e.g., region A approximation, Pythia kinematics matching ATLAS simulation) are not independently verified.

assumptions (5)

domain assumption The efficiency map is model-independent: for given truth-level LLP kinematics (decay position, pT, decay products), the selection probability is the same regardless of the underlying model.
Core premise of the map, stated in Sec. II; it is the basis for applying the map to HAHM events generated by the authors.
domain assumption MadGraph5_aMC@NLO plus Pythia8 with default settings produces truth-level distributions sufficiently close to ATLAS's internal simulation for the map to be applied.
Event generation is described in Sec. III-B; no validation of these distributions against ATLAS internal simulation is provided, only comparisons of final efficiencies.
domain assumption A single-bin fit to region A alone is a valid approximation of the full ABCD simultaneous fit for signal efficiency and limit extraction.
Assumed in Sec. III-B and defended qualitatively in Sec. V-A without a proof or quantitative cross-check.
domain assumption The eta distribution of LLPs in any new model is similar to that in the HAHM benchmark; the map folds in detector acceptance effects.
Stated in Sec. II-A and V-A; the authors note this could cause overestimates for models with different eta distributions.
domain assumption All signal events satisfying the map's implicit selections (trackless jets, missing hadronic energy fraction) are representative of what would pass the full analysis.
The map assumes these selections are passed, as explained in Sec. II; the authors flag this as a caveat.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Notes on recasting the ATLAS-EXOT-2019-23 search for pairs of displaced hadronic jets in the ATLAS calorimeter." pith.science (2026). https://pith.science/paper/EH5GIILW

@misc{pith2026241213976,
  author       = {Pith},
  title        = {Pith review of: Notes on recasting the ATLAS-EXOT-2019-23 search for pairs of displaced hadronic jets in the ATLAS calorimeter},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EH5GIILW}},
  note         = {Machine review of arXiv:2412.13976}
}

read the original abstract

This note describes the validation of material allowing the reinterpretation of an ATLAS search for decays of pair-produced neutral long-lived particles decaying in the hadronic part of the calorimeter, or at the edge of the electromagnetic calorimeter, using the full Run-2 ATLAS dataset. This reinterpretation material includes an efficiency map linking truth-level kinematic information (decay position, transverse momentum and decay products of the LLPs) to the probability of the reconstructed event being selected in the analysis signal region. In this document we describe the validation procedure, i.e. how the map was used to recover the limits presented in the ATLAS publication using events generated with MadGraph5_aMC@NLO and hadronised using Pythia8, and we identify some limitations of this approach. We moreover comment upon issues concerning the validation procedure itself, in particular with regards to whether or not the information included in the existing, published material allows for an external user to test recasting methods.

Figures

Figures reproduced from arXiv: 2412.13976 by the authors.

**Figure 1.** Schematic diagram of the process Φ −→ ss −→ f f f ′ f ′ decay used as a benchmark model. The LLPs couple to SM fermions in a Yukawa-like manner, via their mixing with Φ, and therefore decay primarily to heavy quarks [2]. either towards the edge of the electromagnetic calorimeter (ECal) or in the hadronic calorimeter (HCal). In such cases, the LLP decay products can be reconstructed as a single jet which is typically… view at source ↗

**Figure 2.** The two re-interpretation maps provided by the ATLAS analysis [2]. The High-ET map is shown on top, the Low-ET map is shown at the bottom. The definition of the “Bin Index” is given in the main body of the text. The LLP transverse momentum (pT) is binned in the range of [0,50,100,200,400,1600] GeV (5 bins), the LLP transverse decay position Lx y is binned in [0,1.5,2,2.5,3,3.5,3.9,∞]m (7 bins), and the longitudinal … view at source ↗

**Figure 3.** Comparison of distribution for the transverse momenta without and with the hadronisation for a signal with low mediator mass [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Figures from the paper (5 more)

**Figure 4.** Figure 4: Comparison of distribution for the transverse momenta without and with the hadronisation for a signal with high mediator mass. each event and dividing by the total number of events, we obtain the sample’s selection efficiency. The code also produces limits: for this, a…

**Figure 6.** Figure 6: Efficiencies obtained for samples mφ = 200, 125 GeV and ms = 50, 55 GeV using the Low-ET selection, compared to the original ATLAS analysis. B. Cross section limits results As for the efficiencies, in order to validate the procedure we have plotted the limits that we c…

**Figure 5.** Figure 5: Efficiencies obtained for samples with mφ = 1000, 600, 400 GeV and ms = 275, 150, 100 GeV using the HighET selection, compared to the origin ATLAS analysis [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]

**Figure 8.** Figure 8: Limits for the samples mφ = 200, 125 GeV; ms = 100, 55 GeV using the Low-ET selection, compared to the original ATLAS results. A. Comments on the map Let us first make a few remarks concerning the reinterpretation map, the validity of some of the approximations it re…

**Figure 7.** Figure 7: Limits for the samples mφ = 1000, 600, 400 GeV and ms = 275, 150, 100 GeV using the High-ET selection, compared to the original ATLAS results [PITH_FULL_IMAGE:figures/full_fig_p006_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

10 extracted references · 4 canonical work pages

[1]

Probing exotic long-lived par- ticles from the prompt side using the CONTUR method

Louie Corpe et al. “Probing exotic long-lived par- ticles from the prompt side using the CONTUR method”. In: (July 2024). arXiv: 2407 . 18710 [hep-ph]

work page 2024
[2]

Search for neutral long-lived particles in pp collisions at ps = 13 TeV that decay into displaced hadronic jets in the ATLAS calorime- ter

Georges Aad et al. “Search for neutral long-lived particles in pp collisions at ps = 13 TeV that decay into displaced hadronic jets in the ATLAS calorime- ter”. In: JHEP 06 (2022), p. 005. DOI : 10 . 1007 / JHEP06(2022)005. arXiv: 2203.01009 [hep-ex]

arXiv 2022
[3]

https : / / www.hepdata.net/record/ins2043503

ATLAS-EXOT-2019-23 HEPdata record . https : / / www.hepdata.net/record/ins2043503. Accessed: 11 Jul 2023

work page 2019
[4]

https : / / github

ATLAS-EXOT-2019-23 Recasting Code . https : / / github . com / ThomasChehab / recastingCodes / tree / master / CalRatioDisplacedJet. Accessed: 27 Jul 2023

work page 2019
[5]

Higgs boson decays to four fermions through an abelian hidden sector

Shrihari Gopalakrishna, Sunghoon Jung, and James D. Wells. “Higgs boson decays to four fermions through an abelian hidden sector”. In: Phys. Rev. D 78 (2008), p. 055002. DOI : 10.1103/PhysRevD. 78.055002. arXiv: 0801.3456 [hep-ph]

arXiv 2008
[6]

How to Find a Hidden World at the Large Hadron Collider

James D. Wells. “How to Find a Hidden World at the Large Hadron Collider”. In: (Mar. 2008). Ed. by Gordon Kane and Aaron Pierce, pp. 283–298. arXiv: 0803.1243 [hep-ph] . 7

arXiv 2008
[7]

FeynRules 2.0 - A complete toolbox for tree-level phenomenology

Adam Alloul et al. “FeynRules 2.0 - A complete toolbox for tree-level phenomenology”. In: Comput. Phys. Commun. 185 (2014), pp. 2250–2300. DOI : 10. 1016 / j . cpc . 2014 . 04 . 012. arXiv: 1310 . 1921 [hep-ph]

work page 2014
[8]

Illuminating Dark Photons with High-Energy Colliders

David Curtin et al. “Illuminating Dark Photons with High-Energy Colliders”. In: JHEP 02 (2015), p. 157. DOI : 10.1007/JHEP02(2015)157. arXiv: 1412.0018 [hep-ph]

arXiv 2015

Show all 10 references

[9]

The automated computation of tree- level and next-to-leading order differential cross sections, and their matching to parton shower sim- ulations

J. Alwall et al. “The automated computation of tree- level and next-to-leading order differential cross sections, and their matching to parton shower sim- ulations”. In: JHEP 07 (2014), p. 079. DOI : 10.1007/ JHEP07(2014)079. arXiv: 1405.0301 [hep-ph]

2014 arXiv
[10]

A comprehensive guide to the physics and usage of PYTHIA 8.3

Christian Bierlich et al. “A comprehensive guide to the physics and usage of PYTHIA 8.3”. In: (Mar. 2022). DOI : 10.21468/SciPostPhysCodeb.8. arXiv: 2203.11601 [hep-ph] . 8

2022 arXiv

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.

[1] [1]

Probing exotic long-lived par- ticles from the prompt side using the CONTUR method

Louie Corpe et al. “Probing exotic long-lived par- ticles from the prompt side using the CONTUR method”. In: (July 2024). arXiv: 2407 . 18710 [hep-ph]

work page 2024

[2] [2]

Search for neutral long-lived particles in pp collisions at ps = 13 TeV that decay into displaced hadronic jets in the ATLAS calorime- ter

Georges Aad et al. “Search for neutral long-lived particles in pp collisions at ps = 13 TeV that decay into displaced hadronic jets in the ATLAS calorime- ter”. In: JHEP 06 (2022), p. 005. DOI : 10 . 1007 / JHEP06(2022)005. arXiv: 2203.01009 [hep-ex]

arXiv 2022

[3] [3]

https : / / www.hepdata.net/record/ins2043503

ATLAS-EXOT-2019-23 HEPdata record . https : / / www.hepdata.net/record/ins2043503. Accessed: 11 Jul 2023

work page 2019

[4] [4]

https : / / github

ATLAS-EXOT-2019-23 Recasting Code . https : / / github . com / ThomasChehab / recastingCodes / tree / master / CalRatioDisplacedJet. Accessed: 27 Jul 2023

work page 2019

[5] [5]

Higgs boson decays to four fermions through an abelian hidden sector

Shrihari Gopalakrishna, Sunghoon Jung, and James D. Wells. “Higgs boson decays to four fermions through an abelian hidden sector”. In: Phys. Rev. D 78 (2008), p. 055002. DOI : 10.1103/PhysRevD. 78.055002. arXiv: 0801.3456 [hep-ph]

arXiv 2008

[6] [6]

How to Find a Hidden World at the Large Hadron Collider

James D. Wells. “How to Find a Hidden World at the Large Hadron Collider”. In: (Mar. 2008). Ed. by Gordon Kane and Aaron Pierce, pp. 283–298. arXiv: 0803.1243 [hep-ph] . 7

arXiv 2008

[7] [7]

FeynRules 2.0 - A complete toolbox for tree-level phenomenology

Adam Alloul et al. “FeynRules 2.0 - A complete toolbox for tree-level phenomenology”. In: Comput. Phys. Commun. 185 (2014), pp. 2250–2300. DOI : 10. 1016 / j . cpc . 2014 . 04 . 012. arXiv: 1310 . 1921 [hep-ph]

work page 2014

[8] [8]

Illuminating Dark Photons with High-Energy Colliders

David Curtin et al. “Illuminating Dark Photons with High-Energy Colliders”. In: JHEP 02 (2015), p. 157. DOI : 10.1007/JHEP02(2015)157. arXiv: 1412.0018 [hep-ph]

arXiv 2015