arxiv: 2511.04005 · v3 · submitted 2025-11-06 · ✦ hep-ph · nucl-ex

Validating a Machine Learning Approach to Identify Quenched Jets in Heavy-Ion Collisions

Yilun Wu , Yi Chen , Julia Velkovska This is my paper

Pith reviewed 2026-05-18 01:34 UTC · model grok-4.3

classification ✦ hep-ph nucl-ex

keywords jet quenchingheavy-ion collisionsmachine learningLSTMjet substructurequark-gluon plasmaDelphesphoton-jet

0 comments

The pith

An LSTM neural network trained on jet substructure predicts jet energy loss due to the quark-gluon plasma and works after detector simulation.

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

The paper validates a machine learning approach to identify quenched jets on a jet-by-jet basis in heavy-ion collisions. An LSTM network trained on jet substructure that includes parton shower history from the Jewel generator produces predictions that correlate with the actual energy each jet loses to the quark-gluon plasma. The correlation persists after Delphes simulates detector effects, and the model distinguishes quenching signatures in photon-jet momentum imbalance, fragmentation functions, and jet shapes that were withheld from training.

Core claim

Using photon-jet samples from the Jewel event generator, the LSTM predictions strongly correlate with true jet energy loss. This validates that the model effectively learns the features of jet-QGP interaction. We simulate detector effects using Delphes simulation framework and demonstrate that the method identifies quenching effects in a realistic environment. We test the approach with photon-jet momentum imbalance, jet fragmentation function, and jet shape, which were not included in the training, confirming its ability to distinguish true quenching features.

What carries the argument

LSTM neural network trained on jet substructure incorporating parton shower history to predict jet-by-jet quenching levels.

If this is right

LSTM predictions strongly correlate with true jet energy loss from the Jewel simulation.
The identification of quenching effects remains effective after Delphes detector simulation.
The model distinguishes true quenching in photon-jet momentum imbalance, jet fragmentation function, and jet shape without training on those observables.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The trained model could be applied directly to experimental heavy-ion data to tag quenched jets for targeted analysis of medium properties.
Jet-by-jet quenching estimates might tighten constraints on energy-loss models when combined with existing observables.
Retraining or testing the same architecture on other generators would check whether the learned features are generator-independent.

Load-bearing premise

Features learned from Jewel parton-shower history without detector effects will generalize to Delphes-simulated data and correctly isolate true quenching from other physics and detector effects on unseen observables.

What would settle it

If the predicted quenching levels show no correlation with actual energy loss or fail to separate quenching signals in the jet shape distribution on an independent test set that includes Delphes effects, the claim that the model identifies genuine quenching would not hold.

read the original abstract

Jet quenching is a phenomenon in heavy-ion collisions arising from jet interactions with the quark-gluon plasma (QGP). Its study is complicated by the interplay of multiple physics processes that affect jet observables. In addition, detector effects may influence the results and must be accounted for when identifying quenched jets. We employ a Long Short-Term Memory (LSTM) neural network trained on jet substructure, incorporating parton shower history, to predict jet-by-jet quenching levels. Using photon-jet samples from the \textsc{Jewel} event generator, we show that the LSTM predictions strongly correlate with true jet energy loss. This validates that the model effectively learns the features of jet-QGP interaction. We simulate detector effects using \textsc{Delphes} simulation framework and demonstrate that the method identifies quenching effects in a realistic environment. We test the approach with photon-jet momentum imbalance, jet fragmentation function, and jet shape, which were not included in the training, confirming its ability to distinguish true quenching features.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

LSTM trained on Jewel parton-shower labels predicts per-jet quenching and holds up on held-out observables after Delphes, but the writeup gives almost no numbers or controls.

read the letter

The paper trains an LSTM on jet substructure features from Jewel, using the parton-shower history to supply direct labels for how much energy a jet lost to the medium. It then runs the same model on photon-jet events after Delphes detector simulation and checks whether the predictions still separate quenched from unquenched jets on momentum imbalance, fragmentation functions, and jet shapes that were never shown during training. That choice of held-out observables is the clearest strength; it supplies an external check instead of just re-testing the training labels. Adding the detector step is also the right direction if the goal is a tool that could be used on real data. The approach is straightforward and the validation strategy makes sense for this subfield. The main gaps are quantitative. The abstract claims strong correlation and successful distinction but reports no coefficient, no uncertainty, no baseline comparison, and no training hyperparameters. On the Delphes side, the same substructure variables that feed the network are altered by the detector response, so any remaining separation could partly reflect how Delphes distorts those inputs rather than learned QGP physics. A control sample with no medium but with Delphes would clarify this. For experimentalists looking for a practical tagger this is worth trying out. It is not a major theoretical step, but the idea is grounded and the test design is reasonable. I would send it to referees and ask for the missing metrics plus the detector-only control.

Referee Report

2 major / 2 minor

Summary. The manuscript proposes training an LSTM neural network on jet substructure observables extracted from Jewel simulations of photon-jet events in heavy-ion collisions, using parton-shower history to supply per-jet quenching labels. The central claim is that the resulting model predictions correlate strongly with true jet energy loss, remain effective after Delphes detector simulation, and successfully distinguish quenching signatures on three observables (photon-jet momentum imbalance, fragmentation functions, and jet shapes) that were withheld from training.

Significance. If the quantitative validation holds, the work would supply a jet-by-jet tagger for quenched jets that is grounded in simulated truth labels yet tested on independent observables. This could help separate genuine QGP-induced energy loss from detector smearing and other backgrounds in future heavy-ion analyses. The post-Delphes test on unseen observables is a constructive design choice that supplies external grounding.

major comments (2)

[Abstract] Abstract: the statement that 'LSTM predictions strongly correlate with true jet energy loss' is presented without any reported correlation coefficient, R² value, mean absolute error, or uncertainty bands. Because this correlation is the primary evidence that the network has learned jet-QGP interaction features, the absence of these metrics is load-bearing for the central claim.
[Results] Results section (post-Delphes validation): no quantitative comparison is shown between the LSTM-energy-loss correlation obtained before and after Delphes smearing, nor is there a control test that subtracts the correlation arising from detector effects alone on the withheld observables (momentum imbalance, fragmentation function, jet shape). This leaves open the possibility that residual correlations are driven by shared substructure variables rather than learned quenching physics.

minor comments (2)

[Methods] The manuscript should report the size of the training sample, the precise LSTM architecture (number of layers, hidden units, dropout), and the loss function used for regression on energy loss.
[Figures] Figure captions and axis labels for any correlation or separation plots should explicitly state whether the data are pre- or post-Delphes and whether the plotted observable was used in training.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We are grateful to the referee for their thorough review and valuable suggestions. Below we provide point-by-point responses to the major comments and describe the changes we plan to implement in the revised manuscript.

read point-by-point responses

Referee: [Abstract] Abstract: the statement that 'LSTM predictions strongly correlate with true jet energy loss' is presented without any reported correlation coefficient, R² value, mean absolute error, or uncertainty bands. Because this correlation is the primary evidence that the network has learned jet-QGP interaction features, the absence of these metrics is load-bearing for the central claim.

Authors: We concur that quantitative metrics are essential to substantiate this key claim. In the revised version of the manuscript, we will explicitly report the Pearson correlation coefficient, R² value, mean absolute error, and associated uncertainty bands both in the abstract and in the main text where the LSTM predictions are compared to the true jet energy loss from the Jewel simulation. revision: yes
Referee: [Results] Results section (post-Delphes validation): no quantitative comparison is shown between the LSTM-energy-loss correlation obtained before and after Delphes smearing, nor is there a control test that subtracts the correlation arising from detector effects alone on the withheld observables (momentum imbalance, fragmentation function, jet shape). This leaves open the possibility that residual correlations are driven by shared substructure variables rather than learned quenching physics.

Authors: We appreciate this observation. While the post-Delphes performance on independent observables already provides evidence that the model captures quenching physics beyond detector effects, we agree that a direct quantitative comparison would be beneficial. We will add plots and metrics comparing the LSTM-true energy loss correlation before and after Delphes in the revised results section. Additionally, we will include a discussion or supplementary control analysis to evaluate the contribution from detector smearing on the withheld observables. revision: yes

Circularity Check

0 steps flagged

No significant circularity; validation grounded in independent observables

full rationale

The paper trains an LSTM on Jewel parton-shower substructure features with quenching labels derived from simulation history, then reports correlation of predictions with true energy loss and applies the model after Delphes detector simulation. It explicitly tests generalization on photon-jet momentum imbalance, fragmentation functions, and jet shapes that were withheld from training. This constitutes external grounding against held-out observables and detector response rather than any self-definitional loop, fitted parameter renamed as prediction, or load-bearing self-citation. The central claim therefore retains independent content from the simulation benchmarks and does not reduce to its inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The claim rests on the fidelity of the Jewel generator for providing ground-truth quenching labels and on the assumption that learned substructure features transfer across detector simulation; no new particles or forces are postulated.

axioms (1)

domain assumption Jewel event generator provides sufficiently accurate modeling of jet-QGP interactions to serve as ground truth for supervised training.
All training labels and correlation targets derive from Jewel simulations.

pith-pipeline@v0.9.0 · 5704 in / 1215 out tokens · 30657 ms · 2026-05-18T01:34:36.204240+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

We employ a Long Short-Term Memory (LSTM) neural network trained on jet substructure, incorporating parton shower history, to predict jet-by-jet quenching levels... We test the approach with photon-jet momentum imbalance, jet fragmentation function, and jet shape, which were not included in the training

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

59 extracted references · 59 canonical work pages · 29 internal anchors

[1]

J. D. Bjorken,Energy Loss of Energetic Partons in Quark - Gluon Plasma: Possible Extinction of High p(t) Jets in Hadron - Hadron Collisions,

work page
[2]

PHENIX Collaboration,Suppression of hadrons with large transverse momentum in central Au+Au collisions at√sNN = 130 GeV,Phys. Rev. Lett.88(2002) 022301, [nucl-ex/0109003]

work page internal anchor Pith review Pith/arXiv arXiv 2002
[3]

STAR Collaboration,Centrality dependence of highpT hadron suppression in Au+Au collisions at √sN N = 130-GeV,Phys. Rev. Lett.89(2002) 202301, [nucl-ex/0206011]

work page internal anchor Pith review Pith/arXiv arXiv 2002
[4]

PHENIX Collaboration,Suppressedπ 0 production at large transverse momentum in central Au+Au collisions at√sNN = 200 GeV,Phys. Rev. Lett.91(2003) 072301, [nucl-ex/0304022]

work page internal anchor Pith review Pith/arXiv arXiv 2003
[5]

STAR Collaboration,Transverse momentum and collision energy dependence of highpT hadron suppression in Au+Au collisions at ultrarelativistic energies,Phys. Rev. Lett.91 (2003) 172302, [nucl-ex/0305015]

work page internal anchor Pith review Pith/arXiv arXiv 2003
[6]

STAR Collaboration,Disappearance of back-to-back highpT hadron correlations in central Au+Au collisions at√sN N = 200-GeV,Phys. Rev. Lett.90(2003) 082302, [nucl-ex/0210033]

work page internal anchor Pith review Pith/arXiv arXiv 2003
[7]

ALICE Collaboration,Suppression of charged particle production at large transverse momentum in central Pb–Pb collisions at√sNN = 2.76TeV,Phys. Lett. B696(2011) 30, [1012.1004]

work page internal anchor Pith review Pith/arXiv arXiv 2011
[8]

ALICE Collaboration,Particle-yield modification in jetlike azimuthal dihadron correlations in Pb-Pb collisions at√sNN = 2.76TeV,Phys. Rev. Lett.108(2012) 092301, [1110.0121]

work page internal anchor Pith review Pith/arXiv arXiv 2012
[10]

ATLAS Collaboration,Measurement of charged-particle spectra in Pb+Pb collisions at√sNN = 2.76TeV with the ATLAS detector at the LHC,JHEP09(2015) 050, [1504.04337]

work page internal anchor Pith review Pith/arXiv arXiv 2015
[11]

CMS Collaboration,Study of high-pt charged particle suppression in pbpb compared to pp collisions at √sNN = 2.76 TeV,The European Physical Journal C72(2012) , [1202.2554]

work page internal anchor Pith review Pith/arXiv arXiv 2012
[12]

ATLAS Collaboration,Measurement of the jet radius and transverse momentum dependence of inclusive jet suppression in lead-lead collisions at√sNN = 2.76TeV with the ATLAS detector,Phys. Lett. B719(2013) 220, [1208.1967]

work page internal anchor Pith review Pith/arXiv arXiv 2013
[13]

ALICE Collaboration,Measurement of charged jet suppression in Pb-Pb collisions at√sNN = 2.76 TeV,JHEP03(2014) 013, [1311.0633]

work page internal anchor Pith review Pith/arXiv arXiv 2014
[14]

ALICE Collaboration,Measurement of jet suppression in central Pb-Pb collisions at√sNN = 2.76 TeV,Phys. Lett. B746(2015) 1, [1502.01689]

work page internal anchor Pith review Pith/arXiv arXiv 2015
[15]

CMS Collaboration,Evidence of b-Jet Quenching in PbPb Collisions at√sN N = 2.76TeV, Phys. Rev. Lett.113(2014) 132301, [1312.4198]

work page internal anchor Pith review Pith/arXiv arXiv 2014
[16]

ATLAS Collaboration,Comparison of inclusive and photon-tagged jet suppression in 5.02 TeV Pb+Pb collisions with ATLAS,Phys. Lett. B846(2023) 138154, [2303.10090]

work page arXiv 2023
[17]

ATLAS Collaboration,Measurement of the nuclear modification factor of b-jets in 5.02 TeV Pb+Pb collisions with the ATLAS detector,Eur. Phys. J. C 7283(2023) , [2204.13530]. – 15 –

work page arXiv 2023
[18]

ATLAS Collaboration,Measurements of azimuthal anisotropies of jet production inPb + Pb collisions at √sNN = 5.02TeV with the ATLAS detector,Phys. Rev. C105(2022) 064903, [2111.06606]

work page arXiv 2022
[19]

CMS Collaboration,Azimuthal anisotropy of dijet events in PbPb collisions at√sNN = 5.02 TeV,JHEP2023(2023) , [2210.08325v2]

work page arXiv 2023
[20]

Cumulants and nonlinear response of high $p_T$ harmonic flow at $\sqrt{s_{NN}}=5.02$ TeV

J. Noronha-Hostler, B. Betz, M. Gyulassy, M. Luzum, J. Noronha, I. Portillo et al., Cumulants and nonlinear response of highpT harmonic flow at√sNN = 5.02TeV,Phys. Rev. C95(2017) 044901, [1609.05171v2]

work page internal anchor Pith review Pith/arXiv arXiv 2017
[21]

Holtermann, J

A. Holtermann, J. Noronha-Hostler, A. M. Sickles and X. Wang,Multiparticle correlations, cumulants, and moments sensitive to fluctuations in rare-probe azimuthal anisotropy in heavy ion collisions,Phys. Rev. C108(2023) 064901, [2307.16796v2]

work page arXiv 2023
[22]

CMS Collaboration,Modification of jet shapes in PbPb collisions at√sNN = 2.76TeV,Phys. Lett. B730(2014) 243, [1310.0878]

work page internal anchor Pith review Pith/arXiv arXiv 2014
[23]

ALICE Collaboration,Measurement of jet radial profiles in Pb˘Pb collisions at√sNN =2.76 TeV,Phys. Lett. B796(2019) 204–219, [1904.13118]

work page arXiv 2019
[24]

ATLAS Collaboration,Measurement of angular and momentum distributions of charged particles within and around jets in Pb+Pb andppcollisions at√sNN = 5.02TeV with the ATLAS detector,Phys. Rev. C100(2019) 064901, [1908.05264]

work page arXiv 2019
[25]

CMS Collaboration,Measurement of jet fragmentation in PbPb and pp collisions at√sNN = 2.76TeV,Phys. Rev. C90(2014) 024908, [1406.0932]

work page internal anchor Pith review Pith/arXiv arXiv 2014
[26]

ATLAS Collaboration,Measurement of jet fragmentation in Pb+Pb and pp collisions at√sNN = 2.76TeV with the ATLAS detector at the LHC,Eur. Phys. J. C77(2017) 379, [1702.00674]

work page internal anchor Pith review Pith/arXiv arXiv 2017
[27]

CMS Collaboration,Measurement of the Splitting Function inppand Pb-Pb Collisions at√sNN =5.02 TeV,Phys. Rev. Lett.120(2018) 142302, [1708.09429]

work page internal anchor Pith review Pith/arXiv arXiv 2018
[28]

STAR Collaboration,Differential measurements of jet substructure and partonic energy loss in Au+Au collisions at√SN N =200 GeV,Phys. Rev. C105(2022) 044906, [2109.09793]

work page arXiv 2022
[29]

ALICE Collaboration,Measurement of the groomed jet radius and momentum splitting fraction in pp and Pb−Pb collisions at√sN N = 5.02TeV,Phys. Rev. Lett.128(2022) 102001, [2107.12984]

work page arXiv 2022
[30]

CMS Collaboration,Measurement of the groomed jet mass in PbPb and pp collisions at√sNN =5.02 TeV,JHEP2018(2018) , [1805.05145]

work page internal anchor Pith review Pith/arXiv arXiv 2018
[31]

ATLAS Collaboration,Measurement of substructure-dependent jet suppression inPb + Pb collisions at 5.02 TeV with the ATLAS detector,Phys. Rev. C107(2023) 054909, [2211.11470v2]

work page arXiv 2023
[32]

CMS Collaboration,Energy-energy correlators from PbPb and pp collisions at 5.02 TeV, 2024, https://cds.cern.ch/record/2906425/files/HIN-23-004-pas.pdf

work page arXiv 2024
[33]

CMS Collaboration,Girth and groomed radius of jets recoiling against isolated photons in lead-lead and proton-proton collisions at√sN N = 5.02tev,Physics Letters B861(2025) 139088, [2405.02737]

work page arXiv 2025
[34]

Review of Jet Measurements in Heavy Ion Collisions

M. Connors, C. Nattrass, R. Reed and S. Salur,Jet measurements in heavy ion physics, Reviews of Modern Physics90(2018) , [1705.01974v2]. – 16 –

work page internal anchor Pith review Pith/arXiv arXiv 2018
[35]

Mulligan and M

J. Mulligan and M. Ploskon,Identifying groomed jet splittings in heavy-ion collisions,Phys. Rev. C102(2020) 044913, [2006.01812]

work page arXiv 2020
[36]

Ringer, B.-W

F. Ringer, B.-W. Xiao and F. Yuan,Can we observe jet P-broadening in heavy-ion collisions at the LHC?,Phys. Lett. B808(Sept., 2020) 135634, [1907.12541]

work page arXiv 2020
[37]

Apolinário, R

L. Apolinário, R. K. Elayavalli, N. O. Madureira, J.-X. Sheng, X.-N. Wang and Z. Yang, Flavor dependence of Energy-energy correlators,2502.11406

work page arXiv
[38]

Apolinário, N

L. Apolinário, N. F. Castro, M. Crispim Romão, J. G. Milhano, R. Pedro and F. C. R. Peres, Deep Learning for the classification of quenched jets,JHEP11(2021) 219, [2106.08869]

work page arXiv 2021
[39]

Y.-L. Du, D. Pablos and K. Tywoniuk,Deep learning jet modifications in heavy-ion collisions,JHEP21(2020) 206, [2012.07797]

work page arXiv 2020
[40]

Y. S. Lai, J. Mulligan, M. Płoskoń and F. Ringer,The information content of jet quenching and machine learning assisted observable design,JHEP2022(2022) 11, [2111.14589]

work page arXiv 2022
[41]

Crispim Romão, J

M. Crispim Romão, J. Guilherme Milhano and M. van Leeuwen,Jet substructure observables for jet quenching in quark gluon plasma: A machine learning driven analysis,SciPost Physics16(2024) , [2304.07196v2]

work page arXiv 2024
[42]

L. Liu, J. Velkovska, Y. Wu and M. Verweij,Identifying quenched jets in heavy ion collisions with machine learning,JHEP2023(2023)

work page 2023
[43]

J. a. A. Gonçalves and J. G. Milhano,Apples to Apples in Jet Quenching: robustness of Machine Learning classification of quenched jets to Underlying Event contamination, 2501.14015

work page arXiv
[44]

U. S. Qureshi and R. K. Elayavalli,Model-agnostic tagging of quenched jets in heavy-ion collisions,2411.19389

work page arXiv
[45]

Hochreiter and J

S. Hochreiter and J. Schmidhuber,Long short-term memory,Neural computation9(1997) 1735–1780

work page 1997
[46]

Sherstinsky,Fundamentals of recurrent neural network (rnn) and long short-term memory (lstm) network,Physica D: Nonlinear Phenomena404(2020) 132306

A. Sherstinsky,Fundamentals of recurrent neural network (rnn) and long short-term memory (lstm) network,Physica D: Nonlinear Phenomena404(2020) 132306

work page 2020
[47]

K. C. Zapp, F. Krauss and U. A. Wiedemann,A perturbative framework for jet quenching, JHEP03(2013) 080, [1212.1599]

work page internal anchor Pith review Pith/arXiv arXiv 2013
[48]

Medium response in JEWEL and its impact on jet shape observables in heavy ion collisions

R. Kunnawalkam Elayavalli and K. C. Zapp,Medium response in JEWEL and its impact on jet shape observables in heavy ion collisions,JHEP07(2017) 141, [1707.01539]

work page internal anchor Pith review Pith/arXiv arXiv 2017
[49]

The Angantyr model for Heavy-Ion Collisions in PYTHIA8

C. Bierlich, G. Gustafson, L. Lönnblad and H. Shah,The angantyr model for heavy-ion collisions in pythia8,JHEP2018(2018) , [1806.10820v2]

work page internal anchor Pith review Pith/arXiv arXiv 2018
[50]

A comprehensive guide to the physics and usage of PYTHIA 8.3

C. Bierlich, S. Chakraborty, N. Desai, L. Gellersen, I. Helenius, P. Ilten et al.,A comprehensive guide to the physics and usage of PYTHIA 8.3,SciPost Phys. Codebases (2022) 8, [2203.11601v1]

work page internal anchor Pith review Pith/arXiv arXiv 2022
[51]

CMS Collaboration,Overview of high-density QCD studies with the CMS experiment at the LHC,2405.10785

work page arXiv
[52]

Liu,Github repository for jet quenching studies with machine learning, 2022

L. Liu,Github repository for jet quenching studies with machine learning, 2022

work page 2022
[53]

Y. L. Dokshitzer, G. D. Leder, S. Moretti and B. R. Webber,Better jet clustering algorithms, JHEP08(1997) 001, [hep-ph/9707323]. – 17 –

work page internal anchor Pith review Pith/arXiv arXiv 1997
[54]

A. J. Larkoski, S. Marzani, G. Soyez and J. Thaler,Soft drop,JHEP05(2014) 146, [1402.2657]

work page internal anchor Pith review Pith/arXiv arXiv 2014
[55]

DELPHES 3, A modular framework for fast simulation of a generic collider experiment

J. de Favereau, C. Delaere, P. Demin, A. Giammanco, V. Lemaître, A. Mertens et al., Delphes 3: a modular framework for fast simulation of a generic collider experiment,JHEP 2014(2014) , [1307.6346v3]

work page internal anchor Pith review Pith/arXiv arXiv 2014
[56]

rep., CERN, Geneva, 2009

CMS Collaboration,Particle-Flow Event Reconstruction in CMS and Performance for Jets, Taus, and MET, tech. rep., CERN, Geneva, 2009

work page 2009
[57]

The anti-k_t jet clustering algorithm

M. Cacciari, G. P. Salam and G. Soyez,The anti-kt jet clustering algorithm,JHEP04 (2008) 063, [0802.1189]

work page internal anchor Pith review Pith/arXiv arXiv 2008
[58]

Particle-level pileup subtraction for jets and jet shapes

P. Berta, M. Spousta, D. W. Miller and R. Leitner,Particle-level pileup subtraction for jets and jet shapes,JHEP06(2014) 092, [1403.3108]

work page internal anchor Pith review Pith/arXiv arXiv 2014
[59]

Paszke, S

A. Paszke, S. Gross, F. Massa, A. Lerer, J. Bradbury, G. Chanan et al.,Pytorch: An imperative style, high-performance deep learning library, inAdvances in Neural Information Processing Systems 32(H. Wallach, H. Larochelle, A. Beygelzimer, F. d'Alché-Buc, E. Fox and R. Garnett, eds.), pp. 8024–8035. Curran Associates, Inc., 2019

work page 2019
[60]

Bergstra, D

J. Bergstra, D. Yamins and D. Cox,Making a science of model search: Hyperparameter optimization in hundreds of dimensions for vision architectures, inProceedings of the 30th International Conference on Machine Learning(S. Dasgupta and D. McAllester, eds.), vol. 28 ofProceedings of Machine Learning Research, (Atlanta, Georgia, USA), pp. 115–123, PMLR, 17–1...

work page 2013