Pith. sign in

REVIEW 2 major objections 3 minor 1 cited by

This search for exotic Higgs decays to four photons in the semi-merged topology finds no excess and sets the most stringent upper limits to date on the cross section times branching fraction for a new particle A with mass between 1 and 5 Ge

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-03 13:07 UTC pith:KHWJE6I2

load-bearing objection Solid CMS Run 2 search that closes the semi-merged gap and sets the best limits in the 2-6 GeV region, but the 'most stringent in 1-5 GeV' claim is overstated and the background transfer ratio is validated only in sidebands. the 2 major comments →

arxiv 2601.00183 v2 pith:KHWJE6I2 submitted 2026-01-01 hep-ex

Search for exotic Higgs boson decays H to mathcal{AA} with mathcal{AA} to γγ in events with a semi-merged topology in proton-proton collisions at sqrt{s} = 13 TeV

CMS Collaboration This is my paper
classification hep-ex
keywords exotic Higgs decaysfour photonssemi-merged topologygraph neural networkmass regressionCMS13 TeVupper limits
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The paper targets a previously un-covered mass window for exotic Higgs decays H→AA→4γ, where one diphoton pair is reconstructed as a single photon-like object and the other as two resolved photons. Using 138 fb⁻¹ of 13 TeV proton-proton collisions, the CMS experiment sees data that agree with standard model background expectations. The analysis sets 95% confidence-level upper limits on σ(pp→H)B(H→AA→4γ) ranging from 0.264 pb at m_A=1 GeV to 0.005 pb at m_A=15 GeV, with the best improvement—about two orders of magnitude—around m_A=3 GeV. This closes the gap between fully merged and fully resolved searches and demonstrates a new semimerged reconstruction technique.

Core claim

The central result is that exotic Higgs decays H→AA→4γ in the semimerged topology are not observed, and the resulting upper limits on σ(pp→H)B(H→AA→4γ) are the most stringent yet in the 1–5 GeV mass range. To reach this region, the authors developed a graph neural network that directly regresses the mass of a merged diphoton from energy deposits in the electromagnetic calorimeter, bypassing the need to resolve the two photons individually. The signal is extracted from a two-dimensional template of merged-candidate mass versus resolved-diphoton mass, with the background inferred from data sidebands under a ratio-assumption described in Eq. (5).

What carries the argument

The key machinery is a graph neural network (GNN) called a dynamic reduction network with residual connections, trained end-to-end on calorimeter crystal energy and position information to predict the mass of a semi-merged photon pair. It replaces the need to individually reconstruct both photons and is trained on simulated A→γγ decays with flat mass and transverse-momentum distributions, augmented by single-photon events via domain continuation. The GNN output feeds a two-dimensional (m_A1, m_A2) template where signal appears on the diagonal; background is modeled from sidebands using the sideband-ratio relation.

Load-bearing premise

The background estimate in the signal region relies on the assumption that the ratio of diagonal to off-diagonal events in the two-dimensional mass plane is the same in the Higgs-mass signal region as in the Higgs-mass sidebands (Eq. 5).

What would settle it

Recompute the signal-region background using a different sideband definition (e.g., a different m_A-SR boundary or an alternative m_H sideband split) and compare the predicted signal-region event count with the observed data; a significant discrepancy would show the ratio assumption breaks down and would invalidate the quoted limits.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • The 1–5 GeV mass gap in CMS searches for H→AA→4γ is now covered by a dedicated semimerged analysis, complementing the fully merged and fully resolved searches.
  • The graph neural network mass regression can be applied to other signatures involving merged electromagnetic objects, potentially improving searches for light resonances decaying to photon pairs.
  • The observed upper limits provide new constraints on photophilic models and 2HDM+S scenarios that predict sizable branching fractions for H→AA→4γ with m_A in the few-GeV range.
  • The data-driven background method, validated in sidebands, allows the search to avoid large simulation-based background uncertainties and can be reused in future semimerged analyses.

Where Pith is reading between the lines

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

  • A natural extension is to adapt the GNN regression to include endcap photons or lower photon p_T thresholds, which could extend sensitivity to lower masses or higher pseudorapidities.
  • The sideband-ratio assumption in Eq. (5) might be tested more stringently by using an independent control sample or by loosening the m_A-SR boundary; if the ratio varies with triphoton mass, the background estimate could be biased.
  • The semimerged topology is also relevant for pair production of axion-like particles not necessarily from Higgs decays; the technique could be reused in inclusive searches.
  • With the higher collision energy and luminosity of Run 3, this analysis could be updated to reach lower signal cross sections, provided the trigger and reconstruction thresholds are revisited.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 3 minor

Summary. The paper reports a search for the exotic Higgs decay H→AA→4γ in the 'semi-merged' topology, where one A→γγ decay is reconstructed as a single photon-like object and the other as two resolved photons. The analysis uses 138 fb−1 of 13 TeV pp collision data collected by CMS. A graph neural network regresses the invariant mass of the merged photon candidate from ECAL crystal-level energy deposits. The background is estimated entirely from data sidebands in the two-dimensional (mA1,mA2) plane versus the triphoton mass. No significant excess over the SM background is found, and 95% CL upper limits are set on σ(pp→H)B(H→AA→4γ) from 0.264 pb at mA=1 GeV to 0.005 pb at mA=15 GeV. The paper claims these are the most stringent limits in the 1–5 GeV mA range.

Significance. If the result stands, it fills the gap between the previously published CMS fully-merged search (mA≈0.1–1.2 GeV) and the fully-resolved search (mA≈15–62 GeV), and it provides the first CMS measurement in the 1–15 GeV mass range with a semi-merged signature. The analysis is methodologically strong in several respects: the background is data-driven with sideband and signal-injection validation, the GNN mass regression is calibrated with Z→ee data and the associated scale/smearing uncertainties are propagated, and the results are to be tabulated in HEPData. The main concerns are the unsupported 'most stringent' claim in the abstract and the lack of a direct closure test for the key background-normalization assumption in Eq. (5).

major comments (2)
  1. [Abstract / Section 8] The abstract and Section 9 state that the limits are 'the most stringent to date in the 1–5 GeV mA range'. However, Section 8 says that 'the sensitivity is reduced at the lowest mass compared to the previous CMS search, since the fully merged photon signature still dominates there'. The fully merged search [20] covers up to mA≈1.2 GeV, so at mA≈1 GeV the fully merged limit is expected to be more stringent. This appears to contradict the 'most stringent in 1–5 GeV' claim over part of the stated range. Please either restrict the claim to the mass range above the fully merged coverage, or provide a quantitative comparison (e.g., in Fig. 12 or in HEPData) showing that the new limit is below all previous limits for every mA in 1–5 GeV.
  2. [Section 6.2, Eq. (5)] The final signal-region background normalization is obtained from the assumption that the ratio R(mH) = N(mA-SR)/N(mA-SB) is independent of the triphoton mass, i.e., that the ratio measured in the mH sidebands equals that in the mH signal region. This assumption is not directly validated. Figure 9 only validates the background shape in the mA-SB region, and the signal-injection tests do not test the constancy of R across mH. If R varies between the low sideband (90–100 GeV), the signal region (100–135 GeV), and the high sideband (135–180 GeV), the SR background yield will be biased, directly shifting the extracted limits. Please add a closure test that uses the two mH sidebands as pseudo-signal and pseudo-sideband regions (or vice versa) and reports the observed variation of R, or alternatively assign a systematic uncertainty that covers that variation. This is a load-bearing point becau
minor comments (3)
  1. [Section 6.2.1] The text states that signal contamination in the mA-SB region 'was considered but deemed to be negligible.' To make this quantitative, please report the largest expected signal leakage fraction into the mA-SB region for the signal hypotheses considered.
  2. [Section 5.2] The choice of tuning the number of single-photon events so that the 'density of negative-mass samples is the same as that of the positive-mass samples' is not justified. A brief explanation of why this density balance is optimal for the domain-continuation method would be helpful.
  3. [Section 4] The selection requires exactly three reconstructed photons. The paper does not describe how events with additional photon-like objects (e.g., from pileup) are rejected, nor whether such events cause a significant efficiency loss. A short comment on this would clarify the object selection.

Circularity Check

0 steps flagged

No significant circularity: the result is an experimental measurement whose background and signal models are independent of the observed signal-region yield; self-citations are methodological rather than load-bearing.

full rationale

The paper does not derive its central claim from the inputs it claims to predict. The signal normalization in Eq. (1) is an external parameter multiplied by a simulated efficiency, and the background in the final signal region is constructed from mH and mA sidebands using Eqs. (2)-(7). The target yield N(mH-SR∩mA-SR) is not used as an input to the background model; it is estimated from the diagonal sideband yield and an off-diagonal transfer ratio, so this is a standard sideband/transfer-factor estimate rather than a fit that forces the null result. The mass regression technique is taken from the collaboration's prior work (Refs. [18,20]), but its scale and smearing are calibrated against Z→ee data and propagated as systematic uncertainties, and the signal templates come from independent simulation; the self-citations are therefore methodological, not load-bearing reductions. The only notable caveat is the explicit assumption in Eq. (5) that the diagonal/off-diagonal event ratio is independent of the triphoton mass region. That is an untested modeling assumption that could bias the background estimate, but it is not circular because it does not use the predicted signal-region yield as an input and is not an equation that reduces to itself by construction. No step in the derivation chain reduces to an input, a fitted parameter renamed as a prediction, or an author-imposed uniqueness/ansatz chain.

Axiom & Free-Parameter Ledger

4 free parameters · 6 axioms · 1 invented entities

The analysis rests on standard CMS detector/simulation assumptions, a data-driven background transfer, and a trained GNN. No new physical entity is introduced beyond the hypothesized A that is searched for. The free parameters are ML training/calibration degrees of freedom and background-shape correction coefficients, not physical constants fitted to the target observable.

free parameters (4)
  • GNN mass-regression weights = not specified (trained on ~600k A→γγ and 70k single-photon events)
    The graph neural network that predicts the merged-leg mass is trained on simulation; all subsequent physics results depend on this trained model.
  • Mass regression calibration scale shift = 1.6–6% depending on year and |eta|
    Best-fit scale shifts applied to simulated regressed masses, derived from Z→ee tag-and-probe data in Section 5.3.
  • Mass regression calibration smearing = up to 260 MeV
    Extra Gaussian smearing added to simulated regressed masses to match data resolution, fitted in the same Z→ee calibration.
  • Background shape Chebyshev coefficients = not specified (second order in m_A1, third order in m_A2)
    Coefficients fitted to residual data-versus-background differences in the m_A-SB region and used to define background shape systematics.
axioms (6)
  • domain assumption The background transfer ratio in Eq. (5) is independent of the triphoton mass region.
    The final signal-region background yield is derived by assuming the on-diagonal/off-diagonal ratio measured in m_H sidebands holds in the m_H signal window. This is stated explicitly and validated only in sidebands.
  • domain assumption GEANT4 simulation of the CMS ECAL accurately describes crystal-level energy deposits for merged and low-energy photon showers.
    The GNN is trained entirely on simulated calorimeter images; inaccuracies in shower simulation propagate directly into the mass regression.
  • domain assumption Z→ee electrons with a nearby soft PF deposit are a valid proxy for A→γγ semi-merged showers after scale/smearing corrections.
    Section 5.3 uses electrons to calibrate the regression response; this assumes the electron shower shape after smearing mimics the signal topology.
  • domain assumption The GNN trained on 2017 simulation conditions generalizes to 2016 and 2018 data-taking conditions after per-year scale and smearing corrections.
    Figure 5 shows alignment in simulation across years, but data validation for all years is indirect via the electron calibration.
  • domain assumption The signal process is H→AA→4γ with a prompt A, equal masses, and a generator-level model (MadGraph5 SM+Dark Vector+Dark Higgs) that accurately models the signal kinematics.
    The signal efficiency and templates rely on this model, and only the prompt lifetime scenario is considered.
  • domain assumption The background 2D m_A shape in the m_H signal region is a linear combination of the low and high m_H sideband shapes, Eq. (2).
    The background shape is constructed by weighting the two m_H sidebands by their event fractions; any nonlinear m_H evolution of the 2D shape would bias the estimate.
invented entities (1)
  • A (light spin-0/axion-like particle) no independent evidence
    purpose: Hypothetical Higgs decay product; the search targets H→AA→4γ with A→γγ.
    A is postulated in the BSM literature the paper cites (2HDM+S, ALP models), not discovered here. The paper provides a search, not independent evidence of its existence.

pith-pipeline@v1.3.0-alltime-deepseek · 41179 in / 12889 out tokens · 140676 ms · 2026-08-03T13:07:13.128297+00:00 · methodology

0 comments
read the original abstract

A search for exotic Higgs boson decays H $\to$ $\mathcal{AA}$, with $\mathcal{A}$ $\to$ $\gamma\gamma$ is presented, using events with a semi-merged topology. One of the hypothetical particles, $\mathcal{A}$, is assumed to decay promptly into a semi-merged diphoton system reconstructed as a single photon-like object, while the other $\mathcal{A}$ decays into two resolved photons. The search is performed using proton-proton collision data collected by the CMS experiment at $\sqrt{s}$ = 13 TeV, corresponding to an integrated luminosity of 138 fb$^{-1}$. The data agree with the standard model background expectation. Upper limits are set on the product of the Higgs boson production cross section and the branching fraction, $\sigma$(pp $\to$ H) $\mathcal{B}$(H $\to$ $\mathcal{AA}$ $\to$ 4$\gamma$), which range from 0.264 to 0.005 pb at 95% confidence level, for $\mathcal{A}$ masses in the range 1 $\lt$ $m_\mathcal{A}$ $\lt$ 15 GeV. These limits are the most stringent to date in the 1$-$5 GeV $m_\mathcal{A}$ range.

Figures

Figures reproduced from arXiv: 2601.00183 by CMS Collaboration.

Figure 1
Figure 1. Figure 1: Energy deposit maps in the ECAL from simulated H [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Illustration of the network architecture used for the mass regression. Input energy [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: True vs. predicted mass from the mass regression during training validation using [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Predicted mass distribution from the mass regression on the merged leg of the simu [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Predicted mass distribution from the mass regression on the merged leg of the sim [PITH_FULL_IMAGE:figures/full_fig_p011_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Regressed mass for Z → e +e − electrons in the 2017 data vs. simulation for all events passing the selections (left), showing a peak at zero, and for a subset of those events with a soft photon-like deposit near the electron (right), showing a peak away from zero. The variable f denotes the fraction of events per bin. The area of each distribution is normalized to unity. In the upper panels of each plot, t… view at source ↗
Figure 7
Figure 7. Figure 7: The 2D mA distribution of the signal model from 2017 simulation for the mA = 5 GeV hypothesis, normalized to σ(pp → H)B(H → AA → 4γ) = 1 pb. The region enclosed by the jagged lines defines the mA-SR region, while the area outside corresponds to the mA-SB region [PITH_FULL_IMAGE:figures/full_fig_p013_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: The 2D mA distribution of the background model constructed from data sidebands. The region enclosed by the jagged lines defines the mA-SR region, while the area outside corre￾sponds to the mA-SB region. axis with Chebyshev polynomials. The resulting functions are then applied to the nominal 2D background distribution to produce alternative templates, which are used to evaluate the background shape systemat… view at source ↗
Figure 9
Figure 9. Figure 9: Expected background versus observed data 2D [PITH_FULL_IMAGE:figures/full_fig_p016_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: The 2D mA spectra in the final signal region. Upper plot: The unrolled 2D mA dis￾tribution made by scanning along bins of increasing mA2 at fixed mA1 before incrementing in mA1 . Only the bins in the mA-SR region are included, with the x-axis corresponding to the un￾rolled bin index of the selected bins, listed sequentially. Lower plot: 1D projections on the mA1 (left) and mA2 (right) axes of the 2D mA di… view at source ↗
Figure 11
Figure 11. Figure 11: Observed (solid black line) and median expected (dashed black line) upper limits at [PITH_FULL_IMAGE:figures/full_fig_p020_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Observed (solid line) and expected (dashed line) upper limits, including the 95% [PITH_FULL_IMAGE:figures/full_fig_p020_12.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Transformer-based machine learning using low-level calorimeter signals for collimated photon identification at collider experiments

    hep-ph 2026-07 accept novelty 6.0

    Cell-level Transformers classify collimated ALP photon-jets versus single photons with AUC 0.98 and regress diphoton mass to ~64 MeV, beating shower-shape and other ML baselines in an ATLAS-like GEANT4 simulation.

Reference graph

Works this paper leans on

66 extracted references · 10 canonical work pages · cited by 1 Pith paper

  1. [1]

    Observation of a new particle in the search for the standard model Higgs boson with the ATLAS detector at the LHC

    ATLAS Collaboration, “Observation of a new particle in the search for the standard model Higgs boson with the ATLAS detector at the LHC”,Phys. Lett. B716(2012) 1, doi:10.1016/j.physletb.2012.08.020

    doi:10.1016/j.physletb.2012.08.020 2012

  2. [2]

    Observation of a new boson at a mass of 125 GeV with the CMS experiment at the LHC

    CMS Collaboration, “Observation of a new boson at a mass of 125 GeV with the CMS experiment at the LHC”,Phys. Lett. B716(2012) 30, doi:10.1016/j.physletb.2012.08.021

    doi:10.1016/j.physletb.2012.08.021 2012

  3. [3]

    Observation of a new boson with mass near 125 GeV in pp collisions at √s= 7 and 8 TeV

    CMS Collaboration, “Observation of a new boson with mass near 125 GeV in pp collisions at √s= 7 and 8 TeV”,J. High Energy Phys.2013(2013), no. 06, 081, doi:10.1007/JHEP06(2013)081

    doi:10.1007/jhep06(2013)081 2013

  4. [4]

    Combined measurements and interpretations of Higgs boson production and decay at √s=13 TeV

    CMS Collaboration, “Combined measurements and interpretations of Higgs boson production and decay at √s=13 TeV”, CMS Physics Analysis Summary CMS-PAS-HIG-21-018, 2025

    2025

  5. [5]

    Exotic decays of the 125 GeV Higgs boson

    D. Curtin et al., “Exotic decays of the 125 GeV Higgs boson”,Phys. Rev. D90(2014) 075004,doi:10.1103/PhysRevD.90.075004

    doi:10.1103/physrevd.90.075004 2014

  6. [6]

    Illuminating dark photons with high-energy colliders

    D. Curtin, R. Essig, S. Gori, and J. Shelton, “Illuminating dark photons with high-energy colliders”,J. High Energy Phys.2015(2015), no. 02, 157, doi:10.1007/JHEP02(2015)157

    doi:10.1007/jhep02(2015)157 2015

  7. [7]

    Axion dark matter: What is it and why now?

    F. Chadha-Day, J. Ellis, and D. J. E. Marsh, “Axion dark matter: What is it and why now?”,Sci. Adv.8(2022) abj3618,doi:10.1126/sciadv.abj3618

    doi:10.1126/sciadv.abj3618 2022

  8. [8]

    CP conservation in the presence of instantons

    R. D. Peccei and H. R. Quinn, “CP conservation in the presence of instantons”,Phys. Rev. Lett.38(1977) 1440,doi:10.1103/PhysRevLett.38.1440

    doi:10.1103/physrevlett.38.1440 1977

  9. [9]

    Collider probes of axion-like particles

    M. Bauer, M. Neubert, and A. Thamm, “Collider probes of axion-like particles”,J. High Energy Phys.2017(2017), no. 12, 044,doi:10.1007/JHEP12(2017)044

    doi:10.1007/jhep12(2017)044 2017

  10. [10]

    The strong CP problem and axions

    R. D. Peccei, “The strong CP problem and axions”, inAxions, M. Kuster, G. Raffelt, and B. Beltran, eds. Springer Berlin Heidelberg, 2008. doi:10.1007/978-3-540-73518-2_1

    doi:10.1007/978-3-540-73518-2_1 2008

  11. [11]

    Search for exotic decays of the Higgs boson to a pair of pseudoscalars in theµµbb andττbb final states

    CMS Collaboration, “Search for exotic decays of the Higgs boson to a pair of pseudoscalars in theµµbb andττbb final states”,Eur. Phys. J. C84(2024) 493, doi:10.1140/epjc/s10052-024-12727-4

    doi:10.1140/epjc/s10052-024-12727-4 2024

  12. [12]

    Search for the decay of the Higgs boson to a pair of light pseudoscalar bosons in the final state with four bottom quarks in proton-proton collisions at √s=13 TeV

    CMS Collaboration, “Search for the decay of the Higgs boson to a pair of light pseudoscalar bosons in the final state with four bottom quarks in proton-proton collisions at √s=13 TeV”,J. High Energy Phys.2024(2024), no. 06, 097, doi:10.1007/JHEP06(2024)097. References 21

    doi:10.1007/jhep06(2024)097 2024

  13. [13]

    Search for light pseudoscalar boson pairs produced from Higgs boson decays using the 4τand 2µ2τfinal states in proton-proton collisions at √s=13 TeV

    CMS Collaboration, “Search for light pseudoscalar boson pairs produced from Higgs boson decays using the 4τand 2µ2τfinal states in proton-proton collisions at √s=13 TeV”, 2025.arXiv:2508.06947. Submitted toJHEP

    Pith/arXiv arXiv 2025

  14. [14]

    Search for Higgs boson exotic decays into Lorentz-boosted light bosons in the four-τfinal state at √s=13 TeV with the ATLAS detector

    ATLAS Collaboration, “Search for Higgs boson exotic decays into Lorentz-boosted light bosons in the four-τfinal state at √s=13 TeV with the ATLAS detector”,Phys. Lett. B 870(2025) 139843,doi:10.1016/j.physletb.2025.139843

    arXiv 2025

  15. [15]

    Search for light long-lived particles decaying to displaced jets in proton–proton collisions at √s=13.6 TeV

    CMS Collaboration, “Search for light long-lived particles decaying to displaced jets in proton–proton collisions at √s=13.6 TeV”,Rept. Prog. Phys.88(2025) 037801, doi:10.1088/1361-6633/adaa13

    work page doi:10.1088/1361-6633/adaa13 2025

  16. [16]

    A search for pair production of new light bosons decaying into muons in proton-proton collisions at 13 TeV

    CMS Collaboration, “A search for pair production of new light bosons decaying into muons in proton-proton collisions at 13 TeV”,Phys. Lett. B796(2019) 131, doi:10.1016/j.physletb.2019.07.013

    doi:10.1016/j.physletb.2019.07.013 2019

  17. [17]

    Particle-flow reconstruction and global event description with the CMS detector

    CMS Collaboration, “Particle-flow reconstruction and global event description with the CMS detector”,J. Instrum.12(2017) P10003, doi:10.1088/1748-0221/12/10/P10003

    doi:10.1088/1748-0221/12/10/p10003 2017

  18. [18]

    Reconstruction of decays to merged photons using end-to-end deep learning with domain continuation in the CMS detector

    CMS Collaboration, “Reconstruction of decays to merged photons using end-to-end deep learning with domain continuation in the CMS detector”,Phys. Rev. D108(2023) 052002, doi:10.1103/PhysRevD.108.052002

    doi:10.1103/physrevd.108.052002 2023

  19. [19]

    Search for the exotic decay of the Higgs boson into two light pseudoscalars with four photons in the final state in proton-proton collisions at √s=13 TeV

    CMS Collaboration, “Search for the exotic decay of the Higgs boson into two light pseudoscalars with four photons in the final state in proton-proton collisions at √s=13 TeV”,J. High Energy Phys.2023(2023), no. 07, 148,doi:10.1007/JHEP07(2023)148

    doi:10.1007/jhep07(2023)148 2023

  20. [20]

    Search for exotic Higgs boson decays H→ AA →4γwith events containing two merged diphotons in proton-proton collisions at √s= 13 TeV

    CMS Collaboration, “Search for exotic Higgs boson decays H→ AA →4γwith events containing two merged diphotons in proton-proton collisions at √s= 13 TeV”,Phys. Rev. Lett.131(2023) 101801,doi:10.1103/PhysRevLett.131.101801

    doi:10.1103/physrevlett.131.101801 2023

  21. [21]

    Search for short- and long-lived axion-like particles in H→aa→4γdecays with the ATLAS experiment at the LHC

    ATLAS Collaboration, “Search for short- and long-lived axion-like particles in H→aa→4γdecays with the ATLAS experiment at the LHC”,Eur. Phys. J. C84(2024) 742,doi:10.1140/epjc/s10052-024-12979-0

    work page doi:10.1140/epjc/s10052-024-12979-0 2024

  22. [22]

    Electron and photon reconstruction and identification with the CMS experiment at the CERN LHC

    CMS Collaboration, “Electron and photon reconstruction and identification with the CMS experiment at the CERN LHC”,J. Instrum.16(2021) P05014, doi:10.1088/1748-0221/16/05/P05014

    doi:10.1088/1748-0221/16/05/p05014 2021

  23. [23]

    HEPData record for this analysis

    “HEPData record for this analysis”, 2025.doi:10.17182/hepdata.166011

    work page doi:10.17182/hepdata.166011 2025

  24. [24]

    The CMS experiment at the CERN LHC

    CMS Collaboration, “The CMS experiment at the CERN LHC”,J. Instrum.3(2008) S08004,doi:10.1088/1748-0221/3/08/S08004

    doi:10.1088/1748-0221/3/08/s08004 2008

  25. [25]

    Development of the CMS detector for the CERN LHC Run 3

    CMS Collaboration, “Development of the CMS detector for the CERN LHC Run 3”,J. Instrum.19(2024) P05064,doi:10.1088/1748-0221/19/05/P05064

    doi:10.1088/1748-0221/19/05/p05064 2024

  26. [26]

    Performance of the CMS Level-1 trigger in proton-proton collisions at √s=13 TeV

    CMS Collaboration, “Performance of the CMS Level-1 trigger in proton-proton collisions at √s=13 TeV”,J. Instrum.15(2020) P10017, doi:10.1088/1748-0221/15/10/P10017

    doi:10.1088/1748-0221/15/10/p10017 2020

  27. [27]

    The CMS trigger system

    CMS Collaboration, “The CMS trigger system”,J. Instrum.12(2017) P01020, doi:10.1088/1748-0221/12/01/P01020. 22

    doi:10.1088/1748-0221/12/01/p01020 2017

  28. [28]

    Performance of the CMS high-level trigger during LHC Run 2

    CMS Collaboration, “Performance of the CMS high-level trigger during LHC Run 2”,J. Instrum.19(2024) P11021,doi:10.1088/1748-0221/19/11/P11021

    doi:10.1088/1748-0221/19/11/p11021 2024

  29. [29]

    Performance of the CMS muon detector and muon reconstruction with proton-proton collisions at √s=13 TeV

    CMS Collaboration, “Performance of the CMS muon detector and muon reconstruction with proton-proton collisions at √s=13 TeV”,J. Instrum.13(2018) P06015, doi:10.1088/1748-0221/13/06/P06015

    doi:10.1088/1748-0221/13/06/p06015 2018

  30. [30]

    Description and performance of track and primary-vertex reconstruction with the CMS tracker

    CMS Collaboration, “Description and performance of track and primary-vertex reconstruction with the CMS tracker”,J. Instrum.9(2014) P10009, doi:10.1088/1748-0221/9/10/P10009

    doi:10.1088/1748-0221/9/10/p10009 2014

  31. [31]

    Performance of reconstruction and identification ofτleptons decaying to hadrons andν τ in pp collisions at √s=13 TeV

    CMS Collaboration, “Performance of reconstruction and identification ofτleptons decaying to hadrons andν τ in pp collisions at √s=13 TeV”,J. Instrum.13(2018) P10005,doi:10.1088/1748-0221/13/10/P10005

    doi:10.1088/1748-0221/13/10/p10005 2018

  32. [32]

    Jet energy scale and resolution in the CMS experiment in pp collisions at 8 TeV

    CMS Collaboration, “Jet energy scale and resolution in the CMS experiment in pp collisions at 8 TeV”,J. Instrum.12(2017) P02014, doi:10.1088/1748-0221/12/02/P02014

    doi:10.1088/1748-0221/12/02/p02014 2017

  33. [33]

    Performance of missing transverse momentum reconstruction in proton-proton collisions at √s=13 TeV using the CMS detector

    CMS Collaboration, “Performance of missing transverse momentum reconstruction in proton-proton collisions at √s=13 TeV using the CMS detector”,J. Instrum.14(2019) P07004,doi:10.1088/1748-0221/14/07/P07004

    doi:10.1088/1748-0221/14/07/p07004 2019

  34. [34]

    Pileup mitigation at CMS in 13 TeV data

    CMS Collaboration, “Pileup mitigation at CMS in 13 TeV data”,J. Instrum.15(2020) P09018,doi:10.1088/1748-0221/15/09/P09018

    doi:10.1088/1748-0221/15/09/p09018 2020

  35. [35]

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

    J. Alwall et al., “The automated computation of tree-level and next-to-leading order differential cross sections, and their matching to parton shower simulations”,J. High Energy Phys.2014(2014), no. 07, 079,doi:10.1007/JHEP07(2014)079

    doi:10.1007/jhep07(2014)079 2014

  36. [36]

    Parton distributions from high-precision collider data

    NNPDF Collaboration, “Parton distributions from high-precision collider data”,Eur. Phys. J. C77(2017) 663,doi:10.1140/epjc/s10052-017-5199-5

    doi:10.1140/epjc/s10052-017-5199-5 2017

  37. [37]

    An introduction to PYTHIA 8.2

    T. Sj ¨ostrand et al., “An introduction to PYTHIA 8.2”,Comput. Phys. Commun.191(2015) 159,doi:10.1016/j.cpc.2015.01.024

    doi:10.1016/j.cpc.2015.01.024 2015

  38. [38]

    Extraction and validation of a new set of CMS PYTHIA8 tunes from underlying-event measurements

    CMS Collaboration, “Extraction and validation of a new set of CMS PYTHIA8 tunes from underlying-event measurements”,Eur. Phys. J. C80(2020) 4, doi:10.1140/epjc/s10052-019-7499-4

    doi:10.1140/epjc/s10052-019-7499-4 2020

  39. [39]

    GEANT4—a simulation toolkit

    GEANT4 Collaboration, “GEANT4—a simulation toolkit”,Nucl. Instrum. Meth. A506 (2003) 250,doi:10.1016/S0168-9002(03)01368-8

    doi:10.1016/s0168-9002(03)01368-8 2003

  40. [40]

    Search for a standard model-like Higgs boson in the mass range between 70 and 110 GeV in the diphoton final state in proton-proton collisions at√s=13 TeV

    CMS Collaboration, “Search for a standard model-like Higgs boson in the mass range between 70 and 110 GeV in the diphoton final state in proton-proton collisions at√s=13 TeV”,Phys. Lett. B860(2025) 139067,doi:10.1016/j.physletb.2024.139067

    arXiv 2025

  41. [41]

    Jet reconstruction and performance using particle flow with the ATLAS detector

    ATLAS Collaboration, “Jet reconstruction and performance using particle flow with the ATLAS detector”,Eur. Phys. J. C77(2017) 466, doi:10.1140/epjc/s10052-017-5031-2

    work page doi:10.1140/epjc/s10052-017-5031-2 2017

  42. [42]

    Observation of the diphoton decay of the Higgs boson and measurement of its properties

    CMS Collaboration, “Observation of the diphoton decay of the Higgs boson and measurement of its properties”,Eur. Phys. J. C74(2014) 3076, doi:10.1140/epjc/s10052-014-3076-z. References 23

    doi:10.1140/epjc/s10052-014-3076-z 2014

  43. [43]

    Performance of photon reconstruction and identification with the CMS detector in proton-proton collisions at √s=8 TeV

    CMS Collaboration, “Performance of photon reconstruction and identification with the CMS detector in proton-proton collisions at √s=8 TeV”,J. Instrum.10(2015) P08010, doi:10.1088/1748-0221/10/08/P08010

    doi:10.1088/1748-0221/10/08/p08010 2015

  44. [44]

    Supervised jet clustering with graph neural networks for Lorentz boosted bosons

    X. Ju and B. Nachman, “Supervised jet clustering with graph neural networks for Lorentz boosted bosons”,Phys. Rev. D102(2020) 075014, doi:10.1103/PhysRevD.102.075014

    work page doi:10.1103/physrevd.102.075014 2020

  45. [45]

    Performance of top-quark and W-boson tagging with ATLAS in Run 2 of the LHC

    ATLAS Collaboration, “Performance of top-quark and W-boson tagging with ATLAS in Run 2 of the LHC”,Eur. Phys. J. C79(2019) 375, doi:10.1140/epjc/s10052-019-6847-8

    work page doi:10.1140/epjc/s10052-019-6847-8 2019

  46. [46]

    Energy flow networks: Deep sets for particle jets

    P . T. Komiske, E. M. Metodiev, and J. Thaler, “Energy flow networks: Deep sets for particle jets”,J. High Energy Phys.2019(2019), no. 01, 121, doi:10.1007/JHEP01(2019)121

    doi:10.1007/jhep01(2019)121 2019

  47. [47]

    A simulation study to distinguish prompt photon fromπ 0 and beam halo in a granular calorimeter using deep networks

    S. Ghosh et al., “A simulation study to distinguish prompt photon fromπ 0 and beam halo in a granular calorimeter using deep networks”,J. Instrum.14(2019) P01011, doi:10.1088/1748-0221/14/01/P01011

    work page doi:10.1088/1748-0221/14/01/p01011 2019

  48. [48]

    Extracting low energy signals from raw LArTPC waveforms using deep learning techniques — A proof of concept

    L. Uboldi et al., “Extracting low energy signals from raw LArTPC waveforms using deep learning techniques — A proof of concept”,Nucl. Instrum. Meth. A1028(2022) 166371, doi:10.1016/j.nima.2022.166371

    arXiv 2022

  49. [49]

    End-to-end jet classification of quarks and gluons with the CMS Open Data

    M. Andrews et al., “End-to-end jet classification of quarks and gluons with the CMS Open Data”,Nucl. Instrum. Meth. A977(2020) 164304, doi:10.1016/j.nima.2020.164304

    arXiv 2020

  50. [50]

    Deep neural network for pixel-level electromagnetic particle identification in the MicroBooNE liquid argon time projection chamber

    MicroBooNE Collaboration, “Deep neural network for pixel-level electromagnetic particle identification in the MicroBooNE liquid argon time projection chamber”,Phys. Rev. D99(2019) 092001,doi:10.1103/PhysRevD.99.092001

    work page doi:10.1103/physrevd.99.092001 2019

  51. [51]

    Graph neural networks for particle reconstruction in high energy physics detectors

    Exa.TrkX Collaboration, X. Ju et al., “Graph neural networks for particle reconstruction in high energy physics detectors”, inProc. 33rd Annual Conf. Neural Information Processing Systems. 2020.arXiv:2003.11603

    Pith/arXiv arXiv 2020

  52. [52]

    Mini-AOD: A new analysis data format for CMS

    CMS Collaboration, “Mini-AOD: A new analysis data format for CMS”,J. Phys. Conf. Ser. 664(2015) 072052,doi:10.1088/1742-6596/664/7/072052

    work page doi:10.1088/1742-6596/664/7/072052 2015

  53. [53]

    The NanoAOD event data format in CMS

    CMS Collaboration, “The NanoAOD event data format in CMS”,J. Phys. Conf. Ser.1525 (2020) 012038,doi:10.1088/1742-6596/1525/1/012038

    work page doi:10.1088/1742-6596/1525/1/012038 2020

  54. [54]

    A dynamic reduction network for point clouds

    L. Gray, T. Klijnsma, and S. Ghosh, “A dynamic reduction network for point clouds”, 2020.arXiv:2003.08013

    Pith/arXiv arXiv 2020

  55. [55]

    Deep Learning

    I. Goodfellow, Y. Bengio, and A. Courville, “Deep Learning”. MIT Press, 2016. http://www.deeplearningbook.org

    2016

  56. [56]

    Dynamic graph CNN for learning on point clouds

    Y. Wang et al., “Dynamic graph CNN for learning on point clouds”, 2018. arXiv:1801.07829

    Pith/arXiv arXiv 2018

  57. [57]

    Measurement of the inclusive W and Z production cross sections in pp collisions at √s=7 TeV

    CMS Collaboration, “Measurement of the inclusive W and Z production cross sections in pp collisions at √s=7 TeV”,J. High Energy Phys.2011(2011), no. 10, 132, doi:10.1007/JHEP10(2011)132. 24

    doi:10.1007/jhep10(2011)132 2011

  58. [58]

    Handbook of LHC Higgs cross sections: 4. Deciphering the nature of the Higgs sector

    LHC Higgs Cross Section Working Group, “Handbook of LHC Higgs cross sections: 4. Deciphering the nature of the Higgs sector”,CERN Yellow Rep. Monogr.2(2017) 1, doi:10.23731/CYRM-2017-002

    doi:10.23731/cyrm-2017-002 2017

  59. [59]

    Precision luminosity measurement in proton-proton collisions at√s=13 TeV in 2015 and 2016 at CMS

    CMS Collaboration, “Precision luminosity measurement in proton-proton collisions at√s=13 TeV in 2015 and 2016 at CMS”,Eur. Phys. J. C81(2021) 800, doi:10.1140/epjc/s10052-021-09538-2

    doi:10.1140/epjc/s10052-021-09538-2 2015

  60. [60]

    CMS luminosity measurement for the 2017 data-taking period at√s= 13 TeV

    CMS Collaboration, “CMS luminosity measurement for the 2017 data-taking period at√s= 13 TeV”, CMS Physics Analysis Summary CMS-PAS-LUM-17-004, 2018

    2017

  61. [61]

    CMS luminosity measurement for the 2018 data-taking period at√s= 13 TeV

    CMS Collaboration, “CMS luminosity measurement for the 2018 data-taking period at√s= 13 TeV”, CMS Physics Analysis Summary CMS-PAS-LUM-18-002, 2019

    2018

  62. [62]

    Procedure for the LHC Higgs boson search combination in Summer 2011

    ATLAS and CMS Collaborations, and LHC Higgs Combination Group, “Procedure for the LHC Higgs boson search combination in Summer 2011”, Technical Report CMS-NOTE-2011-005, ATL-PHYS-PUB-2011-11, 2011

    2011

  63. [63]

    Confidence level computation for combining searches with small statistics

    T. Junk, “Confidence level computation for combining searches with small statistics”, Nucl. Instrum. Meth. A434(1999) 435,doi:10.1016/S0168-9002(99)00498-2

    doi:10.1016/s0168-9002(99)00498-2 1999

  64. [64]

    Presentation of search results: TheCL s technique

    A. L. Read, “Presentation of search results: TheCL s technique”,J. Phys. G28(2002) 2693, doi:10.1088/0954-3899/28/10/313

    doi:10.1088/0954-3899/28/10/313 2002

  65. [65]

    Asymptotic formulae for likelihood-based tests of new physics

    G. Cowan, K. Cranmer, E. Gross, and O. Vitells, “Asymptotic formulae for likelihood-based tests of new physics”,Eur. Phys. J. C71(2011) 1554, doi:10.1140/epjc/s10052-011-1554-0. [Erratum: doi:10.1140/epjc/s10052-013-2501-z]

    doi:10.1140/epjc/s10052-011-1554-0 2011

  66. [66]

    The CMS statistical analysis and combination tool: COMBINE

    CMS Collaboration, “The CMS statistical analysis and combination tool: COMBINE”, Comput. Softw. Big Sci.8(2024) 19,doi:10.1007/s41781-024-00121-4. 25 A The CMS Collaboration Yerevan Physics Institute, Yerevan, Armenia A. Hayrapetyan, V . Makarenko , A. Tumasyan1 Institut f ¨ ur Hochenergiephysik, Vienna, Austria W. Adam , J.W. Andrejkovic, L. Benato , T. ...

    doi:10.1007/s41781-024-00121-4 2024