-
Francesco CordellaAI/ML for ePIC and Beyond
Hybrid pixel detectors such as Timepix3 record position, energy (ToT) and time of arrival (ToA) for every hit, yet particle identification (PID) still compresses each cluster into a few engineered scalars before any classifier sees it. We treat each cluster as an unordered point cloud in (x, y, E, t) and extend PointNet++ to four dimensions, with a learnable diagonal metric and three...
Go to contribution page -
Marco Zaccheddu (Argonne National Lab)AI/ML for Data Analysis and Theory
The extraction of Transverse Momentum Dependent distributions (TMDs) from experimental observables represents a classic ill-posed inverse problem in hadronic physics. Traditionally, this challenge is addressed by assuming rigid analytical functional forms, which introduce parameterization bias, obscure unconstrained degrees of freedom, and limit the exploration of the full epistemic...
Go to contribution page -
Amit Bashyal (Brookhaven National Laboratory)Trends in Data Science
The design of next-generation scientific instruments, such as detectors for the Electron-Ion Collider (EIC), presents a challenging optimization problem involving high-dimensional parameter spaces, competing objectives (e.g., cost and performance), and computationally expensive simulation and reconstruction workflows. Efficiently exploring these complex design spaces motivates approaches that...
Go to contribution page -
Aoran Liu (staff@duke.edu;student@duke.edu;member@duke.edu)AI/ML for Data Analysis and Theory
Several nuclear physics studies using the CLAS12 detector rely on the accurate reconstruction of neutrons and photons from its electromagnetic calorimeter system. These studies are often limited by an overabundance of false neutral clusters produced by the existing reconstruction software, COATJAVA, which reduces the reliability of neutral particle measurements and restricts the physics reach...
Go to contribution page -
Dr Wouter Deconinck (University of Manitoba)AI/ML for Data Analysis and Theory
Extracting physics at the Electron-Ion Collider (EIC) means connecting measured observables to theory through chains of definitions, approximations, and fitting or unfolding procedures that are rarely stated precisely enough to be checked. We present EpsilonEridani, an open project that formalizes physics in Lean4 on top of Mathlib, Physlib and TauCeti, with the constraint that all code is...
Go to contribution page -
Raiqa Rasool (Jefferson Lab)Calibration, Monitoring, and Experimental Control in Streaming Environments
Data quality monitoring (DQM) is essential for ensuring reliable detector performance. Hydra is an AI-enabled DQM framework developed at Jefferson Lab that leverages complementary machine learning techniques to automate the evaluation of quality assurance plots. The framework incorporates computer vision models for image classification, autoencoders for anomaly detection, clustering algorithms...
Go to contribution page -
Tasnuva Chowdhury (Brookhaven National Laboratory)AI/ML in Production and Distributed ML
PanDA (Production and Distributed Analysis) is a large-scale workload management system supporting distributed scientific computing for ATLAS and expanding to ePIC. As computing infrastructures and scientific workflows grow in complexity, integrating AI into workload management offers opportunities to improve operational efficiency, resource utilization, and reliability. This talk presents...
Go to contribution page -
Norbert Novitzky (ORNL)Calibration, Monitoring, and Experimental Control in Streaming Environments
The Electron-Ion Collider (EIC) presents exciting opportunities to integrate artificial intelligence (AI) and machine learning (ML) into calorimeter readout, online event reconstruction, and particle identification. Based on the Longitudinally segmented Forward Hadronic CALorimeter (LFHCAL) of ePIC we are exploring an AI-assisted readout and reconstruction framework designed to optimize the...
Go to contribution page -
Zhiwan Xu (Argonne National Lab)AI/ML for ePIC and Beyond
Next-generation nuclear physics experiments at the Electron-Ion Collider (EIC) require advanced reconstruction and analysis techniques capable of exploiting the increasingly detailed data provided by the ePIC detector systems. In this work, we present the design of an AI-ready data framework based on low-level pre-reconstruction detector readout for general detectors, and discuss its...
Go to contribution page -
Rajat Sainju (member@anl.gov)AI/ML in Production and Distributed ML
Large scientific facilities accumulate decades of operational knowledge across logbooks, maintenance records, documents, chat, and control-system data, yet no single interface searches them together. Here, we present APS-RAG, a retrieval-augmented generation (RAG) platform deployed at the Advanced Photon Source (APS), and what it takes to run it in production. A daily diff-and-upsert pipeline...
Go to contribution page -
Dmitry Romanov (Jefferson lab)Trends in Data Science
Autonomous research loops, in which an AI agent proposes a hypothesis, changes code, measures the result, and decides what to try next, have become one of the defining trends of 2026, from Karpathy's autoresearch to AlphaEvolve-style evolution engines. Industry reports are striking, yet they rarely address the central question of this talk: when can such a loop be effective, and when can it be...
Go to contribution page -
Hemalata NayakAI/ML for Data Analysis and Theory
Precise reconstruction of jet kinematics is essential for the physics program of the future Electron-Ion Collider (EIC). Deep-learning-based methods have recently shown strong performance in improving jet energy scale and resolution; however, standard deterministic approaches do not provide the uncertainty quantification required for robust precision measurements. In this work, we introduce a...
Go to contribution page -
Pengyu Li (old dominion university)AI/ML for Accelerators
We address the inverse problem of estimating solenoid field errors and misalignments from downstream viewer images to calibrate a digital twin of the Upgraded Injector Test Facility (UITF) at Jefferson Lab. This study uses the simulation code Cheetah to model a two-solenoid section of the beamline, with the field-strength error and transverse offsets of each solenoid device as unknown...
Go to contribution page -
Hanpu JiangCalibration, Monitoring, and Experimental Control in Streaming Environments
The sPHENIX hadronic calorimeter (HCal), comprising inner and outer sampling calorimeters, provides essential measurements of hadronic energy for jet studies in proton–proton and heavy-ion collisions. Part of the outer HCal is planned for reuse in the ePIC barrel hadronic calorimeter, extending its scientific role to the Electron–Ion Collider. We present the calibration of the sPHENIX inner...
Go to contribution page -
Cameron Clarke (Thomas Jefferson National Accelerator Facility)Calibration, Monitoring, and Experimental Control in Streaming Environments
End-station detectors watch the beam continuously and are a natural candidate for beam optics monitoring, but they only partially capture the beam tune with nontrivial underlying correlations. A change in a detector rate, helicity correlate asymmetry, or spatial profile carries information about the upstream beam state, but inferring that state and how to adaptively steer the beam to restore...
Go to contribution page -
Chun Yuen Tsang (Argonne National Laboratory)AI/ML for ePIC and Beyond
The application of AI, which ranges from task-specific models to large foundation models that are adaptable to different tasks and detectors, has been transformative for nuclear physics. However, realistic detector simulations and/or labeled data are needed for AI to be usable in experiments. Furthermore, given the significant resources required to train a massive foundation model, a baseline...
Go to contribution page -
Dr Ming Liu (Los Alamos)Calibration, Monitoring, and Experimental Control in Streaming Environments
Real-time AI/ML is emerging as a critical capability for future high-rate nuclear physics experiments, such as EIC, where rare physics signals must be identified within complex detector data streams. This talk will summarize key lessons learned from the DOE-funded project “Intelligent Experiments through Real-Time AI: Fast Data Processing and Autonomous Detector Control for sPHENIX and Future...
Go to contribution page -
Dr Jacky Kumar (member@umontreal.ca;staff@umontreal.ca;employee@umontreal.ca)AI/ML for Data Analysis and Theory
Finding new physics (NP) is one of the central challenges in particle physics. The Standard Model Effective Field Theory (SMEFT) provides a model-independent framework to interpret any discrepancies with the Standard Model (SM). However, identifying the relevant SMEFT operators is challenging, because of the enormous operator space and the complex correlations among operator. Consequently,...
Go to contribution page -
Chunyu Hu (Old Dominion University)AI/ML for Data Analysis and Theory
Symbolic regression (SR) yields closed-form models of generalized parton distributions (GPDs), yet a good fit alone does not ensure consistency with known GPD physics. We utilize SR to model the $x$ and $t$ dependence of the flavor isovector combination $H_{u-d}(x,t,\xi)$ at $\xi=0$ under physics constraints. These models are fitted to GPD results from Lattice QCD at the physical pion mass....
Go to contribution page -
Poonam Choudhary (Jefferson Lab)AI/ML for Data Analysis and Theory
Extracting transverse-momentum-dependent parton distributions (TMDs) from experiment is an inverse problem. We present a generative adversarial network (GAN) framework that solves this inverse problem at the event level, without assuming a functional form for the TMD. The GAN generator produces the TMD as an image and is trained against reference events through a discriminator, using a fully...
Go to contribution page -
Derong Xu (Brookhaven National Laboratory)AI/ML for Accelerators
Beam-beam interactions are a major factor limiting luminosity in the Electron-Ion Collider (EIC), motivating reliable and efficient simulations. AI coding agents offer new opportunities for developing high-performance tools to simulate these complex, nonlinear interactions, but their effective use requires explicit physical assumptions, consistent numerical conventions, and **systematic...
Go to contribution page -
Jan-Frederik Schulte (Purdue University)AI/ML in Production and Distributed ML
The tight integration of machine learning (ML) models into detector readout and trigger systems will allow future detectors to move complex reconstruction tasks much closer to the detector. This will enable these detectors to cope with much higher data rates by performing more complex and better targeted event selection or data compression in DAQ and trigger systems. ML algorithms have to be...
Go to contribution page -
Hao-Ren Jheng (Massachusetts Institute of Technology)Calibration, Monitoring, and Experimental Control in Streaming Environments
A neural-network-based method is developed in this work to derive residual per-track momentum-scale corrections for the sPHENIX charged-particle reconstruction. The method addresses both charge-even and charge-odd momentum biases using kinematic constraints from reconstructed $K_{s}^{0}$ and $\Lambda^{0}$ decays. The correction is derived from reconstructed data and applied after track...
Go to contribution page -
Dhanush Hangal (Lawrence Livermore Nat. Laboratory (US))Calibration, Monitoring, and Experimental Control in Streaming Environments
High-rate nuclear physics experiments require fast and accurate detector calibration to support continuous data collection and reconstruction. Time Projection Chambers (TPCs) are particularly sensitive to time-dependent space-charge distortions, which require frequent calibration updates and impose significant computational challenges.We present a machine-learning-based framework developed for...
Go to contribution page -
Cole GrangerTrends in Data Science
Foundation models are emerging as a promising paradigm for scientific computing, offering the potential to replace collections of task-specific models with shared representations that can be adapted across multiple applications. In nuclear and particle physics, this shift is motivating new approaches to detector modeling, ranging from self-supervised representation learning to generative...
Go to contribution page -
Tapasi Ghosh (Ramaiah University of Applied Sciences)AI/ML for ePIC and Beyond
The Zero-Degree Calorimeter(ZDC) is an important sub-detector component in the ePIC detector of the EIC. The detector is situated very close to the beam interaction point and consists of ECAL and HCAL to capture the neutral particle (neutron and photons) kinematics at extremely forward angles covering polar angles θ<5.5 mrad(η>6). Subsequently, this enables to study the three-dimensional...
Go to contribution page -
Pedro Vicente Leitao (Massachusetts Institute of Technology)AI/ML for ePIC and Beyond
Precision tracking at the Electron-Ion Collider (EIC) requires highly granular silicon detectors with stringent constraints on power consumption and material budget. Transmitting raw detector data increases demands on readout bandwidth, power delivery, and cooling. On-chip AI offers a route to reducing data volume before transmission, aiming to preserve the information needed for tracking...
Go to contribution page -
Cristian Rossi (INFN Sezione di Roma)AI/ML in Production and Distributed ML
The Electron-Ion Collider (EIC) under construction at Brookhaven National Laboratory is a next-generation facility designed to systematically investigate the gluonic and spin structure of nucleons and atomic nuclei. To fulfill the stringent physics requirements of the EIC ePIC detector, the dual-radiator Ring Imaging Cherenkov (dRICH) detector provides particle identification (PID) over a wide...
Go to contribution page -
Tyler Hellstern (Duke University)AI/ML for Data Analysis and Theory
Detector simulations serve a critical role in Nuclear and Particle Physics for design and analysis, but these expensive simulation programs also act as computational bottlenecks. This work leverages the PARNASSUS framework to provide an end-to-end generative AI based fast simulation for the CLAS12 detector and reconstruction pipeline at Jefferson Lab. By using Continuous Normalizing Flows with...
Go to contribution page -
Daniel Almonte (Stony Brook University)AI/ML for ePIC and Beyond
In previous work, a deep neural network was shown to improve the reconstruction of inclusive kinematic variables in neutral-current deep inelastic scattering. Building on this approach, we present a particle flow network (PFN) for reconstructing semi-inclusive kinematic variables. The PFN takes as input the kinematics of the scattered electron, the hadronic final state, and the inclusive...
Go to contribution page -
Gian Michele InnocentiAI/ML for ePIC and Beyond
We present performance results for charged-particle tracking on H1 and ePIC simulated collision data as well as archived LEP data using the FM4NPP [1,2] foundation model. This study is part of a DOE Genesis project on multimodal and multifacility applications of FM4NPP, extending the model beyond its original sPHENIX time projection chamber setting to silicon trackers and electron colliders....
Go to contribution page -
Patrick Moran (W&M)Calibration, Monitoring, and Experimental Control in Streaming Environments
As part of the Coupling Experiment to Accelerator Controls (CEAC) project, an AI systems are being developed that adjust CEBAF beam and detector settings in response to experimental conditions measured in real time, with complementary use cases in MOLLER (HALL A) and GlueX (Hall D). For MOLLER, the goal is to minimize backgrounds in the Hall A Compton polarimeter, which must run continuously...
Go to contribution page -
Sandeep Shiraskar (The Catholic University of America)Calibration, Monitoring, and Experimental Control in Streaming Environments
The Electron–Ion Collider (EIC) will generate continuous, high-rate detector streams, creating a need for low-latency identification of rare physics signals amid substantial background and pileup. We propose a streaming-native, physics-informed AI framework that brings inference directly into the detector data-processing pipeline. Detector readout is organized into time-ordered frames that...
Go to contribution page -
Daniel MarxAI/ML for Accelerators
AI capabilities are advancing rapidly, potentially transforming how particle accelerators are designed, commissioned, and operated. For the Electron-Ion Collider (EIC), whose construction and commissioning will precede decades of anticipated operation, this raises an important question: how should decisions made today account for AI capabilities that are difficult to predict?
Recent...
Go to contribution page -
Eric Moffat (Temple University/Argonne National Lab)AI/ML for Data Analysis and Theory
I will present recent results of a Bayesian Monte Carlo extraction of generalized parton distributions (GPDs) from global proton and neutron deeply virtual Compton scattering (DVCS) data as well as the resulting impact parameter distributions.
Go to contribution page -
Armen Kasparian (Jefferson Lab)AI/ML in Production and Distributed ML
Deploying reinforcement learning (RL) for autonomous control on a high power proton accelerators is gated less by the choice of RL algorithm than by a stack of infrastructure and data realities: the fidelity of any available simulator, the effective sample rate of the live machine, the latency imposed by safe machine-interface layers, and the readiness of historical operational data. We report...
Go to contribution page -
Wen Guan (BNL)AI/ML in Production and Distributed ML
Training and tuning ML models for the EIC needs computation spread across grid, HPC and cloud sites, and running that work takes a lot of operator effort. This talk shows how PanDA and iDDS run and increasingly automate distributed ML workflows. iDDS describes training and hyperparameter optimization (HPO) campaigns as workflows of dependent steps, and PanDA runs those steps on GPU resources...
Go to contribution page -
Chi Kin Tam (cua)AI/ML for ePIC and Beyond
Future experiments at Jefferson Lab (JLab) and the Electron-Ion Collider (EIC) are designed to leverage the highest achievable luminosities to probe hadronic structure with unprecedented precision. These measurements require the efficient identification of rare physics events in the presence of substantial backgrounds. To address these challenges, scalable SI/ML algorithms capable of efficient...
Go to contribution page -
Dr Markus Diefenthaler (Jefferson Lab)Calibration, Monitoring, and Experimental Control in Streaming Environments
The EIC offers an opportunity to rethink scientific workflows by combining streaming readout, artificial intelligence, and distributed computing into closed-loop workflows that connect data taking, processing, physics analysis, and operational feedback. We present SPINDLE (Streaming Physics INtelligence and Data Lifecycle Engine), an effort to develop AI-enabled workflows that integrate...
Go to contribution page -
Lara Suzanne Blokland (University of Tennessee)Calibration, Monitoring, and Experimental Control in Streaming Environments
Nuclear targets can be polarized through dynamic nuclear polarization using high magnetic fields, low temperatures, and RF. The polarization of the material can be maximized through adjusting the microwave frequency. Human operators typically make the adjustments, but machine learning techniques may be able to automate this system, as well as optimize the polarization. The targets can also be...
Go to contribution page -
Brandon Kriesten (Argonne National Laboratory)AI/ML for Data Analysis and Theory
Deep-inelastic scattering (DIS) provides a fundamental window into the quark and gluon structure of hadrons and a foundation for precision studies at current and future colliders. I will discuss how advances in AI/ML are reshaping hadron-structure global analysis, from generative approaches to reconstructing parton densities from lattice QCD calculations to MORPH, a multimodal foundation-model...
Go to contribution page -
Todd Satogata (JLAB)AI/ML for Accelerators
The DOE Nuclear Physics AI-Ready Accelerator Data (NARAD) project is building a facility-agnostic way to describe accelerator facilities so that machine-learning tools, agentic assistants, and digital twins can find and use their data. NARAD extends the Particle Accelerator Language Standard (PALS): a flattened PALS lattice supplies the physics model, and NARAD adds facility configuration data...
Go to contribution page -
6. Theoretical challenges with Semi-inclusive deep inelastic production at the electron ion colliderShaswat Tiwari (Brookhaven National Laboratory)AI/ML for Data Analysis and Theory
We study the Semi-Inclusive Deep Inelastic Scattering (SIDIS) cross-section at large values of hadron transverse momentum $P_t \gtrsim Q$, with $Q^2$ being the photon virtuality. This kinematic regime of SIDIS allows for a collinear factorization in terms of parton distribution functions (PDFs) and collinear fragmentation functions (FFs). On the other hand, at high energies, i.e., at small...
Go to contribution page -
Gabor Galgoczi (BNL)AI/ML for ePIC and Beyond
FM4NPP is a foundation model that aims to perform track finding and particle identification from sPHENIX TPC hits with one pretrained representation, adapted to each task by a head trained on a small fraction of labeled data. Its Mamba-2 backbone is pretrained to predict nearby hits at larger radii, then frozen while lightweight supervised heads are trained for downstream tasks. We outline an...
Go to contribution page -
Kishansingh Rajput (Jefferson Lab)AI/ML for Accelerators
Digital twins and learned controllers are becoming central to accelerator operations, but a twin is only useful while it matches the machine, and few ML tools have reached sustained use. Beam settings change from run to run, and hardware ages in ways the diagnostics do not capture. These changes shift the relationship between machine inputs and beam response, so models trained on a fixed...
Go to contribution page -
Chenliang Jin (Stony Brook University)AI/ML for Data Analysis and Theory
The Transferable Particle Tracking in Nuclear Physics (TPTNP) project,
Go to contribution page
supported by the DOE Genesis Mission, aims to develop a foundation model
for tracking across different detector technologies and geometries.
Building on FM4NPP, we investigate how learned representations of raw
detector measurements can support track reconstruction across experimental
settings. We will present ongoing... -
Dr Nihar Sahoo (IISER Tirupati)AI/ML for Data Analysis and Theory
High-granularity electromagnetic calorimeters play a crucial role in photon identification in high-energy particle and nuclear physics experiments. A major challenge is distinguishing isolated photons ($\gamma$) from highly boosted neutral pions ($\pi^0$), whose decay photons can produce overlapping electromagnetic showers that mimic a single-photon signal. The fine spatial segmentation of...
Go to contribution page -
Cynthia Rudin (Duke University)Trends in Data Science
Dimension reduction (DR) for data visualization provides unique insights into the structure of high-dimensional data. DR provides a bird's eye view of a dataset, showing clusters and their relationships, manifolds, branching patterns, and even possible errors in the data. It is extremely effective for scientific discovery and hypothesis generation.
One tension that has always faced DR...
Go to contribution page -
Daniel Piasecki (North Carolina State University)AI/ML for Data Analysis and Theory
The Wigner function in QCD has traditionally been interpreted as a quasiprobability density, but recent work in Koopman–von Neumann (KvN) mechanics suggests it is more naturally understood as a probability amplitude. KvN is a classical statistical theory formulated within the Dirac–von Neumann axioms, complete with wavefunctions, a complex Hilbert space, the Born Rule, Hermitian operators, the...
Go to contribution page
Choose timezone
Your profile timezone: