Reinterpretation of LHC results for new physics: unlocking the LHC legacy

10 Nov 2026, 14:00 → 13 Nov 2026, 17:00 Europe/Zurich

13/2-005 (CERN)

Felicia Volle (University of Birmingham (GB)), Jonas Wurzinger (Technische Universitat Munchen (DE)), Sabine Kraml (LPSC Grenoble), Sezen Sekmen (Kyungpook National University (KR))

With the LHC physics runs concluded, it is crucial to ensure that the plethora of analyses carried out have a lasting impact beyond their paper publications, enabling new research based on reinterpretation and reuse of these important (and unique!) data. 

This workshop marks the first general meeting of the LHC Reinterpretation (REI) Working Group, which succeeds the LHC Reinterpretation Forum (RIF). Its goal is to review recent developments in tools, phenomenology, and experimental approaches, with a particular focus on data and analysis preservation and reuse.

In addition to presentations on the WG's activities, in particular from the ongoing task forces, we welcome contributions on

• Development of public tools 
• Phenomenological reinterpretation studies
• Analysis reproducibility, incl. reusable ML  
• Open Data usage  
• any other topic relevant to the reinterpretation of LHC results 

Contributions from other LHC working groups are strongly encouraged to foster cross-collaboration!

---

The current timetable is PRELIMINARY and may still change a bit in the next 2 weeks.

---

Background

The REI Working Group provides a platform for ongoing discussion of topics related to beyond-the-Standard-Model (BSM) reinterpretation of LHC data, including the development of public recasting tools and supporting infrastructure.

To date, nine RIF workshops have been held at CERN, Fermilab, Imperial College London, and IPPP Durham. These efforts led to key community outputs, including:

• Reinterpretation of LHC Results for New Physics: Status and recommendations after Run 2 (2020, SciPostPhys.9.2.022)
• Publishing statistical models: Getting the most out of particle physics experiments (2021, SciPostPhys.12.1.037)
• Snowmass input: Data and analysis preseration, recasting and reinterpretation (2022, arXiv:2203.10057)
• European Strategy for Particle Physics Update (ESPPU) input: Reinterpretation and preservation of data and analyses in HEP (2025, arXiv:2504.00256).

Since 2025, the Forum has been integrated into the LPCC-organised LHC BSM working group.

** This edition marks the 10th anniversary since the first reinterpretation workshop in 2016 **

Tuesday 10 November

14:00 → 14:15 General organisation: Welcome and introduction

Conveners: Felicia Volle (University of Birmingham (GB)), Jonas Wurzinger (Technische Universitat Munchen (DE)), Sabine Kraml (LPSC Grenoble), Sezen Sekmen (Kyungpook National University (KR))

14:15 → 18:00 Reinterpretation studies (pheno): Block 1

14:15 ATLAS triple Higgs search in CheckMATE and Rivet

I will discuss implementation of the ATLAS triple Higgs search in CheckMATE and Rivet. Several examples of application to BSM models will be shown together with expectations at the HL-LHC.

Speakers: Krzysztof Rolbiecki (University of Warsaw), Tomasz Procter (Jagiellonian University (PL))

14:35 Constraints on Vector-Like Top Dipole Interactions from Top-Associated Photon Measurements at the LHC

Precision measurements of top-quark production with energetic photons provide a complementary probe of vector-like quarks with nonstandard radiative decays. We reinterpret the CMS differential tt̄γ measurement and the ATLAS fiducial tt̄γγ cross section to constrain the electromagnetic and chromomagnetic transition interactions of a charge-2/3 vector-like top partner, T . We consider effective interactions written in terms of physical fields after electroweak symmetry breaking, retaining only the decays T → tγ and T → tg. The CMS signal includes both the mixed (tγ)(tg) topology and double-radiative tt̄γγ events entering the exactly-one-photon fiducial region, while the ATLAS rate is interpreted using a signal-specific detector-response transfer factor. Signal uncertainties are included, and the interpretation is restricted point by point to the domain satisfying the EFT and narrow-width
requirements. Denoting the electromagnetic and chromomagnetic transition coefficients by ctγ and ctg , respectively, we find that, for ctg = 1.0 × 10−5 GeV−1 and mT = 500 GeV, the observed Run 2 exclusions at 95% confidence level extend down to ctγ ≃ 4.7 × 10−7 GeV−1 for CMS and 2.3 × 10−6 GeV−1 for ATLAS within the valid domain. At mT = 1000 GeV, the corresponding values are 2.1 × 10−6 and 8.5 × 10−6 GeV−1 . Projections to the HL-LHC indicate improved sensitivity to these radiative transition interactions. These results establish top-associated photon measurements as complementary probes of radiative top-partner
interactions beyond searches targeting conventional decay modes.

Speaker: Mohammad Sahraei (Institute for Research in Fundamental Sciences (IR))

14:55 Composite top partners in exotic colour representations

Publicly available recasts allow LHC searches to probe signals beyond those for which they were originally designed. I will discuss their application to colour-sextet fermionic resonances predicted in composite Higgs models with partial compositeness. We use existing MadAnalysis 5 recasts of ATLAS and CMS supersymmetry searches to study simulated sextet-pair production and its decay chains. These produce high-multiplicity final states rich in top and bottom quarks and, in some channels, missing transverse momentum. I will outline the signal simulation and show how the existing searches constrain the benchmark scenarios considered.

Speaker: Jan Hadlik (University Würzburg)

15:15 The recasting landscape of vector-like quarks

We present a recasting study of pair production of vector-like quarks. We consider decays into a third generation quark and either a SM boson or a color singlet BSM scalar of charge 0, 1 or 2. Motivated by composite Higgs models, we assume the scalars to decay to third generation quarks. We obtain recasting bounds from MadAnalysis5 and CheckMATE on all 43 combinations of decay channels. We highlight results on selected individual channels, then show how to combine bounds across decay channels. Our recasting results are publicly available, allowing constraints to be set on VLQ models without the need for additional recasting.

Speaker: Manuel Kunkel (University of Würzburg)

15:35 Coffee break

16:10 Constraining the higgsino mass in the MSSM by reinterpreting the limits set by electroweakino searches at the LHC

We present a method to set exclusion limits on the Minimal Supersymmetric Standard Model (MSSM) higgsino mass parameter $\mu$ using the experimental results from supersymmetric electroweak searches performed at the Large Hadron Collider (LHC).
The method uses available exclusion limits on the two lightest neutralino masses ($\tilde{\chi}_1^0$, $\tilde{\chi}_2^0$) together with the branching ratio of $\tilde{\chi}_2^0 \rightarrow Z\tilde{\chi}_1^0$. Exploiting the dependence of this branching ratio on $\mu$, $M_1$, $M_2$, and $\tan\beta$, we extend the experimental
limits into this parameter space.

An application of the method is also presented, combining the likelihoods made available by the ATLAS collaboration that target two relevant models, $WZ(2\ell 2j)$ and $Wh(1\ell bb)$, in which the $\tilde{\chi}_2^0$ is assumed to decay exclusively either by a $Z$ or a $h$ boson, following the prescription of the simplified models.
By scanning over the relative branching ratio between the two channels, exclusion limits are obtained not only as a function of the neutralino masses, but also of $\mathrm{BR}(\tilde{\chi}_2^0 \rightarrow Z\tilde{\chi}_1^0)$ itself.

This procedure makes full use of the information available from collider searches,
extending its reach to the MSSM parameter $\mu$ as well as to other fundamental
parameters such as $\tan\beta$, while requiring minimal computational resources.

Furthermore, since $\mu$, $M_1$, $M_2$, and $\tan\beta$ also control the $h\tilde{\chi}_1^0\tilde{\chi}_1^0$ and $Z\tilde{\chi}_1^0\tilde{\chi}_1^0$ couplings, the new limits are translated into constraints on the Spin-Independent (SI) and Spin-Dependent (SD) neutralino-nucleon cross section $\sigma_{p,n}$, enabling a direct comparison between collider results and direct Dark Matter detection experiments (e.g. LUX-ZEPLIN). A comparison with existing ATLAS pMSSM scans targeting the same phase space is also presented.

Speaker: Annalisa Berti (Laboratory of Instrumentation and Experimental Particle Physics (PT))

16:30 Electroweak-inos in the pMSSM with SModelS v3.2

In arXiv:2512.14502, we discussed the coverage of the electroweakino sector of the phenomenological minimal supersymmetric standard model (pMSSM) with SModelS v3.0, comparing with the ATLAS pMSSM study and investigating what can be gained by a global likelihood analysis. I will present an update of this study based on the recently released SModelS v3.2. The major novelties are the inclusion of mono-jet, disappearing-tracks and soft-leptons searches from ATLAS and/or CMS, and a much faster statistical evaluation in SModelS owing to machine-learned likelihoods.

Speaker: Leo CONSTANTIN

16:50 Reinterpretation of searches for supersymmetry models with long-lived particles using the ATLAS experiment at the LHC

Traditional SUSY searches focus on either R-parity-conserving or prompt R-parity-violating (RPV) scenarios, leaving a gap for long-lived lightest supersymmetric particles (LSPs). This study reinterprets 13 published ATLAS SUSY searches, including both prompt SUSY and long-lived particle searches, using the full Run 2 dataset of 140 fb⁻¹ at $\sqrt{s} = 13 \text{ TeV}$ to scan a wide parameter space of variable RPV couplings (λ, λ', λ''). The object systematic uncertainties are studied, and the automated RECAST framework is utilised for a subset of the searches. Limits on various sparticle masses are established at the 95% confidence level. These findings substantially broaden collider sensitivity to long-lived particle signatures.

Speakers: Valentina Tudorache (Horia Hulubei National Institute of Physics and Nuclear Engineering (RO)), Yuchen Cai (Universita e INFN, Bologna (IT))

17:10 Clustering for large-scale reinterpretation of new physics searches at the LHC

Beyond the Standard Model (BSM) searches show no statistically significant sign of new physics to date. However, several analyses reported small excesses, higher than 2σ SD beyond the SM expectation. In this work, clustering algorithms are used to extract more insight from existing searches and motivate a next round of BSM analyses. The flexible framework of the phenomenological Minimal Supersymmetric Standard Model (pMSSM) is used for a large-scale reinterpretation of new physics models using existing analyses. Models consistent with the observed excesses are explored using clustering methods. In this work, clustering algorithms are benchmarked for this high-dimensional problem. Optimal algorithms are deployed to perform a large-scale reinterpretation of new physics searches to help inform the direction of future searches at the HL-LHC.

Speaker: Judita Mamužić (Jožef Štefan Institute (SI))

Wednesday 11 November

09:00 → 12:30 Experiment(-related) contributions, incl. Open Data

09:00 ATLAS efforts regarding reinterpretation

Speaker: Tomasz Procter (Jagiellonian University (PL))

09:30 CMS efforts regarding reinterpretation

10:00 Coffee break

10:30 LHCb efforts regarding reinterpretation

Speaker: Jamie Gooding (CERN)

11:00 LHCb's open data

Speaker: Piet Nogga (University of Bonn (DE))

11:30 BSM at ALICE and efforts regarding reinterpretation

12:00 t.b.a. (excesses ...)

12:30 → 14:00 Lunch break

14:00 → 18:00 Experiment(-related) contributions, incl. Open Data

14:00 AgentRivet: an automated system for producing Rivet routines from journal publications

Particle physics collider experiments provide Rivet routines as part of the analysis preservation strategy for model-independent measurements. Rivet is a C++ toolkit that allow new theoretical models to be compared to the measurements, thus aiding the development and tuning of Monte Carlo event generators as well as searches for physics beyond the Standard Model. However, analysis coverage is known to be incomplete, with only 39% of measurements having documented and publicly available Rivet routines.

In this talk, we present the design and implementation of an automated workflow based on Large Language Models with the goal of providing the missing routines. This multi-step workflow, referred to as AgentRivet, extracts the physics analysis information from published papers and writes the missing Rivet routines, with intermediate code- and physics- reviews as part of an autonomous quality control. We report the results obtained using commercial Large Language Models, provided by OpenAI, Anthropic, and Google, for two recent measurements from the ATLAS and CMS experiments. Talk based on https://arxiv.org/abs/2606.13535.

Speakers: Christian Gutschow (UCL (UK)), Sukanya Sinha (Nikhef (NL))

14:25 Reinterpreting searches with generator-level transformer surrogates

Analyses that select events with ML classifiers acting on low-level reconstructed objects are challenging to reinterpret outside of experimental collaborations, because reproducing the selection requires both the trained model and a complete detector simulation. We describe a surrogate approach in which a transformer trained only on generator-level particles predicts the decision the detector-level tagger would have made, so that the signal-region efficiency of a new model follows from a truth record alone. We develop the method on a toy emerging-jet analysis simulated end to end with Pythia and Delphes, with a detector-level transformer jet tagger trained to discriminate from Standard Model QCD. A second transformer of the same architecture, seeing only the generator particles inside a truth jet, is trained to predict the decision of that tagger, and its per-jet probabilities are combined into the efficiency of the signal region. We report closure on held-out samples and on dark-sector parameter points never seen in training. The trained surrogates are served through PRISM, a public web service hosting a library of preserved analyses, where an uploaded HepMC file returns the predicted efficiency and new analyses are contributed by pull request.

Speaker: Jackson Carl Burzynski (University of Oklahoma (US))

14:50 Mapping Heavy-Flavour Decays with Self-Supervised Learning and LHCb Open Data

Can we map heavy-flavour decay environments directly from collision data, without specifying a particular decay hypothesis? We present a self-supervised transformer architecture that learns the structure of heavy-hadron hadronization and decays by predicting masked particle-identification information and reconstructing missing jet constituents. Using publicly available LHCb simulation and 2017 proton-proton collision Open Data, we demonstrate that such a map can be learned without flavour or exclusive-decay labels.

We investigate the physics encoded in this representation through controlled removal of reconstructed charm and beauty decay products. In simulation, removing correlated decay products produces a stronger response than removing kinematically matched tracks from the same jet. In collision data, we validate the approach across eight reconstructed heavy-flavour decay channels, observing enhanced responses in the signal regions relative to adjacent mass sidebands for the high-statistics charm modes.

Finally, we explore the learned map through similarity searches among anomalous jets, identifying recurring reconstructed topologies that can be subjected to further physics and detector cross-checks. These results establish a proof of principle for using LHCb Open Data to build a data-driven cartography of heavy-flavour decays, opening new opportunities to explore rare, unusual, and partially reconstructed decays, including signatures involving invisible particles.

Speaker: Marko Stamenkovic (Brown University (US))

15:20 Coffee break

15:50 Open Event Generation

Speaker: Enrico Bothmann (CERN)

16:20 A Common BSM Signal Benchmark Pool

Reinterpreting LHC results for new theories depends on access to well-documented signal simulations. Over the years, ATLAS has produced a large number of Beyond-the-Standard-Model (BSM) signal samples across many searches, but they remain scattered and hard to find, which limits their reuse. We present an effort to catalogue the publicly simulated BSM signal samples into a common, searchable pool, starting with resonant signatures. We analyse their metadata and classify each model by resonant observable, model category, production mode, decay topology and event context. This lets users find existing samples that match a given final state or model class. The pool can help validate recasting tools, compare model coverage across analyses, find regions of theory space that no simulated benchmark covers, and reuse existing samples rather than generating new ones. About 2,000 samples have already been released publicly through ATLAS Open Data, laying the groundwork for broader and more systematic reinterpretation of LHC searches.

Database and additional information: https://bsm-signal-pool.docs.cern.ch/

Speaker: Theresa Reisch (Universite de Geneve (CH))

16:30 Open Event Generation task force discussion

18:00 → 19:00 Reception

Thursday 12 November

09:00 → 12:30 Public tools, likelihoods and all that: Block 1

09:00 News from CheckMATE

I will present recent developments in CheckMATE.

Speaker: Krzysztof Rolbiecki (University of Warsaw)

09:20 Recent Developments in MadAnalysis 5 and the Road to MadAnalysis 6

I will present a short overview of recent developments in MadAnalysis 5, with emphasis on features relevant for LHC reinterpretation and phenomenological analyses. After a brief recap of the current framework, I will discuss ongoing work towards MadAnalysis 6 and highlight some of the new capabilities and structural developments currently being explored. The talk will therefore serve both as a status update and as a preview of the future evolution of the MadAnalysis ecosystem.

Speaker: Benjamin Fuks

09:40 The Analysis Description Language and CutLang: recent developments for LHC reinterpretation

We present recent developments in the Analysis Description Language (ADL) and its runtime interpreter CutLang in the context of reinterpretation. We report on validation studies of ATLAS and CMS searches, released as a public database with efficiency maps and validation material. We also introduce Bakery, a Snakemake-based rewrite of EM Creator, which runs the full efficiency-map production chain from a single configuration file. A major focus is the ADL Overlap Analyzer, which determines the relations between signal regions of different analyses directly from their ADL descriptions, without generating events. Cross-validation with an event-based method using CutLang confirms all its predicted nesting relations. Finally, we discuss ongoing work on translating public analysis codes, starting with SimpleAnalysis, into ADL using AI agents based on locally fine-tuned open-weight models, which will also provide realistic inputs for the ADL overlap analyzer.

Speaker: Ekin Yoruk (Boğaziçi University (TR))

10:00 Coffee break

10:30 ENCORE: reinterpreting and combining public LHC likelihoods

ATLAS and CMS publish the statistical models and efficiency maps of many Run 2
searches, so new models can be tested and searches combined from outside the
collaborations. We present ENCORE, a framework that turns such a study into one
validated configuration file and checks every result against a published one. Where
ATLAS has published the answer, we reproduce it within a few percent, and we show
limits that were never published, including joint fits of ATLAS and CMS likelihoods.
Because each model point costs seconds, the same approach can scan the whole public
record and show the next generation of searches where coverage ends and what would close each gap.

Speaker: Theresa Reisch (Universite de Geneve (CH))

10:50 Rivet platform developments for BSM

The Rivet MC-analysis toolkit is being increasingly used for BSM studies as well as measurements. This is supported by evolutions of the machinery for truth observables to also support detector emulation, lifetime-based selections of final-state particles sharing much in common with long-lived particle signatures, and driven by more complex searches relying on detailed jet structure or particle-level machine-learning classifiers. In this short talk I will update on the status of the Rivet core, particularly the SimpleAnalysis interface for reco/search-analysis use, the extension of particle-selection cuts to include lifetime and vertexing variables, and increasing support for using preserved ML algorithms via ONNX Runtime.

Speaker: Andy Buckley (University of Glasgow (GB))

11:10 Reusing preserved collider analyses for new physics constraints: a Contur update

The LHC is a unique machine, it is essential that its measurements remain reusable and accessible, so that new theories and improved calculations can be confronted with data even if they were not conceived of at the time the measurements were made. This is the challenge addressed by Contur (Constraints On New Theories Using Rivet). Using a library of model-independent analyses preserved in Rivet, Contur allows rapid statistical testing of new models against published results from across the LHC measurement programme.

In this talk, we will present an overview of the Contur method, highlighting some recent physics results. We will then discuss recent developments to the software, including an overhaul of the statistical backend, allowing Contur to perform fits to more than one parameter of interest. We will conclude with efforts to aid interoperability between Contur and the wider ecosystem of analysis preservation and re-interpretation software.

Speaker: Jonathan Butterworth (University College London (GB))

11:30 t.b.a. (tools discussion, agentic approches, ...)

12:30 → 14:00 Lunch break

14:00 → 18:00 Public tools, likelihoods and all that: Block 2 & task forces

14:00 OLLL: Open library of learned likelihoods from LHC experiments

Open statistical models released by the LHC experiments are transforming the reinterpretation of collider searches by providing access to the full likelihood information of experimental analyses. However, evaluating these likelihoods remains computationally expensive, particularly for analyses with many signal and control regions, limiting their use in large-scale phenomenological studies. In this contribution, we present the Open Library of Learned Likelihoods (OLLL), a collection of neural-network surrogates for profiled likelihoods from public LHC statistical models. The surrogate models are trained on likelihood evaluations generated with pyhf, optimized using a heteroscedastic loss that enables them to predict both the profiled likelihood and an associated uncertainty estimate, and serialized in the framework-independent ONNX format for seamless integration into reinterpretation frameworks. We demonstrate the approach on five ATLAS SUSY searches of increasing complexity, where the learned likelihoods accurately reproduce both the official ATLAS and full pyhf exclusion contours while reducing the computational cost of likelihood evaluation by orders of magnitude. The resulting library provides a practical and extensible infrastructure for fast and statistically faithful reinterpretation of LHC searches

Speaker: Dr Humberto Reyes-González (RWTH Aachen)

14:20 Spey: smooth inference for reinterpretation studies

Reinterpreting LHC results depends on access to publishable, reusable likelihoods, yet these are distributed across a growing range of statistical frameworks, including simplified likelihoods, machine-learned statistical models, HistFactory-style JSON, RooFit workspaces, and CMS Combine datacards. Each framework has its own conventions and interfaces, making it difficult to develop analysis-independent reinterpretation tools. Spey addresses this challenge through a backend-agnostic, plug-in-based statistical inference framework. Once a model can provide a log-probability, Spey provides a common interface for hypothesis testing, exclusion limits, and profile-likelihood inference, independent of the framework in which the model was constructed.

In this talk, I will introduce Spey and present several developments since its original release. I will first discuss new and ongoing integrations with RooFit, HS3, and CMS Combine, extending the range of likelihood representations that can be used directly within Spey. I will then introduce support for multi-parameter-of-interest fits across multiple histograms with correlated nuisance parameters, enabling global EFT fits within the same framework. Together, these developments bring Spey closer to a unified statistical interface for HEP reinterpretation, allowing likelihoods from different statistical frameworks to be analysed using the same inference tools. I will demonstrate the new functionality through worked examples and discuss the remaining challenges in achieving robust interoperability across HEP statistical frameworks.

Speaker: Jack Y. Araz (UCL & City St George's, UoL)

14:40 HS3 review

Speaker: Simon Cello (Technische Universitaet Dortmund (DE))

15:05 HS3 task force discussion

15:30 Coffee break

16:00 pMSSM task force discussion

Friday 13 November

09:00 → 12:00 Reinterpretation studies (pheno): Block 2

09:00 BSM Interpretations of Multitop Production at CMS

Multitop production at the LHC provides a powerful probe of physics beyond the Standard Model, offering sensitivity to new interactions involving the top quark. This contribution presents recent CMS results from searches for three- and four-top-quark production in proton–proton collisions at a center-of-mass energy of 13 TeV, using the full Run 2 dataset corresponding to an integrated luminosity of 138/fb. The results are interpreted in several BSM scenarios, including constraints on dimension-six operators within the Standard Model Effective Field Theory, searches for heavy resonances coupled preferentially to top quarks, and modifications of the top Yukawa interaction involving CP-even and CP-odd couplings.

Speaker: Mojtaba Mohammadi Najafabadi (Institute for Research in Fundamental Sciences (IR))

09:20 Probing Matter Unification through Higgs Physics at the LHC

We investigate the Higgs sector phenomenology of the minimal low scale quark-lepton unification theory. In addition to the Standard Model-like Higgs boson, the theory predicts a new CP-even neutral Higgs, a CP-odd neutral Higgs, and two charged Higgses, together with heavy right-handed neutrinos and leptoquarks. We analyze the scalar spectrum, the Yukawa structure responsible for charged-fermion masses, and the neutrino sector realized through an inverse-seesaw mechanism, emphasizing that the couplings of the new Higgs bosons are directly tied to the structure of matter unification and therefore lead to characteristic decay patterns dominated by third-generation fermions. We compute the branching ratios and production cross sections of the new Higgs states at the Large Hadron Collider (LHC) and assess their impact on present and future collider measurements. We find that current LHC data probe significant regions of parameter space for the neutral heavy Higgs bosons, with the strongest sensitivity arising from channels involving tau leptons, tops, and related final states, while the charged Higgs remains less constrained because of its electroweak production mechanism. Our results show that Higgs measurements provide a powerful and complementary probe of low-scale matter unification beyond direct leptoquark searches and open a new avenue for testing quark-lepton unification at present and future colliders, with discovery potential in high-luminosity LHC running.

Speaker: Mark Brettell (UCL)

09:40 From EFT Interpretation to Reinterpretation

Speaker: Peter Stangl (Johannes Gutenberg University Mainz)

10:10 Coffee break

10:40 Bounds on Massive Graviton-Like Particles from Searches for Photon-Coupled ALPs

Limits on spin-0 axion-like-particles (ALPs) coupling to photons are reinterpreted as constraints on massive spin-2 graviton-like-particles (GLPs) with universal coupling $\alpha_G/M_P$​ (where $M_P$​ is the reduced Planck mass) to the Standard Model fields. A minimally model-dependent recasting is performed, exploiting the formally analogous production and detection mechanisms for both particle types, based on the Primakoff and Gertsenshtein effects, i.e., photon-axion/graviton conversion. Constraints originally derived in the ALP mass vs. photon-coupling plane $(m_a,g_{a\gamma}​)$ are translated into the corresponding bounds in the GLP $(m_G,\alpha_G/M_P​)$ parameter space over the full mass range, $m_{a,G} = 10^{-20}-10^{14}$ eV probed in current and future experimental setups including cavity-based detectors (haloscopes and resonant upconversion devices), helioscopes, magnetometers, optical interferometers, beam dumps, fixed-target, and collider experiments, as well as astrophysical and cosmological constraints. Whereas current ALP searches do not set stronger bounds on massive spin-2 particles than fifth-force tests, future magnetometers, two-beam interferometers, and upconversion experiments have the potential to provide very strong sensitivity, down to $\alpha_G/M_P \approx 10^{-32}$ GeV$^{-1}$, for light graviton-like particles with $m_G\lesssim10^{-8}$ eV. These future detectors exhibit comparatively greater sensitivity to massive gravitons than to axions. For massive gravitons at the TeV scale, exclusive diphoton decay searches, employed in ALP studies at the LHC, offer a complementary approach to standard searches for spin-2 resonances in other inclusive final states.

Biblio: J. Gue and D. d'Enterria, "Bounds on massive graviton-like particles from searches for axion-like particles coupling to photons", JHEP to appear; enter link description here[arXiv:2605.00549 [hep-ph]]

Speaker: David d'Enterria (CERN)

11:00 NJL-Composite Model at LHC and HL-LHC

Within a new physics scenario, four-fermion operators of the Nambu–Jona-Lasinio (NJL) type exhibit a strong-coupling ultraviolet (UV) fixed point at the TeV scale [1]. At this scale, two Standard Model (SM) elementary fermions, form a bound state, resulting in a composite bosons, which couples to its constituents via effective contact interactions at the compositeness scale $\Lambda \approx \mathcal{O}(\text{TeV})$ [2].
Specifically, we are exploring the production mechanisms of composite scalar particles that decay into a diphoton final state. A central goal is to extract constraints on these final states using existing experimental data.
The CMS analysis [3] is translated into a MadAnalysis recast code, which defines the exact event selection cuts, object definitions, and background estimation procedures as reported by the experiment. This recast will allow us to assess the sensitivity of current LHC data to the NJL-inspired scenario and identify the most stringent bounds, thereby guiding future searches at the High-Luminosity LHC.

Speaker: Ms Sehar Ajmal (Universita e INFN, Perugia (IT))

11:20 Rethinking search signatures: Hadronic decays of GeV-scale feebly coupled particles

Searches for GeV-scale feebly interacting particles commonly target simple signatures with two charged particles. Hadronic modes often dominate, but their simulation relies on fragmentation models calibrated at much higher energies, which can populate symmetry-forbidden states and misidentify the dominant allowed channels. We present exHad, a publicly available hadronization model constrained by hadronic data, conservation laws and exclusive decay calculations, for use in Pythia-based simulations. Using SHiP as an example, we show that the dominant accepted signatures can shift from two-particle to multiparticle decays, including mixed charged-neutral states. The model is also applied to the ATLAS $h\to Z+a$ search, with a hadronically decaying light resonance $a$, where we demonstrate substantial changes in jet composition after jet and track kinematic cuts. These results highlight the importance of realistic modelling of hadronic decays for reinterpreting existing LHC searches and designing future ones.

Speaker: Maksym Ovchynnikov (CERN)

11:40 t.b.a.

12:00 → 12:30 General organisation: Final discussion

Conveners: Felicia Volle (University of Birmingham (GB)), Jonas Wurzinger (Technische Universitat Munchen (DE)), Sabine Kraml (LPSC Grenoble), Sezen Sekmen (Kyungpook National University (KR))