arxiv: 2605.09022 · v1 · submitted 2026-05-09 · ✦ hep-ph · hep-ex· hep-th· nucl-ex· nucl-th

Recognition: no theorem link

Inferring identified hadron production in pp collisions with physics-informed machine learning at the LHC

Rishabh Gupta , Kangkan Goswami , Suraj Prasad , Raghunath Sahoo

Authors on Pith no claims yet

Pith reviewed 2026-05-12 02:13 UTC · model grok-4.3

classification ✦ hep-ph hep-exhep-thnucl-exnucl-th

keywords physics-informed neural networkhadron spectrapp collisionsLHCrapidity extrapolationparticle productionmachine learning

0 comments

The pith

A physics-informed neural network infers identified hadron pT spectra in unmeasured rapidity regions from LHC simulation data.

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

The paper trains a neural network exclusively on PYTHIA8 Monte Carlo events for 13.6 TeV proton-proton collisions to predict transverse-momentum spectra of pions, kaons, protons, and lambdas over a broader rapidity range than detectors can access. Additional terms in the loss function enforce physical relations such as particle yield ratios, expected spectral shapes, and smoothness in momentum. The resulting model reports yield uncertainties of roughly 1.5 percent inside the training domain, 1.8 percent for interpolation, and 5.83 percent for extrapolation while also matching multiplicity dependence of mean transverse momentum and kinetic freeze-out parameters. A sympathetic reader would care because forward-rapidity data are otherwise inaccessible, so a reliable inference tool could complete the kinematic picture of hadron production without new detector coverage.

Core claim

The central claim is that a neural network trained solely on PYTHIA8 pp collisions at 13.6 TeV, with physics-motivated constraints on particle yield ratios, spectral shape, and smoothness added to its loss function, can infer pT spectra for pi^pm, K^pm, p/pbar, Lambda/Lambdabar, and K0s across different rapidity regions, achieving the quoted uncertainties in training, interpolation, and extrapolation regimes while reproducing key observables such as yield ratios and multiplicity dependence of mean pT.

What carries the argument

Physics-informed neural network whose loss function augments the standard error with explicit terms for particle yield ratios, spectral shape, and smoothness regularization.

If this is right

The inferred spectra can supply hadron yields in forward and backward rapidity regions that lie outside current detector acceptance.
The model reproduces multiplicity dependence of average transverse momentum and kinetic freeze-out parameters, indicating it has captured essential features of the production process.
The same framework outperforms standard gradient-boosting regressors such as XGBoost and LightGBM on this regression task.
Predictions can be generated for any rapidity slice once the network is trained, enabling studies of full-phase-space particle production without additional data collection.

Where Pith is reading between the lines

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

If the approach transfers to real collision data, it could allow experiments to reduce reliance on wide-acceptance detectors for certain observables.
The method offers a template for embedding known physical relations directly into machine-learning pipelines for other high-energy regression problems.
Retraining the same architecture on simulations that include detector effects might test whether the constraints also mitigate differences between Monte Carlo and data.

Load-bearing premise

That the physics constraints in the loss function plus training only on PYTHIA8 are sufficient for the network to generalize correctly to real experimental data and unmeasured rapidity regions rather than learning simulation-specific artifacts.

What would settle it

Direct comparison of the model's extrapolated yields against measured data in a forward-rapidity bin that was deliberately withheld from training would show whether the reported uncertainties hold or whether systematic deviations exceed them.

Figures

Figures reproduced from arXiv: 2605.09022 by Kangkan Goswami, Raghunath Sahoo, Rishabh Gupta, Suraj Prasad.

**Figure 1.** Figure 1: FIG. 1: Machine learning pipeline including dataset preparation, staged optimization, and physics-informed neural [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗

**Figure 2.** Figure 2: FIG. 2: Relation between the true and predicted output for the trained (left), interpolation (center), and [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗

**Figure 3.** Figure 3: FIG. 3: Comparison of extrapolation yield error across [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

**Figure 4.** Figure 4: FIG. 4: Extrapolation yield error as a function of [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗

**Figure 5.** Figure 5: FIG. 5: PINN predictions across the trained, interpolation, and extrapolation regions for different multiplicity [PITH_FULL_IMAGE:figures/full_fig_p010_5.png] view at source ↗

**Figure 6.** Figure 6: FIG. 6: Comparison of particle ratio predictions with PYTHIA8 results in the interpolation and extrapolation [PITH_FULL_IMAGE:figures/full_fig_p011_6.png] view at source ↗

**Figure 7.** Figure 7: FIG. 7: Mean transverse momentum as a function of charged-particle multiplicity [PITH_FULL_IMAGE:figures/full_fig_p011_7.png] view at source ↗

**Figure 8.** Figure 8: FIG. 8: Kinetic freeze out temperature versus mean [PITH_FULL_IMAGE:figures/full_fig_p012_8.png] view at source ↗

**Figure 9.** Figure 9: FIG. 9: Comparison of [PITH_FULL_IMAGE:figures/full_fig_p014_9.png] view at source ↗

**Figure 10.** Figure 10: FIG. 10: Simultaneous BGBW function fit to the identified charged particles’ [PITH_FULL_IMAGE:figures/full_fig_p016_10.png] view at source ↗

read the original abstract

Machine learning has become a powerful tool in high-energy collider experiments, which enables the studies based on data-driven approaches to complex reconstruction and regression tasks. The study of identified hadron spectra in pseudorapidity regions beyond detector acceptance, which is limited to mid-rapidity regions, carries important information about particle production, yet remains unmeasured. In this work, we develop a physics-informed neural network, trained on PYTHIA8 $pp$ collisions at $\sqrt{s}=13.6$ TeV, to infer $p_{\rm T}$ spectra of $\pi^{\pm}$, $K^{\pm}$, $p/\bar{p}$, $\Lambda/\bar{\Lambda}$, and $K^{0}_{\mathrm{s}}$ in different rapidity regions. Physics-motivated constraints, including particle yield ratios, spectral shape, and smoothness, are incorporated into the loss function. A staged hyperparameter optimization strategy is used to ensure stability. The model achieves yield uncertainties of ${\sim}1.5\%$, $1.8\%$, and $5.83\%$ in the training, interpolation, and extrapolation regimes, respectively, outperforming XGBoost and LightGBM. It further reproduces key observables such as particle yield ratios, the multiplicity dependence of $\langle p_{\rm T} \rangle$, and kinetic freeze-out parameters, indicating that the model captures the underlying physics and provides reliable predictions beyond the measured phase space.

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

The paper trains a PINN on PYTHIA8 alone to extrapolate identified hadron spectra to unmeasured rapidities with added physics constraints, but all checks stay inside the same generator.

read the letter

The core contribution is a physics-informed neural network that takes PYTHIA8 pp events at 13.6 TeV and fills in pT spectra for pions, kaons, protons, lambdas and K0s at higher rapidities. It folds yield ratios, spectral shape and smoothness directly into the loss, uses staged hyperparameter tuning, and reports yield uncertainties of roughly 1.5 % in training, 1.8 % in interpolation and 5.8 % in extrapolation while beating XGBoost and LightGBM on the same task. It also shows the network can recover multiplicity dependence of mean pT and kinetic freeze-out parameters inside the simulation. That combination of constraints and the concrete numbers on extrapolation error are the parts that feel new for this specific LHC use case. The method is straightforward to implement and the staged optimization looks like a practical way to keep the network stable. The main limitation is that every number, every loss term and every validation split comes from the identical PYTHIA8 sample. There is no comparison to real mid-rapidity data in the overlap region and no test against a second generator. Because of that, the claim that the network has captured model-independent physics rather than PYTHIA-specific patterns rests on internal consistency alone. The 5.8 % extrapolation uncertainty is therefore best read as an internal figure of merit until external checks appear. This work is aimed at people who already run or tune event generators and need a fast way to extend existing spectra. A reader who wants to see how physics priors can be added to a regression network will find the details useful. I would send it to peer review; the idea is clear enough and the implementation is described well enough that referees can ask for the missing cross-checks without starting from scratch.

Referee Report

2 major / 3 minor

Summary. The paper claims to develop a physics-informed neural network (PINN) trained solely on PYTHIA8 Monte Carlo events for pp collisions at √s=13.6 TeV. The PINN infers p_T spectra for identified hadrons (π±, K±, p/¯p, Λ/¯Λ, K0s) across different rapidity intervals, including extrapolation to unmeasured regions. Constraints on particle yield ratios, spectral shapes, and smoothness are embedded in the loss function. Reported performance includes yield uncertainties of ∼1.5% (training), 1.8% (interpolation), and 5.83% (extrapolation), superior to XGBoost and LightGBM, with reproduction of yield ratios, ⟨p_T⟩ multiplicity dependence, and kinetic freeze-out parameters.

Significance. If the generalization holds beyond the training generator, the work could provide a useful tool for extending LHC measurements of identified hadron production to forward rapidity regions not covered by detectors. The physics-informed approach helps mitigate some risks of pure ML methods. However, the absence of external validation means the significance for real data applications remains to be demonstrated.

major comments (2)

[Validation procedure] The network is trained and validated entirely within PYTHIA8, with all reproduced observables generated from the same model. This raises the risk that the model learns simulation-specific artifacts rather than universal physics, directly impacting the reliability of the 5.83% extrapolation uncertainty for application to experimental data.
[Comparison with data] There is no mention of testing the model's predictions against real LHC data in the mid-rapidity region where measurements are available. Including such a test would be essential to support the claim that the physics constraints enable model-independent inference.

minor comments (3)

The abstract states that a 'staged hyperparameter optimization strategy is used to ensure stability'; providing more specifics on this strategy, such as the hyperparameters tuned and the metrics used, would enhance the reproducibility of the results.
[Abstract] The uncertainties are given with approximate symbols (∼); more precise values or the method of their calculation (e.g., from model ensembles) should be detailed in the main text.
Consider adding a discussion on potential limitations, such as the dependence on the choice of PYTHIA8 tune or parameters.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the thoughtful and constructive report. We address each major comment below and indicate the revisions we will make to the manuscript.

read point-by-point responses

Referee: [Validation procedure] The network is trained and validated entirely within PYTHIA8, with all reproduced observables generated from the same model. This raises the risk that the model learns simulation-specific artifacts rather than universal physics, directly impacting the reliability of the 5.83% extrapolation uncertainty for application to experimental data.

Authors: We agree that the exclusive use of PYTHIA8 for both training and validation introduces a genuine risk of capturing generator-specific features. The physics-informed constraints (yield ratios, spectral shapes, and smoothness) are intended to encode general principles, yet they cannot fully eliminate dependence on the underlying event generator. The quoted uncertainties are therefore conditional on this training set. In the revised manuscript we will add an explicit limitations subsection that states this caveat, clarifies that the 5.83 % figure applies only within the PYTHIA8 ensemble, and outlines the need for future tests with additional generators or real data before claiming broader applicability. revision: partial
Referee: [Comparison with data] There is no mention of testing the model's predictions against real LHC data in the mid-rapidity region where measurements are available. Including such a test would be essential to support the claim that the physics constraints enable model-independent inference.

Authors: The original manuscript does not contain a direct comparison with experimental data because the study was designed to quantify performance against known truth in a controlled Monte Carlo environment. We accept that this omission weakens the support for model-independent inference. In the revised version we will add a dedicated comparison in the results section: the trained network will be evaluated on mid-rapidity multiplicity bins drawn from published ALICE 13 TeV pp data, and the inferred p_T spectra will be overlaid on the measured spectra. Any observed differences will be discussed, with explicit attribution to possible PYTHIA8–data discrepancies versus limitations of the inference method itself. revision: yes

Circularity Check

0 steps flagged

No significant circularity: validation is internal consistency test within PYTHIA8, not reduction by construction

full rationale

The paper trains a physics-informed NN exclusively on PYTHIA8 events at 13.6 TeV, incorporates yield ratios/spectral shape/smoothness into the loss, and reports performance (1.5%/1.8%/5.83% uncertainties) plus reproduction of observables on training/interpolation/extrapolation splits that are all drawn from the same generator. This is a standard ML generalization test inside one simulation framework rather than a derivation that reduces to its inputs by construction. No equations are shown to be tautological, no self-citations are invoked as load-bearing uniqueness theorems, and the physics constraints are external additions rather than redefinitions of the target. The claim that the model 'captures the underlying physics' rests on an assumption about PYTHIA8 fidelity plus the constraints, which is a correctness issue, not circularity. The derivation chain remains self-contained against its stated benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the assumption that PYTHIA8 provides a faithful training distribution and that the added loss constraints enforce physical behavior without introducing new biases.

axioms (1)

domain assumption PYTHIA8 Monte Carlo accurately models the underlying particle production physics in 13.6 TeV pp collisions
The entire training and validation pipeline uses PYTHIA8 events as ground truth.

pith-pipeline@v0.9.0 · 5576 in / 1355 out tokens · 59903 ms · 2026-05-12T02:13:46.472056+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

65 extracted references · 65 canonical work pages · 1 internal anchor

[1]

Dataset Generation As the primary objective of this work is to study par- ticle yield distributions as a function ofp T,η, andN ch, we generate minimum-biasppcollisions at √s= 13.6 TeV using PYTHIA8 within the MAGE pipeline [19]. PYTHIA8 is a widely used Monte Carlo event genera- tor designed to simulate leptonic, hadronic, and heavy- ion collisions over ...

work page
[2]

The generated dataset is trans- formed intop T spectra for differentN ch, PID, andη ranges

Dataset Preprocessing and Feature Selection To evaluate the applicability of machine learning mod- els across differentηranges, we construct a spectral rep- resentation that captures the dependence on track and event-level variables. The generated dataset is trans- formed intop T spectra for differentN ch, PID, andη ranges. For this study, we consider thr...

work page
[3]

It is widely used due to its flexibility across a range of tasks, including classifica- tion, regression, and ranking

XGBoost XGBoost, short for eXtreme Gradient Boosting, is an advanced implementation of gradient-boosting decision trees (GBDT) that incorporates several optimizations to improve speed and accuracy. It is widely used due to its flexibility across a range of tasks, including classifica- tion, regression, and ranking. XGBoost formulates gra- dient boosting u...

work page
[4]

LightGBM LightGBM (Light Gradient Boosting Machine) is an ef- ficient implementation of gradient-boosted decision trees designed to reduce computational cost and memory us- age compared to conventional boosting methods [32]. In contrast to the traditional level-wise tree growth strategy, where all nodes at a given depth are expanded uniformly, LightGBM ad...

work page
[5]

Physics-Informed Neural Network Physics-informed neural networks are a class of neural networks trained to respect laws of physics described by general nonlinear partial differential equations. Instead of solely relying on data, PINNs enforce governing equa- tions as constraints, typically expressed as differential equations, to guide the model towards ph...

work page
[6]

Physics-Informed Staged Hyperparameter Optimization Optimizing the model over the full hyperparameter search space, along with physics-informed loss functions, is computationally expensive and, across many trials, can lead to non-physical predictions. After experimenting with different optimization strategies, we find that the HPO workflow for the PINN ca...

work page
[7]

Particle ratios The collective expansion of the hot, dense medium formed in ultrarelativistic collision generates a radially outward flow of the particles that imparts a momentum boost to the produced hadrons [6, 44], known as radial flow. Since the magnitude of this boost scales with the particle mass, heavier species acquire a larger momen- tum than lig...

work page
[8]

Driven by the growing MPI activity and color reconnection, the⟨p T⟩is expected to increase with⟨N ch⟩in pp collisions [5, 48, 49]

Mean transverse momentum The mean transverse momentum (⟨p T⟩) as a function of charged-particle multiplicity (N ch) provides comple- mentary and integrated validation of the model’s spectral prediction quality. Driven by the growing MPI activity and color reconnection, the⟨p T⟩is expected to increase with⟨N ch⟩in pp collisions [5, 48, 49]. This rise is mo...

work page
[9]

Indian participation in the ALICE experiment at CERN,

Kinetic freeze out parameters validation In ultrarelativistic collisions, the spacetime evolution of the system proceeds through several distinct phases. Starting from the pre-equilibrium phase, the system evolves into the QGP phase, followed by the mixed phase, chemical freeze-out, and the hadron gas phase [50, 51]. Finally, when the interaction rate dro...

work page 2025
[10]

L. D. McLerran, Rev. Mod. Phys.58, 1021 (1986)

work page 1986
[11]

C. Y. Wong,Introduction to High-Energy Heavy-Ion Col- lisions(World Scientific, Singapore, 1994)

work page 1994
[12]

Gyulassy and L

M. Gyulassy and L. McLerran, Nucl. Phys. A750, 30 (2005)

work page 2005
[13]

Shuryak, Rev

E. Shuryak, Rev. Mod. Phys.89, 035001 (2017)

work page 2017
[14]

Acharyaet al.(ALICE Collaboration), Eur

S. Acharyaet al.(ALICE Collaboration), Eur. Phys. J. C80, 693 (2020)

work page 2020
[15]

Schnedermann, J

E. Schnedermann, J. Sollfrank and U. W. Heinz, Phys. Rev. C48, 2462 (1993)

work page 1993
[16]

Prasad, N

S. Prasad, N. Mallick, D. Behera, R. Sahoo and S. Tri- pathy, Sci. Rep.12, 3917 (2022)

work page 2022
[17]

Abelevet al.(ALICE Collaboration), Phys

B. Abelevet al.(ALICE Collaboration), Phys. Rev. C 88, 044910 (2013)

work page 2013
[18]

Menon Kavumpadikkal Radhakrishnan, S

A. Menon Kavumpadikkal Radhakrishnan, S. Prasad, S. Tripathy, N. Mallick and R. Sahoo, Eur. Phys. J. Plus 140, 110 (2025)

work page 2025
[19]

Aamodtet al.(ALICE Collaboration), JINST3, S08002 (2008)

K. Aamodtet al.(ALICE Collaboration), JINST3, S08002 (2008)

work page 2008
[20]

Acharyaet al.(ALICE Collaboration) JINST19, P05062 (2024)

S. Acharyaet al.(ALICE Collaboration) JINST19, P05062 (2024)

work page 2024
[21]

Sahoo, K

R. Sahoo, K. Goswami and S. Prasad, arXiv:2509.00712 (Phys. Rev. D: In press)

work page arXiv
[22]

Goswami, S

K. Goswami, S. Prasad, N. Mallick, R. Sahoo and G. B. Mohanty, Phys. Rev. D110, 034017 (2024)

work page 2024
[23]

Prasad, N

S. Prasad, N. Mallick and R. Sahoo, Phys. Rev. D109, 014005 (2024)

work page 2024
[24]

Mallick, S

N. Mallick, S. Tripathy, A. N. Mishra, S. Deb and R. Sa- hoo, Phys. Rev. D103, 094031 (2021)

work page 2021
[25]

Mallick, S

N. Mallick, S. Prasad, A. N. Mishra, R. Sahoo and G. G. Barnaf¨ oldi, Phys. Rev. D107, 094001 (2023)

work page 2023
[26]

Mallick, S

N. Mallick, S. Prasad, A. N. Mishra, R. Sahoo and G. G. Barnaf¨ oldi, Phys. Rev. D105, 114022 (2022)

work page 2022
[27]

R. K. Tiwari, K. Goswami, S. Prasad, C. R. Singh, R. Sa- hoo and M. Y. Jamal, arXiv:2604.05858

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

Gupta, K

R. Gupta, K. Goswami, S. Prasad, and R. Sahoo, MAGE, paper in preparation

work page
[29]

o nnblad and T. Sj \

C. Bierlich, L. L¨ onnblad and T. Sj¨ ostrand, arXiv:2603.01744

work page arXiv
[30]

Sj¨ ostrand, S

T. Sj¨ ostrand, S. Ask, J. R. Christiansen, R. Corke, N. De- sai, P. Ilten, S. Mrenna, S. Prestel, C. O. Rasmussen and P. Z. Skands, Comput. Phys. Commun.191, 159 (2015)

work page 2015
[31]

Sjostrand, S

T. Sjostrand, S. Mrenna and P. Z. Skands, JHEP05, 026 (2006)

work page 2006
[32]

Prasad, S

S. Prasad, S. Tripathy, B. Sahoo and R. Sahoo, Phys. Rept.1181, 1 (2026)

work page 2026
[33]

Prasad, B

S. Prasad, B. Sahoo, S. Tripathy, N. Mallick and R. Sa- hoo, Phys. Rev. C111, 044902 (2025)

work page 2025
[34]

Acharyaet al.(ALICE Collaboration), JHEP05, 184 (2024)

S. Acharyaet al.(ALICE Collaboration), JHEP05, 184 (2024)

work page 2024
[35]

Adamet al.(ALICE Collaboration), JHEP09, 148 (2015)

J. Adamet al.(ALICE Collaboration), JHEP09, 148 (2015)

work page 2015
[36]

Thakur, S

D. Thakur, S. De, R. Sahoo and S. Dansana, Phys. Rev. D97, 094002 (2018)

work page 2018
[37]

S. Deb, D. Thakur, S. De and R. Sahoo, Eur. Phys. J. A 56, 134 (2020)

work page 2020
[38]

Brun and F

R. Brun and F. Rademakers, Nucl. Instrum. Meth. A 389, 81 (1997)

work page 1997
[39]

Skands, S

P. Skands, S. Carrazza and J. Rojo, Eur. Phys. J. C74, 3024 (2014)

work page 2014
[40]

Proceedings of the 22nd

T. Chen and C. Guestrin, inProceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining(ACM, New York, 2016), pp. 785,10.1145/2939672.2939785

work page doi:10.1145/2939672.2939785 2016
[41]

G. Ke, Q. Meng, T. Finley, T. Wang, W. Chen, W. Ma, Q. Ye and T. Y. Liu, inAdvances in Neural Informa- tion Processing Systems 30 (NIPS 2017)(Curran Asso- ciates, Inc., 2017), pp. 3149,https://dl.acm.org/doi/ 10.5555/3294996.3295074

work page doi:10.5555/3294996.3295074 2017
[42]

Shokr, A

E. Shokr, A. De Roeck and M. A. Mahmoud, Sci. Rep. 12, 8449 (2022)

work page 2022
[43]

Available Online:https://xgboost.readthedocs.io/ en/stable/

work page
[44]

Available Online:https://lightgbm.readthedocs.io/ en/latest/

work page
[45]

G. E. Karniadakis, I. G. Kevrekidis, L. Lu, P. Perdikaris, S. Wang and L. Yang, Nat. Rev. Phys.3, 422 (2021)

work page 2021
[46]

Raissi, P

M. Raissi, P. Perdikaris, and G. E. Karniadakis, J. Com- put. Phys.378, 686 (2019)

work page 2019
[47]

J. C. Bezdek and R. J. Hathaway, inAdvances in Soft Computing, Vol.2275(Springer, Berlin, Heidelberg, 2002), pp. 288

work page 2002
[48]

arXiv preprint arXiv:2304.11127 , year=

S. Watanabe, arXiv:2304.11127

work page arXiv
[49]

Optuna: A Next-generation Hyperparameter Optimization Framework

T. Akiba, S. Sano, T. Yanase, T. Ohta and M. Koyama, arXiv:1907.10902

work page Pith review arXiv 1907
[50]

Jiang and A

J. Jiang and A. Mian, arXiv:2409.00584

work page arXiv
[51]

BOHB: Robust and Efficient Hyperparameter Optimization at Scale

S. Falkner, A. Klein and F. Hutter, arXiv:1807.01774

work page Pith review arXiv
[52]

Understanding the exploding gradient problem.CoRR abs/1211.5063 (2012)

R. Pascanu, T. Mikolov and Y. Bengio, arXiv:1211.5063

work page arXiv
[53]

P. J. Siemens and J. O. Rasmussen, Phys. Rev. Lett.42, 880 (1979)

work page 1979
[54]

Ortiz Velasquez, P

A. Ortiz Velasquez, P. Christiansen, E. Cuautle Flores, I. Maldonado Cervantes and G. Pai´ c, Phys. Rev. Lett. 111, 042001 (2013)

work page 2013
[55]

Retiere and M

F. Retiere and M. A. Lisa, Phys. Rev. C70, 044907 (2004)

work page 2004
[56]

Adamet al.(ALICE Collaboration), Nature Phys.13, 535 (2017)

J. Adamet al.(ALICE Collaboration), Nature Phys.13, 535 (2017)

work page 2017
[57]

Acharyaet al.(ALICE Collaboration), Phys

S. Acharyaet al.(ALICE Collaboration), Phys. Rev. C 99, 024906 (2019)

work page 2019
[58]

B. B. Abelevet al.(ALICE Collaboration), Phys. Lett. B727, 371 (2013)

work page 2013
[59]

Braun-Munzinger, J

P. Braun-Munzinger, J. Stachel, J. P. Wessels and N. Xu, Phys. Lett. B344, 43 (1995)

work page 1995
[60]

Huovinen, P

P. Huovinen, P. F. Kolb, U. W. Heinz, P. V. Ruuskanen and S. A. Voloshin, Phys. Lett. B503, 58 (2001)

work page 2001
[61]

Acharyaet al.(ALICE Collaboration), Eur

S. Acharyaet al.(ALICE Collaboration), Eur. Phys. J. C84, 813 (2024)

work page 2024
[62]

Ortiz, G

A. Ortiz, G. Pai´ c and E. Cuautle, Nucl. Phys. A941, 78 (2015)

work page 2015
[63]

I. C. Arseneet al.(BRAHMS Collaboration), Phys. Rev. C94, 014907 (2016)

work page 2016
[64]

Z. Tang, L. Yi, L. Ruan, M. Shao, H. Chen, C. Li, B. Mo- hanty, P. Sorensen, A. Tang and Z. Xu, Chin. Phys. Lett. 30, 031201 (2013)

work page 2013
[65]

Adcoxet al.(PHENIX Collaboration), Phys

K. Adcoxet al.(PHENIX Collaboration), Phys. Rev. C 69, 024904 (2004)

work page 2004