High Energy Physics - Experiment

Foundation models often obtain powerful results through very large unlabeled training datasets that dwarf available labeled datasets in size, but high energy physics already provides large samples of high-fidelity labeled simulation. We therefore ask whether self-supervised objectives improve over supervised pretraining alone when large labeled samples already exist. We introduce SuperJEPA, a joint-embedding predictive architec- ture augmented with a supervised objective, and find improved downstream jet-tagging performance across multiple tasks, model sizes, and pretraining dataset sizes. The gains persist for both frozen and fine-tuned encoders, motivating predictive self-supervision in future large-scale particle-physics foundation models.

Over the years, the COMPASS experiment at CERN has pursued an extensive SIDIS programme aimed at a comprehensive understanding of the structure of the nucleon, using a high-energy muon beam and polarised targets. Within this programme, measurements with a transversely polarised deuteron target were performed in 2022 using the 160 GeV/c muon beam. In single-hadron SIDIS on such a target, eight transverse-spin-dependent azimuthal asymmetries arise, providing sensitivity to several transverse-momentum-dependent parton distribution functions and to subleading-twist partonic correlations. We report the simultaneous extraction of all eight asymmetries from the complete 2022 deuteron sample. The Collins and Sivers asymmetries, previously presented using about 70% of the 2022 statistics, are updated with the full sample and found consistent with the earlier result, while the six remaining asymmetries are presented for the first time with the full 2022 statistics. Combined with the existing proton results, these measurements complete a comprehensive COMPASS programme on transverse-spin effects in SIDIS, providing essential input for global analyses.

We investigate the prospects for determining charm mixing, indirect $C\!P$ violation and hadronic parameters in charm decays with quantum-correlated neutral charm mesons at the proposed Super Tau-Charm Facility. The study is based on time-integrated double tag rates of coherent $D\overline{D}$ pairs produced at $\sqrt{s}=3.773$ and $4.030$ GeV, from $e^{+}e^{-}\to D^{(*)}\overline{D}^{(*)}$ processes. Here, one $D$ is tagged with the quasi flavour-specific decays $D\to K^-\pi^+\pi^+\pi^-$, $K^-\pi^+\pi^0$, and $K^-\pi^+$, while the other $\overline{D}$ is detected with the final states in flavour-specific mode, $C\!P$ eigenstate mode, or self-conjugate mode. The expected sensitivities are evaluated assuming an integrated luminosity of $1~{\rm ab}^{-1}$ at each energy point, demonstrating the capability on precise determination of the charm mixing parameters, the indirect $C\!P$ violation parameters, the strong-phase differences, and the coherence factors. The constraints of the hadronic-parameters provide unique inputs to the CKM angle $\gamma$ fit, while the measurements of charm mixing and indirect $C\!P$ violation parameters are complementary to the time-dependent measurements in the Belle II and LHCb experiments.

A commonly used technology for axion and other wave-like dark matter searches is a microwave resonator. In such experiments, the detected signal strength depends on the coupling between the resonator and the microwave transmission line that delivers power to the amplifier. In Axion Dark Matter eXperiment (ADMX) and other cavity searches, this coupling is typically achieved using a compact wire antenna inserted into the cavity. While this approach is adequate for right-cylindrical cavities, it can become difficult to efficiently extract power from large-volume, multi-wavelength ($V \gg \lambda^3$) resonators targeted for higher frequencies, such as those designed by ADMX-VERA (Volume Enhanced Resonant Axion). Here, we present a solution based on a microwave slot-antenna array capacitively coupled to a microstrip "summing network," which coherently combines signals from multiple transmission lines into a single output. This summing architecture is adapted from designs originally developed for cosmic microwave background (CMB) detectors. We demonstrate the design and characterization of a new high-frequency resonator that employs this readout scheme in the $19-21$ GHz range. Finally we discuss the potential to reach Dine-Fischler-Srednicki-Zhitnitsky (DFSZ) sensitivity using multiple summing network cavities fitted into a large volume.

Using $e^+e^-$ collision data collected with the BESIII detector at 27 center-of-mass energies between 4.26 and 4.95 GeV corresponding to a total integrated luminosity of $13.0~\text{fb}^{-1}$, we report the first search for the process $e^{+}e^{-} \to \pi^0 K^+K^-J/\psi$. We additionally probe the cross section spectra for potential charmonium-like vector states and invariant mass distributions for intermediate $Z_c/Z_{cs}$ state signals, and find no resonant structures consistent with these exotic candidates. Given the absence of observed resonant signals, upper limits at the 90\% confidence level are determined for the Born cross sections, along with their best-fit values, as well as for the production of the intermediate states.

Using a sample of $(2712.4 \pm 14.3) \times 10^6$ $\psi(3686)$ events collected with the BESIII detector, we perform a partial wave analysis of the decay $\psi(3686)\to\pi^{+}\pi^{-}\eta$ with the $\pi^{+}\pi^{-}$ invariant mass in the $\rho(770)^0$ mass region, [0.3, 2.2] $\mathrm{GeV}/c^2$. The cross section line-shape around the $\psi(3686)$ resonance is measured using energy-scan data sets, enabling the estimation of continuum contribution and its interference with the $\psi(3686)$ resonance. The branching fraction of $\psi(3686)\to\rho(770)^0\eta$ is determined to be $(1.34 \pm 0.03\pm0.13) \times 10^{-5}$. The $\psi(3686)\to a_{2}(1320)^{\pm}\pi^{\mp}$ decay is observed for the first time with a significance of $35.7\sigma$, and its partial branching fraction is measured to be $(2.13 \pm 0.05\pm0.32) \times 10^{-5}$. The $\eta$-$\eta^{\prime}$ pseudoscalar mixing angle $\theta_P$ is then found to be $(-11.2 \pm 1.6\pm3.6)^\circ$ or $(-19.4\pm 3.1\pm5.7)^\circ$, where the first uncertainty is statistical and the second systematic.

We present a precision measurement of the branching fraction of the rare decay $J/\psi\to\mu^+\mu^-\mu^+\mu^-$, and a search for a new gauge boson $Z^{\prime}$ that couples only to the second and third lepton generations, using a data sample of $(8998 \pm 39) \times 10^6$ $J/\psi$ events collected by the BESIII detector. The branching fraction of the $J/\psi\to\mu^+\mu^-\mu^+\mu^-$ decay is measured to be $B(J/\psi\to\mu^+\mu^-\mu^+\mu^-)=[0.91\pm0.04~{\rm (stat.)}\pm0.07~{\rm (syst.)}]\times 10^{-6}$, representing the most precise measurement to date and consistent with the theoretical prediction within one standard deviation. Additionally, we search for $CP$ asymmetry in the $J/\psi\to\mu^+\mu^-\mu^+\mu^-$ decay and determine the T-odd observable to be $A_T = [0.05\pm 0.04~{\rm (stat.)}\pm0.01~{\rm (syst.)}]$, which is consistent with $CP$ conservation. In the search for $J/\psi\to Z^{\prime}\mu^+\mu^-$ via $Z^{\prime}\to\mu^+\mu^-$, no significant $Z^{\prime}$ signal is observed in the $\mu^+\mu^-$ invariant mass region of $(0.212,~2.6)$~GeV$/c^2$, and we set upper limits on the ${\rm U}(1)_{L_{\mu}-L_{\tau}}$ model coupling constant $g^{\prime}$ ranging from 0.001 to 0.294 at the 90\% confidence level, which is comparable to the upper limits set by Belle II for $Z^{\prime}$ masses below $0.7$~GeV$/c^2$.

We report the most precise measurement to date of coherent elastic neutrino-nucleus scattering (CE$\nu$NS) from reactor antineutrinos using the CONUS+ high-purity germanium detectors at the Leibstadt nuclear power plant. With an exposure of 1296 kg$\cdot$days reactor-on, a blind likelihood analysis yields $1555 \pm 152$ CE$\nu$NS events, with a significance of 10.2 $\sigma$, in agreement with the Standard Model expectation of $1701 \pm 198$ events within 0.6 standard deviations. The result also determines the weak mixing angle at a momentum transfer of about 10 MeV, $\sin^2\theta_W = 0.225 \pm 0.022$, consistent with the Standard Model prediction.

Coherent elastic neutrino-nucleus scattering (CE$\nu$NS) provides a uniquely sensitive probe of neutrino properties, but its detection requires pushing energy thresholds to the sub-keV scale while controlling backgrounds. The CONUS+ experiment, located at the Leibstadt nuclear power plant, reported the first observation of CE$\nu$NS using reactor antineutrinos in 2025. Following a detector upgrade that replaced three of the original 1 kg high-purity germanium detectors with larger 2.42 kg diodes, this paper presents the analysis of the CONUS+ run 2 dataset, including the detector characterization, the background model, the signal prediction and expected sensitivity, and the extraction of the CE$\nu$NS signal. The new detectors reach a background level per unit mass reduced by approximately a factor of two relative to the original detectors, while reaching energy thresholds between 140 and 170 eV$_{ee}$. With an exposure of 1296 kg$\cdot$days reactor-on and 86 kg$\cdot$days reactor-off, a blind likelihood analysis fitting directly the number of signal counts yields $N_s = 1555 \pm 152$ CE$\nu$NS events, corresponding to a significance of 10.2$\,\sigma$. The result is in agreement with the Standard Model expectation of $1701 \pm 198$ events, with $N_s/N_{\mathrm{SM}} = 0.91 \pm 0.14$.

By analyzing $(2.712\pm0.014)\times10^9$ $\psi(3686)$ events collected by the BESIII detector, we report the first experimental study of the decays $\chi_{cJ}\to p K^- K^- \bar \Xi^{+}+c.c.~(J=0,1,2)$. The decays $\chi_{c1}\to p K^- K^- \bar \Xi^{+}+c.c.$ and $\chi_{c2}\to p K^- K^- \bar \Xi^{+}+c.c.$ are observed for the first time with significances of $9.0\sigma$ each, and their branching fractions are determined to be $\mathcal{B}_{\chi_{c1}\to p K^- K^- \bar \Xi^{+}+c.c.} =(9.6^{+2.2}_{-2.0}\pm0.8)\times10^{-6}$, and $\mathcal{B}_{\chi_{c2}\to p K^- K^- \bar \Xi^{+}+c.c.} =(7.5^{+1.8}_{-1.5}\pm0.5)\times10^{-6}$, where the first and second uncertainties are statistical and systematic, respectively. No significant signal is observed for $\chi_{c0}\to p K^- K^- \bar \Xi^{+}+c.c.$, and an upper limit on its branching fraction of $12.7\times10^{-6}$ is established at the 90\% confidence level.

The Short-Baseline Near Detector (SBND) is a liquid argon time projection chamber in Fermilab's Short-Baseline Neutrino Program, positioned 110 meters from the neutrino target along the Booster Neutrino Beam (BNB). In addition to its neutrino physics goals, SBND enables searches for new physics, including long-lived particles (LLPs) produced via neutrino-nucleus upscattering that decay into visible final states. Although neutrino data taking with the full SBND experiment began in December 2024, a subset of the Cosmic Ray Tagger (CRT) system operated from 2017--2018 as a standalone detector, referred to as the CRT Beam Telescope, collecting data corresponding to $2.95\times 10^{20}$ protons on target from the BNB. Originally intended for commissioning and performance studies, the substantial dataset collected also enables searches for displaced LLP decays. This paper presents the first search for physics beyond the Standard Model using data from the SBND CRT Beam Telescope, targeting LLP decays into electron-positron pairs. We observe no significant excess above the expected background and set 90% confidence-level upper limits on an effective production cross section at the level of $10^{-38}$ cm$^2$ for laboratory decay lengths of 0.3-100 m. These results complement existing searches by probing displaced decay signatures at meter-scale decay lengths.

Data analysis in particle physics often involves optimizing a loss function that quantifies how well a theoretical model describes the observed data, in order to estimate the parameters of the model. This is referred to as fitting. As datasets grow larger and more expensive analysis methods are used, the time required to perform this parameter estimation can be a limiting factor. This work explores techniques that can significantly accelerate the fitting process, and measures their performance on a set of example models chosen to be representative of a variety of physics fits. We compare the speed and precision of variants of several well-known optimization algorithms, namely Newton's method, BFGS, and the Migrad algorithm (commonly used in physics fits). We observe significant speedups when using some algorithms: most notably, some variants of BFGS are consistently about $10\times$ faster than Migrad across our benchmarks, while achieving the same precision. For quasi-Newtonian algorithms like BFGS and Migrad, we find that the initial value of the Hessian matrix has a large impact on the runtime. We additionally discuss methods for accelerating the evaluation of the loss function and its derivatives, using algebraic simplifications, automatic differentiation, and modern hardware features on the CPU and GPU. We find that basic algebraic simplifications of the loss function can significantly improve speed and precision. On the GPU we explore using mixed precision computation, and show that this achieves much higher throughput on most accelerators without significant precision loss. Lastly we discuss efficient parallelization strategies for both the CPU and the GPU. All of the techniques and algorithms in this paper are implemented in our new FastFit library, which was used for all of our benchmarks.

Recent years have seen a surge of interest in adapting non-classical states to search for physics beyond the Standard Model, including both qubit- and boson-based systems. These two classes of systems are closely related through the Holstein--Primakoff transformation (HPT). In this work, we show that the HPT offers novel insights into inherently quantum-mechanical detection of new physics. We begin by discussing how the HPT can be used to unify certain existing qubit- and boson-based proposals in a single framework. We then show how it naturally reveals a novel generalization of GHZ states which offer superior performance for the detection of rare interactions. We demonstrate that these novel states can be readily prepared, and combine the coherent advantage of GHZ states with superior error rejection. Finally, we show that the HPT also reveals a natural extension of the Hamiltonian-Not-in-Lindblad-Span theorem for collections of bosonic modes.

Machine learning tasks in particle physics are typically expressed in terms of regression, or relaxation toward some optimal solution as defined by an objective or loss function. In this paper, we introduce a large multiplicity expansion for analyzing regression tasks, which enables significant analytic understanding. For concreteness, we study regression with energy flow polynomials (EFPs), a complete linear basis of infrared and collinear safe observables on Lorentz invariant phase space. While the expectation value of any EFP scales like a constant in the large multiplicity limit, we prove that the covariance on uniform phase space of any two EFPs is highly suppressed and scales like the inverse of the squared multiplicity. This very unintuitive result relies on detailed cancellation guaranteed by the interplay of permutation invariance, momentum conservation, and rotational invariance. This scaling of the covariance with multiplicity further implies that the coefficients in an expansion of EFPs must scale quadratically in multiplicity. We present an example of regression with EFPs as applied to $e^+e^-\to$ hadrons events, and show how the regression can be solved analytically through the complementary expansions of the large multiplicity limit and of familiar fixed-order perturbation theory of QCD.

We present a unified projection of the sensitivity to the Arkani-Hamed-Dimopoulos-Dvali (ADD) model of large extra dimensions in the monojet plus missing transverse momentum final state at the HL-LHC ($\sqrt{s}=14$ TeV, 3 ab$^{-1}$), the HE-LHC (27 TeV, 15 ab$^{-1}$) and the FCC-hh (100 TeV, 30 ab$^{-1}$), for $n=2,3,4,6$ extra dimensions, with one selection, significance estimator and calibration throughout. Signal samples are generated with Pythia 8 and combined with the post-fit Standard-Model background yields released by ATLAS and CMS on HEPData; the expected significance combines the thirteen exclusive $E_{\mathrm{T}}^{\mathrm{miss}}$ bins of the ATLAS signal region, matching the binning of the published likelihood fit. Two elements are new. First, the energy scaling of the background to higher $\sqrt{s}$, including the $13\to14$ TeV step, is measured as a function of $E_{\mathrm{T}}^{\mathrm{miss}}$ from dedicated $Z(\nu\bar\nu)$+jets and $W(\ell\nu)$+jets samples rather than assumed constant; this reduces the projected FCC-hh reach by 39% relative to a constant-factor treatment. Second, the framework is anchored to the published ATLAS Run-2 expected 95% CL limits through a transparent $n$-dependent calibration, determined directly for all five values of $n$ and found consistent with unity ($1.001\pm1.4\%$), so that it acts as a consistency check rather than a sizeable correction; the selection response is validated against the published ATLAS cutflow. The projected 95% CL exclusion reach in $M_D$ for $n=2$ is 12.5, 15.2 and 22.0 TeV at the three colliders, with $5\sigma$ discovery reach of 9.8, 12.0 and 17.0 TeV, extending the ATLAS Run-2 bound of $M_D>11.2$ TeV by factors of 1.1-2.0. The calibration scheme provides a template for analogous future-collider studies; the projections hold within the stated Run-2-anchored detector and uncertainty assumptions.

The first measurement of the elliptic-flow coefficient ($v_{2}$) of $\phi$, K$^0_{\rm S}$, and $\Lambda (\overline{\Lambda})$ hadrons performed in OO collisions at $\sqrt{s_{\mathrm{NN}}}=5.36$ TeV with the ALICE detector is presented as a function of transverse momentum ($p_{\rm T}$) in central and semi-central collisions. For the $\phi$ meson, in particular, this is the first measurement of the elliptic flow in a small collision system. At low $p_{\rm T}$, a characteristic mass ordering of $v_{2}$ is observed, consistent with collective expansion of the system, in which heavier hadrons are shifted to higher $p_{\rm T}$ by a common velocity field. At intermediate $p_{\rm T}$, above 3 GeV/$c$, a baryon$-$meson splitting with a $10\sigma$ significance emerges, with the $\Lambda (\overline{\Lambda})$ baryon exhibiting larger $v_{2}$ values than mesons, while, despite its mass being close to the mass of the $\Lambda$, the $\phi$ follows the K$^{0}_{\rm S}$. The $\phi$ meson behavior disfavors a purely mass-driven interpretation of the observed patterns and supports the role of hadronization via quark recombination from a collectively flowing partonic medium. These results provide complementary evidence that key signatures of partonic collectivity observed in Pb$-$Pb collisions persist in light nucleus$-$nucleus systems.

The recent measurements of event-by-event correlations between the antideuteron and antiproton numbers in Pb$-$Pb collisions at the LHC deviate from the statistical hadronization model parameterizations that describe the yields and fluctuations of light-flavored hadrons. This discrepancy may stem from nucleosynthesis processes involved in the formation of composite states like antideuterons. In this letter, we present the first measurement of the antideuteron$-$net-$\Lambda$ number correlation obtained in Pb$-$Pb collisions with the ALICE detector. This quantity is mostly sensitive to the quantum-number correlations induced by the hadronization process, as $\Lambda$ hyperons are not constituents of antideuterons. Both the antideuteron$-$net-$\Lambda$ number correlation and the normalized second-order cumulant of the net-$\Lambda$ number are well described by the statistical hadronization model implemented in Thermal-FIST. This establishes strong evidence that the antideuteron$-$antiproton correlation is driven by the nucleosynthesis processes, while the correlation between antideuterons and non-nucleonic hadrons is primarily sensitive to the hadronization mechanism.

We present a real-time field-programmable gate array (FPGA) implementation of a convolutional neural network (CNN) for event classification on full calorimeter images that retain crystal-level shower information. To meet the resource and latency constraints of trigger electronics, the network parameters and hidden activations were quantized to 8 bits with quantization-aware training, and a lossless space-to-depth reorganization rearranged the input image into fewer spatial positions, reducing the estimated CNN inference latency by more than an order of magnitude. The approach was evaluated with the J-PARC KOTO detector, recorded data, and data-acquisition (DAQ) constraints as a benchmark, targeting the online separation of the rare $K^{0}_{L}\rightarrow\pi^{0}e^{+}e^{-}$ decay from its dominant $K^{0}_{L}\rightarrow\pi^{+}\pi^{-}\pi^{0}$ background. As a proof of principle, the firmware was implemented and validated on an AMD Versal Premium VPK120 evaluation board. On 100,000 test events, inference on the board rejected $(99.34\pm0.04)\%$ of the recorded background sample enriched in $K^{0}_{L}\rightarrow\pi^{+}\pi^{-}\pi^{0}$ decays at a nominal $K^{0}_{L}\rightarrow\pi^{0}e^{+}e^{-}$ efficiency of 95%, with a measured CNN inference latency of 17.9 $\mu$s. Combining the measured image-assembly and CNN inference latencies with the estimated link-transfer and score-return times gives a projected end-to-end latency of 25.9 $\mu$s, within the adopted 32 $\mu$s DAQ latency budget. These results demonstrate the feasibility of using full-granularity calorimeter images for CNN-based event selection that achieves high background rejection while keeping the trigger latency within tens of microseconds.

A light pseudoscalar lighter than the 125 GeV Higgs boson remains allowed within the type-I Two-Higgs Doublet Model (2HDM), consistent with current collider constraints, including Higgs signal strength measurements and direct searches for additional Higgs bosons. At the Large Hadron Collider (LHC), the light pseudoscalar $A$ can be pair-produced through the undetected decay of the Standard Model-like Higgs boson, $h_{125} \to AA$, when this channel is kinematically accessible (i.e., $m_A < m_{h_{125}}/2$). For the charged scalar $H^\pm$, once open, the bosonic decay channel $H^\pm \to W^{\pm(*)} A$ naturally dominates over the fermionic $H^\pm \to \tau^\pm\nu$ and $H^\pm \to tb$ channels, while relaxing the existing $H^\pm$ limits. As a result, we propose searching for the charged Higgs boson through $H^\pm \to W^{\pm(*)} A$ in the $4b+2j+\ell\nu$ final state. This channel is particularly sensitive to moderate-to-large $\tan\beta$ values and provides an excellent complementary probe to searches based on fermionic $H^\pm$ decays at the LHC. Our collider analysis demonstrates that this channel can probe a substantial region of the type-I 2HDM parameter space at the high-luminosity LHC.

Antineutrinos produce hyperons without kaons through Cabibbo-suppressed quasi-elastic scattering on a single nucleon. A two-body current acting on a correlated nucleon pair, nubar [NN] -> l+ Y N', gives the same hyperons plus a second nucleon. A companion paper constructs this current without fixing its size. StrangeMC, the only generator with the channel, set it to a flat 15% of the quasi-elastic hyperon cross section. We determine the ratio f = sigma_2p2h/sigma_QE of the two-body to the quasi-elastic hyperon cross section on carbon versus antineutrino energy, and the detector signatures of two-body events. The strange two-body current and its non-strange counterpart are computed on the same nucleon pair. Their ratio is converted into f with the Valencia non-strange 2p2h and quasi-elastic cross sections for carbon, assuming that nuclear effects missing from the pair are common to both. Under this assumption f rises from 0.14 at 0.6 GeV to 0.20-0.22 at 2-5 GeV. Averaged over the Booster antineutrino spectra it is 0.15-0.17, close to the 15% placeholder, which StrangeMC now replaces with f(E). The largest evaluated uncertainty comes from separating the two-body current from one-body production followed by rescattering: across the separation prescriptions explored, f at 1 GeV ranges from 0.12 to 0.22. Interference with one-body production is not calculated and could be as large as the two-body rate. In a simulated SBND antineutrino-mode exposure on argon with the carbon f, two-body events are about 11% of events that produce a strange hadron. They supply more than a quarter of the kaon-less Sigma+ events used to tag hyperon charge exchange, partly as Sigma+ made at the weak vertex. They reconstruct 12-30% low in energy under the quasi-elastic hypothesis. After final-state interactions half of them have a proton above 40 MeV at the vertex, against a fifth of quasi-elastic events.

Multiparticle correlations of event-plane angles, known as symmetry plane correlators (SPCs), and correlations of anisotropic flow amplitudes of different harmonics, known as symmetric cumulants (SCs), help constrain the initial conditions and the transport properties of the system formed in heavy-ion collisions. Such multiparticle correlations are robust against nonflow biases in heavy-ion collisions; however, in small systems the nonflow contribution is sizeable and can distort SPCs and SCs at low multiplicity, mimicking a flow-like signal. Evidence of nuclear-geometry-induced collectivity in light-ion collisions, such as OO and Ne--Ne, has been established at the LHC. Measurements of SPCs and SCs in these systems, and their comparison to heavy-ion collisions, can therefore clarify the system-size dependence of these correlations and the strength of nonlinear mode coupling. In this work, an improved nonflow subtraction framework based on multiplicity reweighting ($\czero$) and $v_1$-estimator-based ($\cone$) scaling is validated for selected SPCs and SCs using a controlled toy model and the \textsc{Pythia~8~Angantyr} model, where collective flow is absent by construction. We find that $\cone$ provides the more reliable nonflow control and apply it to predict the SPCs and normalized SCs in OO collisions at $\sqrt{s_{\rm NN}}=5.36$~TeV with the \textsc{ampt} string-melting model. The results are compared with published Pb--Pb measurements from ALICE and from \textsc{ampt} as a function of midrapidity charged-particle multiplicity $\langle N_{\rm ch}\rangle(|\eta|<0.5)$, placing light- and heavy-ion systems on a common system-size axis.

Lund-plane and Particle Transformer (ParT) jet taggers can differ in performance because one representation is weaker, or because a weaker downstream computation reads an adequate one. Standalone performance cannot separate these. Model stitching can: it links the frozen early layers of one network to the frozen late layers of another through a single trained linear map. As a positive control we use an incomplete Lund representation, the primary declustering branch, whose deficit is known independently. Stitched into ParT's downstream, it reaches a background rejection at least as much as its own downstream recovers, yet far below ParT's background rejection: the deficit is representational, not computational. A hypergraph reader over splitting triplets does no better, ruling out higher-order correlations as the cause. The gap is invisible in AUC, which stays within 1\% of its target value, and appears only in background rejection at low signal efficiency. It localizes to 0.6\% of background jets, at and above the top mass, with more constituents but fewer primary splittings than typical background. Rebuilding the representation on the full Cambridge/Aachen declustering tree repairs the deficit as predicted: standalone rejection rises, closing 72\% of the gap. Masking nodes at inference shows the repaired network still relies almost entirely on the primary branch; the secondary splittings make the representation learnable rather than supplying information it reads. Along the primary branch, the widest-angle splitting, near the top and $W$ opening angles, is worth nearly twice a random one. For quark/gluon tagging the primary branch is far less deficient, retaining 77.6\% of ParT's rejection (46.9\% for top), and the full tree closes only 33\% of the gap (72\% for top).

Density estimation with generative models is a common task in the physical sciences, for example in background estimation in particle physics. Existing machine learning techniques, however, lack methods to quantify uncertainties that can be propagated into a statistical test. That limits the validation of generative models to standard goodness-of-fit tests, in which the generative model is treated as exact (point-null hypothesis). But a model fitted to a finite training set is never exact, so if its uncertainties are neglected, any sufficiently powerful goodness-of-fit test will eventually reject it once the test set is large enough to resolve this estimation error. We propose SIGMA, an end-to-end framework combining density estimation with principled uncertainty quantification. We use Gaussian mixture models, for which uncertainties are tractable, and further construct a new goodness-of-fit test in which these uncertainties are profiled as nuisance parameters to account for the statistical error of the fit, promoting the conventional point-null test to a composite one. A rejection then signals genuine mismodeling rather than a test set large enough to resolve the error of the fit, provided the propagated uncertainties faithfully cover that error. We demonstrate the framework on a controlled two-dimensional toy model and on a collider physics benchmark modeling the QCD background process. With uncertainties included, the same model can be trusted on test sets up to an order of magnitude larger than the standard test allows. Our work thus provides a statistically grounded basis for a robust use of generative models in the physical sciences.

A reinterpretation of a prior narrow-resonance search is performed to investigate the resonant production of pairs of dijet resonances via broad mediators. This analysis targets events with four resolved jets, requiring dijet invariant masses greater than 0.2 TeV and four-jet invariant masses greater than 1.6 TeV. The search uses a data sample corresponding to an integrated luminosity of 138 fb$^{-1}$ collected by the CMS experiment in proton-proton collisions at $\sqrt{s}$ = 13 TeV. The reinterpretation considers the production of new heavy four-jet resonances, with widths ranging from 1.5 to 10% of their mass, which decay to a pair of dijet resonances. This analysis probes resonant production in the four-jet and dijet mass distributions. Upper limits at 95% confidence level and significances are reported on the production cross section of new resonances as functions of their widths and masses, between 2 and 10 TeV. In particular, at a four-jet resonance mass of 8.6 TeV, the local (global) significance ranges from 3.9 (1.6) to 3.6 (1.4) standard deviations (s.d.) as the resonance width is increased from 1.5 to 10%. This relative insensitivity to the choice of width indicates that a broad resonance is an equally valid interpretation of this excess. The broad resonance hypothesis at a resonance mass of 8.6 TeV is supported by the presence of an event with a four-jet mass of 5.8 TeV and an average dijet mass of 2.0 TeV. Also, we report the reinterpretation of a second effect, at a four-jet resonance mass of 3.6 TeV, which has a local (global) significance of up to 3.9 (2.2) s.d.

Jets are collimated clusters of particles formed by the hadronization of partons following a hard interaction. In proton-proton ($pp$) collisions at the Relativistic Heavy Ion Collider (RHIC), jet production is dominated by $gg$ and $qg$ partonic processes, allowing us to directly probe the gluon parton distribution function (PDF) in the proton in a way complementary to deep inelastic scattering. In this paper, we report the double-differential inclusive-jet cross sections as a function of jet transverse momentum, $p_{\rm T}$, and pseudorapidity, $\eta$, at center-of-mass energies $\sqrt{s} = 200$ and $510$~GeV, from $pp$ collisions studied with the STAR detector. The jet $p_{\rm T}$ is corrected for underlying event contributions by applying an off-axis cone method. At mid-pseudorapidity, $|\eta| < 0.9$, the kinematic coverage of our data extends to $0.07 < x_{\rm T} \text{ (}= 2p_{\rm T}{} / \sqrt{s} \text{)} < 0.5$ and $0.03 < x_{\rm T} < 0.31$ at $\sqrt{s} = 200$~and 510 GeV, respectively, where the gluon PDF is poorly constrained by the TeV-scale $pp$~($p\bar{p}$) colliders. The inclusive jet cross sections are compared to the next-to-next-to-leading order perturbative quantum chromodynamics calculations using several recent PDF sets as inputs. These results will further constrain the gluon PDF, help tune Monte Carlo generators, and provide critical reference data needed to study the quark-gluon plasma.

We present the first amplitude analysis and branching fraction measurement of $D^{+} \rightarrow K_{S}^{0}K_{L}^{0}\pi^{+}$ decay. The analysis uses a dataset corresponding to an integrated luminosity of 20.3~$\rm fb^{-1}$, which was recorded at a center-of-mass energy 3.773~GeV by the BESIII detector. The measured branching fraction is $\mathcal{B}(D^{+} \rightarrow K_{S}^{0}K_{L}^{0}\pi^{+})=(5.780\pm0.085\pm 0.052)\times10^{-3}$, where the first uncertainty is statistical and the second is systematic. Using the known value of ${\cal B}(D^+ \to \phi \pi^+,\,\phi \to K^+K^-)$, we determine the relative branching fraction between $\phi \to K_{S}^0K_{L}^0$ and $\phi \to K^+K^-$ to be $\mathcal{B}(D^{+} \to \phi \pi^{+}, \phi \to K_{S}^0K_{L}^0)/\mathcal{B}(D^{+} \to \phi \pi^{+}, \phi \to K^+K^-)= 0.628\pm0.022\pm 0.015\pm0.017$, where the third uncertainty is related to $\mathcal{B}(D^{+} \to \phi \pi^{+}, \phi \to K^+K^-)$. This result is significantly lower than the previous world average and is consistent with the isospin expectation for the $\phi$ meson's coupling to charged and neutral kaon pairs.

Reconstructing the mass of a heavy resonance from its decay products with missing energy is one of the central tasks that directly determine the sensitivity in new physics searches at collider experiments. Supervised learning approaches to this problem often struggle to generalize well due to the presence of various systematic uncertainties and distribution shifts. Exhausting all possible variations in the labeled data can be very compute-intensive, while a failure of the model to generalize can corrupt the reconstructed resonance widths that are critical in peak-hunting analyses. In this work, following the foundation model paradigm, we use a self-supervised approach to pre-train a transformer encoder with VICReg to learn an embedding invariant to various corruptions, then fine-tune it for mass regression on a heavy resonance with masses ranging from 2.5 to 6.5 TeV and a SUSY-like cascade decay into an eleven-body final state. We show that the pre-trained model reconstructs sharper resonance peaks and has a more stable performance under various realistic corruptions, compared to a supervised model of the same architecture trained on the same augmented data from scratch.

Many self-supervised methods for training foundation models at the Large Hadron Collider (LHC) rely on data augmentations to encourage the model to embed events into a representation space invariant to certain physical or detector symmetries. A common challenge arises from the large freedom in choosing a proper set of augmentations on which downstream performance depends. The implementation of augmentations involves either modifying existing events, potentially breaking the event fidelity, or simulating more event variants, which is computationally intensive. In this work, we present a data-driven method of pairing events by their similarity via the energy mover's distance (EMD), which measures how similar two events are in terms of the work required to transform one into the other. With this approach, distinct events are sampled and matched by their similarity to serve as views for learning invariance, keeping the physics content of each event intact without handcrafted distortions. We demonstrate this augmentation-free pairing method by pre-training on QCD jets via self-distillation and show that it can yield semantic jet embeddings with downstream discrimination power comparable to or better than an augmentation-based baseline.

Using $e^+e^-$ collision data collected by the BESIII detector at center-of-mass energies from $3.80$ to $4.95~\text{GeV}$, the inclusive production cross sections of open-charm mesons, $\sigma(e^{+}e^{-}\to D^{0}/D^{+}/D_{s}^{+}+X)$, are measured by tagging a $D^0$, $D^+$, or $D_s^+$ meson, where $X$ denotes any accompanying particles. Open-charm production dominates the total hadronic yield above threshold. Therefore these measurements provide a vital reference for understanding hadron production and for validating global $R$ value results. The results offer crucial inputs for investigating heavier charmonia and exotic states coupled to open-charm final states, while serving as a benchmark for comparison with the sum of measured exclusive channels. This comparison directly evaluates the completeness of current exclusive datasets and clarifies the underlying hadronization mechanisms in the open-charm region.

Is the usual treatment of axion dark matter as a classical field reliable? We show that the answer is subtle: the axion field could well be in a quantum state that has no complete classical description, but realistic detectors cannot tell the difference. To see this, we solve a fully quantum model of axion detection using quantum optics techniques. We show that intrinsically quantum effects are washed out by mode averaging or small amounts of noise, and significantly suppressed by the weakness of the axion coupling. Our work exemplifies that there should always be a classical analog for axion dark matter effects, extends to other wave (ultralight) dark-matter candidates, and gives a general method to compute the effects of exotic dark-matter states.

A study of two-particle differential number ($B$) and transverse momentum ($P_{2}^\mathrm{CD}$) balance functions in photon-proton ($\gamma p$) and proton-proton (pp) collisions at $\sqrt{s}=$ 5.36 TeV is presented. The analysis focuses on inclusive charged hadrons within the pseudorapidity coverage $|\eta|<2.4$ and the transverse momentum interval $0.3 < p_\mathrm{T} < 3.0$ GeV and examines their correlations in terms of relative pseudorapidity ($\Delta\eta$) and relative azimuthal angle ($\Delta\phi$). The correlation functions are evaluated for same- and opposite-sign pairs, and their combinations are used to extract charge-dependent (CD) and charge-independent (CI) components. The evolution of the near-side peak of the CD correlations is investigated in terms of $\Delta\eta$ and $\Delta\phi$ as a function of charged-particle multiplicity ($N_\mathrm{ch}$) for $\gamma p$ collisions and compared to pp collisions at a similar multiplicity range. A clear multiplicity dependence of the balance function width is obtained. The width is found systematically lower in $\gamma p$ events than in pp collisions. This study provides valuable information on particle correlations and production mechanisms in low-$N_\mathrm{ch}$ regimes for upcoming measurements in small systems.

The nonlinear inductance of the Josephson junction has enabled the development of a wide range of continuous-variable amplifiers and qubit-based devices with unprecedented sensitivity. We present an alternative use of the Josephson junction in the context of broadband impedance matching. The idea poses a potential solution to a longstanding problem in the field of high energy particle physics. The axion, a compelling candidate for the dark matter, converts to a weak electromagnetic signal at an as-yet unknown frequency. As such, the ideal axion detector does not compromise bandwidth for sensitivity, a trade-off intrinsic to all linear, time-invariant and passive circuits. We propose a circuit that uses a Josephson junction in an impedance matching network to overcome these gain-bandwidth constraints and increase the scan rate of axion searches. The Josephson junction can be biased to exhibit negative inductance capable of canceling geometric inductance similar to a capacitor but across a wider frequency range.

Optical time projection chambers (OTPCs) are well suited for applications that require the highest spatial resolution for particle track reconstruction. The MIGDAL experiment uses a glass GEM-based OTPC and observes a systematic excess in both the intensity and width of particle tracks in its optical readout, when compared with charge readout simulations. One hypothesis is that scintillation light produced inside a GEM hole during the avalanche propagates through the GEM substrate and exits neighboring holes. We present lab measurements testing this hypothesized optical broadening effect in three types of GEM substrates: glass, ceramic, and FR4. Our observations quantify this optical broadening and demonstrate it to be strongest in glass GEMs. Additionally, we use Geant4 simulations to both reproduce our observations and quantify optical broadening effects in realistic charge avalanches. Applying our glass GEM effects to simulated particle tracks yields increases of track intensity and widths by up to approximately 26% and 31%, respectively. This may explain the larger-than-expected track intensities and widths observed in the MIGDAL OTPC. This effect may be relevant to GEM-based OTPCs, with its magnitude depending on the GEM structure and detector-specific optical configuration.

The photoproduction of $J/\Psi$ in peripheral Oxygen - Oxygen ($OO$) collisions at the Large Hadron Collider (LHC) is investigated considering distinct assumptions for the modeling of the nuclear photon flux, overlap function and dipole - proton scattering amplitude. Predictions for the associated rapidity distributions and total cross - sections are presented. Our results indicate that the experimental study of the photoproduction of $J/\Psi$ in peripheral $OO$ collisions is, in principle, feasible. In addition, they point out that the combination of the results for this final state in $OO$ collisions with those obtained for $PbPb$ collisions will allow us to derive important constraints on the description of photon - induced process at peripheral collisions.

We study the possibility of probing the dynamical inverse seesaw mechanism for the origin of light neutrino masses via the detection of stochastic gravitational waves (GW) in the low-frequency regime currently being probed by pulsar timing arrays. As the lepton number-violating term in inverse seesaw typically remains in the sub-MeV ballpark, its dynamical origin naturally brings the possibility of a low-scale first-order phase transition, which can be probed at low-frequency GW experiments. We also find interesting complementarity with heavy neutral lepton searches, as GW experiments remain sensitive to parameter space with small active-sterile mixing, which is out of reach for most particle physics experiments.

We investigate the entanglement in the momenta of particle pairs produced at colliders, discarding the information on their spin state. We show how to quantify this entanglement by reducing the continuous momentum variables to a two-qubit system through hemispherical projections. We discuss in detail the production of $\tau$-lepton pairs at an electron collider. In the Standard Model, the discretized momenta are in a genuinely maximally entangled pure state for suitable choices of the hemispheres.

Hyperon single-particle potentials U_Y(rho) govern hyperon propagation in nuclei and the appearance of hyperons in dense matter, where they soften the neutron-star equation of state (EOS) and lower the maximum mass M_max, the hyperon puzzle. Charged-current (anti)neutrino interactions on argon-40 create Lambda and Sigma hyperons inside the nucleus. The reconstructed Lambda yield and momentum and a kaon-vetoed final-state-interaction (FSI) Sigma+ tag respond to U_Lambda and U_Sigma at SBND and DUNE energies. We write the potential at low density as U_Y = U_Y(rho_0) (rho/rho_0)^gamma, plus a correction that vanishes at saturation. A detector-level Fisher forecast, with production cross sections, fluxes and efficiencies as nuisance parameters and the muon kinematics calibrating the production model, determines U_Lambda best at rho* = 0.82 rho_0, independently of gamma. The statistical precision is 1.5 MeV; with the production model profiled and the remaining systematic uncertainties added in quadrature, the total is +-31 MeV, of which the largest item, 18 MeV, is the convention by which the cascade applies the potential. The measurement is a check, with different systematics, of the hypernuclear constraint, +-3.3 MeV at rho*. U_Sigma is not constrained. In relativistic mean-field EOSs anchored to hypernuclear and Sigma-atom depths, M_max for n p Lambda Sigma- matter is 1.94 M_sun for GM1 and 2.03 M_sun for FSU2H with the phi meson; it depends on a high-density Lambda NN term c_Lambda that the neutrino data do not constrain. The calculations use the StrangeMC Monte Carlo generator.

Since its first observation by the ATOMKI experiment in 2018, the $\rm ^*Be$ anomaly has attracted considerable interest within the dark sector community as it may indicate the existence of a new fundamental particle with a mass of about 16.9 MeV, the $X_{17}$. However, the minimal model describing $X_{17}$ as a new vector boson is severely constrained by null results from legacy beam-dump experiments. Among these, the E141 experiment at SLAC places stringent limits on the $X_{17}$ coupling to electrons $\varepsilon_e$ in the $5.1\times10^{-5}\lesssim \varepsilon_e\lesssim1.7\times10^{-4}$ region. This excludes the possibility of a long-lived $X_{17}$, potentially in contrast with the preliminary estimate of the particle lifetime recently reported by the ATOMKI collaboration. The E141 limits commonly adopted in the literature rely on reinterpretations of the original analysis under solid but simplifying assumptions. While these studies provide a reliable estimate of the experiment's reach, they rely on approximations that were well justified when the boson mass was largely unconstrained. With the $X_{17}$ mass now confined to a narrow region by recent experimental observations, a more refined treatment of the E141 sensitivity becomes necessary. In this work, we revisit the E141 exclusion limits in the $X_{17}$ scenario by performing a dedicated reanalysis that incorporates a more accurate treatment of the experimental setup and signal prediction. We quantify the impact of these refinements on the excluded parameter space and discuss their implications for the compatibility between the E141 constraints and the vector boson interpretation of the ATOMKI anomalies.

We report the first temperature-dependent characterization of a NaI(Tl) crystal readout by two silicon photomultipliers (SiPMs) directly coupled to opposite ends of the crystal for rare-event searches. A $6 \text{ mm} \times 6 \text{ mm} \times 13 \text{ mm}$ NaI(Tl) crystal was directly coupled to two SiPMs and characterized in a liquid nitrogen-cooled cryostat over a temperature range of 94$-$293 K. The light yield, energy resolution, and scintillation decay time were measured using $\gamma$-ray peak from a $^{241}$Am source. After correcting for optical crosstalk contributions, the light yield increased, reaching $17.7 \pm 1.1$ photoelectrons/keV at 238 K, corresponding to a 34.5% enhancement relative to room temperature (293 K). Furthermore, dual-channel configuration effectively suppresses random thermal noise via coincidence triggers, which together with the observed increase in light yield, provides a critical pathway toward lowering the energy threshold for dark matter and coherent elastic neutrino$-$nucleus scattering searches.