Search for soft unclustered energy patterns containing muons in the final state in pp collisions at sqrt s = 13 TeV with the ATLAS detector

arXiv:2605.20015 · hep-ex · Submitted 2026-05-19 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: High Energy Physics: Experiment. The day in high-energy experiment: detectors, searches and limits.

Kai: We're looking at the paper: "Search for soft unclustered energy patterns containing muons in the final state in pp collisions at sqrt s = 13 TeV with the ATLAS detector".

Mira: Here's the gist of the paper. My goal is to present a detailed, accurate description suitable for understanding the core methodology, results, and conclusions of the study.

Kai: Here is the combined summary: This research details an ATLAS search targeting Soft Unclustered Energy Patterns (SUEPs), which are hypothesized high-multiplicity, isotropic distributions of low-momentum particles potentially originating from strongly-coupled hidden sector theories.

Mira: The analysis specifically investigates SUEP signatures arising from gluon–gluon fusion production of a scalar mediator within Hidden Valley scenarios, utilizing 140 fb−1 of proton–proton collision data collected by the ATLAS detector at sqrt s = 13 TeV between 2015 and 2018.

Kai: The event selection strategy is highly specialized, focusing on prompt final states characterized by a high multiplicity of muons.

Paper summary: Kai: So to wrap up this paper, the title "Search for soft unclustered energy patterns containing muons in the final state in pp collisions at sqrt s = thirteen TeV with the ATLAS detector" points directly to what they did here <ref:2605.20015#pg1,Search for soft unclustered energy patterns containing muons in the final state>.

Mira: It’s a search specifically looking for those SUEP signatures that involve muons as decay products, using that specific set of collision data from ATLAS at thirteen TeV <ref:2605.20015#pg1>.

Lev: What does this all mean for us who are building the real hardware? Does finding nothing mean we can stop looking in this area, or does it guide where we need to focus our error correction efforts?

Kai: It shows that multi-muon based strategies are a viable way to search for these specific types of signals at the LHC. It’s a concrete experimental approach that works with the data they collected.

Mira: The implication is that we can place new constraints on the parameter space of Hidden Valley models, specifically those involving gluon–gluon fusion production of a scalar mediator.

Lev: So, for someone who just listens to this show, it means we’ve put a ceiling on how strongly these specific hidden sector couplings can be if they produce these SUEPs in this way. It’s not ruling out the physics entirely, but it narrows down the possibilities.

Kai: Exactly. It tells us that for a mediator mass of four hundred GeV, you can expect an upper limit on the cross section times branching fraction of zero point four fb based on these results from this paper <ref:2605.20015#pg1,for a mediator mass of>.

Mira: That number is specific and tied to that particular model assumption they used in their analysis. It’s not a general statement about all hidden sectors, but about this one pathway they explored.

Lev: So the future work would probably involve pushing those limits further by looking at different kinematic cuts or by using more sophisticated background estimation techniques to reduce those statistical uncertainties.

Kai: That seems right. They also flag that their method is limited because it can't fully capture all possible backgrounds, which is a standard caveat for any search like this.

Mira: And that limitation on capturing all backgrounds is what makes the search complementary; it probes regions where other searches might be less sensitive, especially those involving these specific muon signatures.

Lev: So we’re seeing how the theoretical predictions from Hidden Valley models are constrained by actual detector performance and data limitations at the LHC. It connects the theory to what’s actually measurable with hardware like ours.

Conclusion: Kai: So, we're wrapping up this look at that ATLAS search for soft unclustered energy patterns containing muons in proton–proton collisions at twelve teraelectronvolts.

Mira: It’s an ATLAS experiment looking for these very specific distributions of low-momentum particles, and the authors are exploring what those patterns might tell us about some hidden sector theories.

Lev: They used a lot of data from fifteen to eighteen years of runs, specifically focusing on high-multiplicity events with many muons.

Kai: The core idea is looking for these isotropic muon distributions, which they call SUEPs, because they think those signatures could come from gluon–gluon fusion producing some scalar mediator in Hidden Valley scenarios.

Mira: They set up a selection strategy to filter out the normal background from quantum chromodynamics multijet production by looking at things like muon sphericity and charged-particle track multiplicity.

Lev: That’s important because that background is huge, so they had to use a likelihood-based method called ABCD to estimate it, taking into account how those two observables are weakly correlated.

Kai: And what they found was actually pretty straightforward—they didn't see any significant deviation from the background-only hypothesis in that search region.

Mira: So instead of finding a signal, they ended up setting upper limits on the cross section times the branching fraction of that mediator into those SUEPs.

Lev: The limits depend heavily on what mass they assume for this mediator, like if it’s at seven hundred fifty gigaelectronvolts or four hundred gigaelectronvolts.

Kai: For example, if you assume a seven hundred fifty gigaelectronvolt mass for the mediator, their upper limit on the cross section times branching fraction is about zero point zero five femtobarns.

Mira: And when they assume it’s at one twenty-five gigaelectronvolts, which corresponds to the standard model Higgs boson, the upper limit on that branching fraction drops down to around zero point two percent.

Lev: The strongest constraints they found generally come from a mediator mass of three gigaelectronvolts, though sensitivity drops off if you look at much lower or much higher masses.

Kai: They also mentioned a specific model with several parameters—four hundred gigaelectronvolts for the mediator, zero point seven gigaelectronvolts for another parameter called m A', and three gigaelectronvolts for phi and T.

Mira: For that particular signal model, their fitted value for the cross section times branching fraction is calculated to be about one point six four femtobarns.

Lev: But they noted that the observed limits are a little weaker than what they expected, except for those signal models where the track multiplicity values were lower.

Kai: Overall, this search confirms that using multi-muon strategies works well and gives us a way to probe regions of particle production that other searches, which focus more on hadronic final states, might miss.

Mira: It’s a complementary probe to existing searches because it looks at different kinds of final states with different particle multiplicities.

Lev: This means we can put constraints on the Hidden Valley model parameters by looking at muon signatures rather than just looking at the standard quark and gluon jets.

Kai: The next step is figuring out how these constraints feed into our models for high-energy physics beyond the Standard Model.

The ATLAS Collaboration

EUROPEAN ORGANISATION FOR NUCLEAR RESEARCH (CERN)

hep-ex

Submitted: 2026-05-19

Updated: 2026-05-19

Comments: 45 pages in total, author list starting page 28, 12 figures, 3 tables, submitted to JHEP. All figures including auxiliary figures are available at https://atlas.web.cern.ch/Atlas/GROUPS/PHYSICS/PAPERS/EXOT-2018-56/

Journal ref: JHEP 10 (2026) 206

DOI: 10.1007/JHEP10(2026)206

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 86/100

The gist: My goal is to present a detailed, accurate description suitable for understanding the core methodology, results, and conclusions of the study.

Key concepts

Soft Unclustered Energy Patterns (SUEPs)
These are hypothesized energy distributions characterized by high multiplicity and isotropy of low-momentum particles. The researchers believe they could originate from strongly-coupled hidden sector theories, distinguishing them from standard QCD backgrounds.
Muon Sphericity
This observable measures the spatial distribution of muons in the rest frame of the muon system. It is used to discriminate signal events from background by identifying whether the muons are clustered isotropically or in a more directional pattern.
QCD Multijet Production Background
This is the dominant background process, where quarks and gluons interact to produce many particles. The search strategy specifically designed selection criteria to filter out this overwhelming QCD noise by favoring events with more isotropic muon distributions.
Upper Limits on Cross Section
Since no signal was definitively observed, the study provides upper limits on the product of a particle's production cross section and its branching fraction into SUEPs. These limits help constrain theoretical models predicting such new physics.

Terminology

Summary

My goal is to present a detailed, accurate description suitable for understanding the core methodology, results, and conclusions of the study.

Here is the combined summary:

This research details an ATLAS search targeting Soft Unclustered Energy Patterns (SUEPs), which are hypothesized high-multiplicity, isotropic distributions of low-momentum particles potentially originating from strongly-coupled hidden sector theories. The analysis specifically investigates SUEP signatures arising from gluon–gluon fusion production of a scalar mediator within Hidden Valley scenarios, utilizing 140 fb−1 of proton–proton collision data collected by the ATLAS detector at sqrt s = 13 TeV between 2015 and 2018. p. 0

Where the paper says it

A search for SUEPs whose decay products contain muons in the final state is presented using 140 fb−1 of proton–proton collision data at √ 𝑠 = 13 TeV collected by the ATLAS detector at the Large Hadron Collider during 2015–2018. The analysis targets SUEP signatures via gluon–gluon fusion production of a scalar mediator in Hidden Valley scenarios.

The event selection strategy is highly specialized, focusing on prompt final states characterized by a high multiplicity of muons. p. 2

Where the paper says it

The search presented here complements these approaches by targeting final states with a high multiplicity of prompt muons, which are expected for a broad range of dark-photon masses 𝑚𝐴′. This strategy provides strong background rejection while maintaining sensitivity to scenarios with moderate particle multiplicities, thereby extending the coverage of the SUEP parameter space well beyond that of previous searches.

Events were initially selected using multi-muon triggers and refined by requiring relatively low average muon transverse momentum, large charged-particle track multiplicity associated with the primary interaction, and an isotropic spatial distribution of the muons. p. 3
Where the paper says it

Events are selected using multi-muon triggers and target prompt final states with a high multiplicity of muons. They are further characterized by relatively low average muon transverse momentum, large charged-particle track multiplicity associated with the primary interaction, and isotropic spatial distribution of the muons.

This selection criteria are designed to preferentially isolate events exhibiting more isotropic muon distributions compared to the dominant background from Quantum Chromodynamics (QCD) multijet production. p. 3
Where the paper says it

They are further characterized by relatively low average muon transverse momentum, large charged-particle track multiplicity associated with the primary interaction, and isotropic spatial distribution of the muons. The signal region is defined to be sensitive to a broad range of charged-particle multiplicities and to preferentially select events in which muons are distributed more isotropically than in the dominant background from QCD multijet production, where muons are produced mainly in heavy-flavor decays, and which is estimated using a data-driven method.

To effectively discriminate between the signal and this dominant QCD background, two key observables were employed: muon sphericity—constructed in the rest frame of the muon system—and charged-particle track multiplicity, which was corrected for pile-up effects. p. 3

Where the paper says it

They are further characterized by relatively low average muon transverse momentum, large charged-particle track multiplicity associated with the primary interaction, and isotropic spatial distribution of the muons. The signal region is defined to be sensitive to a broad range of charged-particle multiplicities and to preferentially select events in which muons are distributed more isotropically than in the dominant background from QCD multijet production, where muons are produced mainly in heavy-flavor decays, and which is estimated using a data-driven method.

The background contribution was estimated using a likelihood-based ABCD method that leverages the weak correlation between these two observables. p. 7
Where the paper says it

As a result, the background is dominated by QCD multijet events, where muons are predominantly produced in decays of heavy-flavor hadrons, as well as kaons and pions. As described in Section 5, the background contribution is estimated using a likelihood-based ABCD method based on the muon sphericity 𝑆𝜇 and the track multiplicity 𝑁tracks, defined below.

Crucially, no significant deviation from the background-only hypothesis was observed within this search region. p. 0

Where the paper says it

The analysis targets SUEP signatures via gluon–gluon fusion production of a scalar mediator in Hidden Valley scenarios. No significant excess over the Standard Model expectation is observed.

Consequently, the results are presented in terms of upper limits set at the 95% confidence level on the product of the mediator production cross section (sigma S) and its branching fraction into a SUEP (B(S to SUEP)). p. 0
Where the paper says it

No significant excess over the Standard Model expectation is observed. Exclusion limits are set on the product of the mediator production cross section and the branching fraction for its decay into a SUEP, down to 0.

The derived exclusion limits vary significantly depending on the assumed mediator mass (m S). p. 0

Where the paper says it

Exclusion limits are set on the product of the mediator production cross section and the branching fraction for its decay into a SUEP, down to 0. 05 fb for a mediator mass of 𝑚𝑆 = 750 GeV, 0.

For m S = 750 GeV, the upper limit is approximately 0. p. 0
Where the paper says it

05 fb for a mediator mass of 𝑚𝑆 = 750 GeV, 0. 4 fb for 𝑚𝑆 = 400 GeV, and 70 fb for 𝑚𝑆 = 125 GeV; the last value, when identifying the mediator with the Standard Model Higgs boson, translates to an upper limit on the branching fraction of around 0.

05 fb. For m S = 400 GeV, the limit is set at 0. p. 0
Where the paper says it

05 fb for a mediator mass of 𝑚𝑆 = 750 GeV, 0. 4 fb for 𝑚𝑆 = 400 GeV, and 70 fb for 𝑚𝑆 = 125 GeV; the last value, when identifying the mediator with the Standard Model Higgs boson, translates to an upper limit on the branching fraction of around 0.

4 fb, and for m S = 125 GeV—when identifying the mediator with the Standard Model Higgs boson—the upper limit on the branching fraction is around 0. p. 0
Where the paper says it

05 fb for a mediator mass of 𝑚𝑆 = 750 GeV, 0. 4 fb for 𝑚𝑆 = 400 GeV, and 70 fb for 𝑚𝑆 = 125 GeV; the last value, when identifying the mediator with the Standard Model Higgs boson, translates to an upper limit on the branching fraction of around 0.

2%. The strongest constraints are generally found for a mediator mass of m phi = T = 3 GeV, although sensitivity is reduced at both lower and higher values of m phi and T. p. 2
Where the paper says it

A wide range of values for the dark-meson mass 𝑚𝜙 is theoretically motivated. Generally, values of order 1–10 GeV are well suited for SUEP scenarios, as they allow for a rich set of decay channels to SM particles while still producing the high-multiplicity, soft final states characteristic of this signature, which are not efficiently targeted by conventional searches.

Furthermore, the sensitivity analysis shows that the search performs better for a specific parameter space defined by m A' = 0.5 GeV (corresponding to branching fractions of 0. p. 4
Where the paper says it

A dark-photon mass of 𝑚𝐴′ = 0. 5 GeV, corresponding to branching fractions 𝐵(𝐴 ′ → 𝑒 + 𝑒 − , 𝜇+𝜇 − , 𝜋+𝜋 − ) = (0.

4 for mu+ mu- and pi+ pi- decays) compared to m A' = 0.7 GeV, where the muon branching fraction is reduced to 0. p. 0
Where the paper says it

Exclusion limits are set on the product of the mediator production cross section and the branching fraction for its decay into a SUEP, down to 0. 05 fb for a mediator mass of 𝑚𝑆 = 750 GeV, 0.

15, and less pronounced improvement is seen for m A' = 0.3 GeV (with a muon branching fraction of 0. p. 0
Where the paper says it

Exclusion limits are set on the product of the mediator production cross section and the branching fraction for its decay into a SUEP, down to 0. 05 fb for a mediator mass of 𝑚𝑆 = 750 GeV, 0.

5).

For a specific signal model—characterized by m S = 400 GeV, m A' = 0.7 GeV, m phi = 3 GeV, and T = 3 GeV—the fitted sigma S times B(S to SUEP) is calculated to be approximately 1. p. 2

Where the paper says it

Generally, values of order 1–10 GeV are well suited for SUEP scenarios, as they allow for a rich set of decay channels to SM particles while still producing the high-multiplicity, soft final states characteristic of this signature, which are not efficiently targeted by conventional searches. The momenta of the dark mesons are expected to follow an approximately thermal distribution [21] and are modeled using a Maxwell–Boltzmann spectrum parameterized by a Hagedorn temperature 𝑇 [22], which sets the characteristic momentum scale of the hadronizing hidden sector and is expected to be of the same order as 𝑚𝜙.

64 fb. The observed limits are noted as being slightly weaker than the expected limits, with the exception of signal models that predict lower track multiplicity values. p. 9
Where the paper says it

As shown in Figure 4, for signal models the value of 𝑆𝜇 is larger than that obtained when computed in the laboratory frame, while in data the 𝑆𝜇 distribution is shifted towards lower values compared to the laboratory-frame definition. 2 Track multiplicity The number of reconstructed tracks is a powerful observable for discriminating signal from background.

Overall, this search validates the effectiveness of multi-muon-based strategies and offers a complementary probe to existing searches that focus primarily on hadronic final states, extending sensitivity into regions characterized by moderate particle multiplicities and sizable muon fractions. p. 2
Where the paper says it

The search presented here complements these approaches by targeting final states with a high multiplicity of prompt muons, which are expected for a broad range of dark-photon masses 𝑚𝐴′. This strategy provides strong background rejection while maintaining sensitivity to scenarios with moderate particle multiplicities, thereby extending the coverage of the SUEP parameter space well beyond that of previous searches.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improve signal discrimination for hidden sector models by utilizing combined observables of muon sphericity and corrected track multiplicity to define search regions where the background is modeled using a multi-bin likelihood-based ABCD method. This allows for a more robust estimation of the dominant background from QCD multijet production, as seen in Figure 7.

  2. Enhance sensitivity to low dark-meson mass scenarios by optimizing trigger requirements, noting that The trigger efficiency ranges from approximately 1% for all signal models with mphi = 1.5 GeV and T = 1 GeV to nearly 100% for the model with M = 750 GeV, mA' = 0.5 GeV, and mphi = T = 5 GeV.

  3. Improve background modeling accuracy by incorporating a data-driven estimate of pile-up tracks, defined as Ntracks = Nbare tracks - NPU tracks, which accounts for the difference between the predicted and measured inelastic cross-sections. This correction is crucial for accurately modeling track multiplicity in high pile-up environments.

  4. Refine signal hypothesis testing by employing profile likelihood ratios that compare the background-only and signal-plus-background hypotheses to determine significance, as described in Section 7. This method ensures a consistent determination of background model normalization parameters simultaneously with the scale of the signal contribution.

  5. Establish stringent upper limits on new physics couplings by setting limits on the product of the mediator production cross section and the branching fraction for its decay into a SUEP, which currently reach values as low as approximately 0.05 fb for mS = 750 GeV, 0.4 fb for mS = 400 GeV, and 70 fb for mS = 125 GeV.

Abstract

Soft unclustered energy patterns (SUEPs) refer to high-multiplicity, isotropic distributions of low-momentum particles that could arise in strongly-coupled hidden sector theories. A search for SUEPs whose decay products contain muons in the final state is presented using 140 fb-1 of proton-proton collision data at sqrt s = 13 TeV collected by the ATLAS detector at the Large Hadron Collider during 2015-2018. The analysis targets SUEP signatures via gluon-gluon fusion production of a scalar mediator in Hidden Valley scenarios. No significant excess over the Standard Model expectation is observed. Exclusion limits are set on the product of the mediator production cross section and the branching fraction for its decay into a SUEP, down to 0.05 fb for a mediator mass of m S = 750 GeV, 0.4 fb for m S = 400 GeV, and 70 fb for m S = 125 GeV; the last value, when identifying the mediator with the Standard Model Higgs boson, translates to an upper limit on the branching fraction of around 0.2%.

Sources

Related papers