Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Accelerator Applications”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 433 records · Page 24
All sourcesNASA NTRSDOE OSTI
DOE OSTI2021

Towards Automatic and Agile AI/ML Accelerator Design with End-to-End Synthesis

Domain-specific designs offer greater energy efficiency and performance gain than general-purpose processors. For this reason, modern system-on-chips have a significant portion of their silicon area with custom accelerators. However, designing hardware by hand is laborious and time-consuming, given the large design space and the performance, power, and area constraints that are not realized in the software. Moreover, domain-specific algorithms (e.g., machine learning models) are evolving quickly, challenging the accelerator design further. To address these issues, this paper presents SODA Synthesizer, an automated open-source high-level ML framework to Verilog modular compiler targeting AI/ML Application-Specific Integrated Circuits (ASICs) accelerators. SODA tightly couples the Multi- Level Intermediate Representation (MLIR) compiler infrastructure [24] and open-source HLS approaches. Thus, SODA can support various ML frameworks and algorithms and can perform optimizations that combine specialized architecture templates and conventional HLS to generate the hardware modules. In addition, SODA’s closed-loop design space exploration (DSE) engine allows developers to perform end-to-end design space explorations on different metrics and technology nodes.

Zhang, Jeff↗
DOE OSTI2025

Two-level overlapping additive Schwarz preconditioner for training scientific machine learning applications

In this work we introduce a novel two-level overlapping additive Schwarz preconditioner for accelerating the training of scientific machine learning applications. The design of the proposed preconditioner is motivated by the nonlinear two-level overlapping additive Schwarz preconditioner. The neural network parameters are decomposed into groups (subdomains) with overlapping regions. In addition, the network’s feed-forward structure is indirectly imposed through a novel subdomain-wise synchronization strategy and a coarse-level training step. Through a series of numerical experiments, which consider physicsinformed neural networks and operator learning approaches, we demonstrate that the proposed two-level preconditioner significantly speeds up the convergence of the standard (LBFGS) optimizer while also yielding more accurate machine learning models. Moreover, the devised preconditioner is designed to take advantage of model-parallel computations, which can further reduce the training time.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Development of a second COMB dipole magnet with STAR wires

Rare-Earth Barium Copper Oxide (REBCO) coated conductors are an attractive option for application in high-field accelerator magnets due to their high critical field and the convenience of fabrication without heat treatment compared to some other superconductors. Fermilab plays a vital role in superconducting accelerator magnet R&D under the framework of the U.S. Magnet Development Program. An integral part of that program is the accelerator magnet development based on REBCO conductors to demonstrate self-fields of 5 T or greater compatible with operation in hybrid configurations to generate fields beyond 16 T for future High Energy Physics (HEP) applications. A small, two-layer REBCO dipole magnet based on the Conductor on Molded Barrel (COMB) magnet technology, developed at Fermilab with Symmetric Tape Round (STAR®) wires from AMPeers LLC was recently fabricated and tested in liquid helium, demonstrating the design feasibility. A second dipole magnet based on the same technology is currently under development at Fermilab. The coil has a 60-mm clear bore and a 120-mm OD and consists of six layers of STAR® conductor, which allows to significantly increase the magnetic field compared to the previous design, and potentially probe the 5 T self-field target. This presentation reports the progress in the magnet development.

Kashikhin, Vadim [Fermilab]↗
DOE OSTI2025

Using machine learning to jointly harness the strength of microscopic, fundamental-science driven and macroscopic, application-driven experiments

The PARADIGM project aims at accelerating progress in science by quantitatively answering the following question: What is the optimal combination of fundamental-science and application driven experiments to maximally reduce pertinent data uncertainties? Hence, we are bridging between microscopic experiments and data, and macroscopic simulations and experiments. Answering this question entails solving a high-dimensional and complex optimization problem which we solve with machine learning techniques.

LANSCE↗
DOE OSTI2021

A Grand Challenge "Uncertainty Project" to Accelerate Advances in Earth System Predictability: AI-Enabled Concepts and Applications

This proposal is emerging from GISS ModelE3 ESM development in the area of cloud physics, so we begin with an example of research needs/gaps from that work. Here, some of our greatest development concerns arise where we lack fundamental process-level understanding, as in ice formation. Namely, it is currently unclear what is the main process that is forming the majority of ice crystals in commonly occurring convection, apparently via secondary ice production at warm temperatures. We are keenly awaiting laboratory data for candidate mechanisms, which is not yet in hand to crucially establish their efficiency. Our progress is also hampered by a lack of uncertainty characterization in currently available measurements of ice crystal number size distributions. Furthermore, the same multiplication process may be responsible for a majority of ice crystals in many extratropical mixed-phase clouds, whose variable representation in CMIP6 ESMs may be a leading cause of differences in cloud phase feedback and ECS. Yet we have been required to deliver an ESM with the cloud physics knowledge at hand. The proposed grand challenge project is AI-enabled via application of machine learning (ML) to climate model and observational data streams (focal area 3), and applications include AI-guided observing system design and model/component/parameterization selection (areas 1 and 2). The project is structurally agnostic as to whether model or observing system components use AI approaches or not, but uncertainties must be estimated and propagatable in both.

58 GEOSCIENCES↗
DOE OSTI2024

Synthesis of motif and symmetry for accelerated learning, discovery, and design of electronic structures for energy conversion applications (Final Technical Report)

The overall goal of the projects is to develop a framework to incorporate structure motifs and crystal/orbital symmetries into the data-driven materials discovery infrastructure. The PI proposed to develop structure-motif- and symmetry-based graph convolutional networks for effective learning and efficient predictions of electronic structures and related properties. Fundamental understanding of the roles of structure motif and symmetry will establish new hypothesis and design rules, which will be combined with high-throughput computations based on density functional theory to discover novel light absorbers, transparent conductors, as well as 2D light emitting materials and heterojunctions for optoelectronics.

36 MATERIALS SCIENCE↗
DOE OSTI2024

12th World Conference on Neutron Radiography Program

This program is for the 12th Annual World Neutron Conference hosted by INL, held in Idaho Falls. Included is the conference schedule, abstracts from presenters, outings, a tour of EBR-1, and facilities at MFC including: TREAT, HFEF, and IMCL. INL researcher abstracts are found on the following pgs: 26, 42, 52, 103, 119, 169, 170, 173, and 182. The 12th World Conference on Neutron Radiography (WCNR-12) is an international forum that brings together researchers, engineers, industry practitioners and university students to promote, discuss and disseminate a wide range of topics in neutron imaging. This weeklong conference focuses on the latest methods, instrumentation and facilities, improvements in data processing and interpretation, and new applications of neutron imaging. Ever-improving neutron imaging instruments and new methods push the frontiers of science and address industrial needs. Many user facilities at large research centers with powerful neutron imaging beamlines address the research needs of an increasingly diverse range of applications. Recent developments in accelerator-based sources could expand the user base for neutron imaging by making compact neutron sources available at facilities beyond large neutron sources at major research centers. Industrial practitioners continue to use neutron imaging for practical industrial applications following industry standards. The World Conference on Neutron Radiography series is organized by the International Society for Neutron Radiography (ISNR) to promote the field of neutron imaging and foster interaction and communication between experts and users in the field (see isnr.de). WCNR-12 continues a series of meetings initiated and started in San Diego, California, in December 1981. In intervals of about four years, the conference moved between the United States, Japan and Europe until the most recent one (WCNR-11) in Sydney, Australia, in 2018 (see previous ISNR proceedings here: https://www.isnr.de/index.php/conferences.)

99 - GENERAL AND MISCELLANEOUS↗
DOE OSTI2026

Consideration of HTS rapid-cycling magnet for staged muon acceleration

The HTS conductor hysteresis dominates magnet cable power loss but is independent of the magnetic field ramping rate. This makes the HTS conductor suitable to power the rapid-cycling accelerator magnet. We present a possible application of the HTS rapid-cycling magnet as outlined in [1,2] for the staged muon acceleration including the front-end Recirculating Linear Accelerator and the followed-up Rapid Cycling Synchrotrons delivering the muon beams to the Muon Collider.[1] H. Piekarz, S. Otten, A. Kario, H. ten Kate, “Rapid-cycling HTS magnet for muon acceleration”, US MC Inaugural Meeting, FERMILAB-POSTER-24-0219-AD, August 7-9, 2024[2] H. Piekarz, B. Claypool, S. Hays, M. Kufer, V. Shiltsev, “Record High Ramping Rates in HTS Based Supercond. Accelerator Magnet”, MT 27, IEEE Trans. on Applied Superccond, 32 (2022) 6, 4100404

Piekarz, Henryk [Fermilab]↗
DOE OSTI2024

Celeritas R&D Report: Accelerating Geant4

Celeritas is a new Monte Carlo (MC) detector simulation code designed for computationally intensive applications on high-performance heterogeneous architectures. In the past two years Celeritas has advanced from prototyping a Graphics Processing Unit (GPU)-based single physics model in infinite medium to implementing a full set of electromagnetic (EM) physics processes in complex geometries. The current release of Celeritas, version 0.4, has incorporated full device-based navigation, an event loop in the presence of magnetic fields, and detector hit scoring. New functionality incorporates a scheduler to offload electromagnetic physics to the GPU within a Geant4-driven simulation, enabling straightforward integration of Celeritas into the high energy physics (HEP) experimental frameworks CMSSW and ATLAS FullSimLight. On the Perlmutter supercomputer, Celeritas performs EM physics between 3× and 18× faster using the machine’s Nvidia GPUs compared to using only CPUs, corresponding to an electrical power efficiency up to a factor of 5. When running a multithreaded Geant4 ATLAS test beam application with full hadronic physics, using Celeritas to accelerate the EM physics results in an overall simulation speedup of 1.7–2.2× on GPU and 1.2× on CPU. In a CMS test application using tt¯ events and the prototype Run 4 configuration, compared to Geant4 CPU, Celeritas with a Nvidia A100 improves overall throughput up to a factor of 2.7× but cannot be efficiently shared with more than 8 cores.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2026

A Benchmark Suite for Evaluating Scientific AI Workloads on GPUs

AI applications have been steadily increasing in the allocation portfolio among leadership computing facilities. These applications depend on deep learning frameworks with hardware acceleration and underlying software systems. With the rapid development of applications, software stacks, and hardware devices, it is essential to evaluate the performance of core operations in AI workloads for direction of optimizations and procurement of next-generation high-performance computing (HPC) infrastructures. Currently, most benchmarks lack scientific AI workloads. So, we present DeepKernelBench and the experimental results of evaluating the benchmark suite for early observations and performance comparisons on datacenter GPUs using representative workloads for scientific AI, including Attentions, General matrix multiplications, Geometrics and Fourier neural operations.

Jin, Zheming [Advanced Micro Devices (AMD)]↗
DOE OSTI2024

Materials Science of the Interstitial Doping Process

Particle accelerators are an increasingly important tool for frontier science. Growing initial and operating costs are a significant barrier for upgrades and for new machines. While everything matters, the major cost contributor is the SRF cavities and their ancillary facilities (e.g., cryoplant). Accordingly, the accelerator science community devotes much R&D effort to improving their energy efficiency (increased Q o ) and gradient (E acc ). While improved gradient is at the forefront for certain machines (ILC), improved quality factor has broader impact, benefitting all SRF applications. An important opportunity for accelerator science and technology to move forward arose in the course of building the LCLS-II, the second generation Linac Coherent Light Source at SLAC. At more or less the same time, researchers at Fermilab discovered that introducing a small amount of nitrogen to the niobium surface could improve the mid-range quality factor as much as three-fold. A firm resolution of how nitrogen confers its benefit attracts much current research interest. The key elements of the “nitrogen doping” process were vacuum bake at 800 °C, brief exposure to mTorr of nitrogen at several hundred degrees followed by electropolish (EP) to remove several microns from the surface to eliminate unwanted nitrides: While the process as a whole was novel, the comprising unit operations are familiar to the accelerator community. It was judged reasonable to adopt it as a cost-reduction technology for LCLS-II. Researchers carried out a program of varying process parameters and measuring performance in single-cell cavities, leading to a consensus stable protocol for the project. The transition to vendor fabrication and multiple niobium sources has presented unforeseen challenges of performance variation evidently not connected to anything that could be incorporated in a purchase specification. Moving beyond the high temperature N-doping process above, researchers reported a simplified process consisting entirely of tens of hours anneal in a N atmosphere at low temperatures (~120°C – 160°C) after 800 °C UHV bake. These processes typically yield a few-nm doped layer while the high temperature process yields at least a few to many micron doped layer. Even more recently, oxygen has been used to dope instead of nitrogen which leaves a few-µm O-alloyed layer upon vacuum annealing for 300 °C for ~3 hours. The investigation of these materials is just beginning, but the process simplification they may offer is surely attractive. The very low quantity of material that appears to be significant in the “infusion” process indicates the need for very careful control of gas species available for diffusion into the surface during low temperature treatment, both for process control and research to characterize the underlying dynamics. Oxygen alloying offers the further opportunity to utilize the decomposition of the surface native oxide as the dopant source. It is necessary to understand and (thus) manage this process. We have been supported by the Department of Energy Offices of High Energy Physics and Nuclear Physics to pursue this goal.

36 MATERIALS SCIENCE↗
DOE OSTI2023

Probe microscopy is all you need *

We pose that microscopy offers an ideal real-world experimental environment for the development and deployment of active Bayesian and reinforcement learning methods. Indeed, the tremendous progress achieved by machine learning (ML) and artificial intelligence over the last decade has been largely achieved via the utilization of static data sets, from the paradigmatic MNIST to the bespoke corpora of text and image data used to train large models such as GPT3, DALL·E and others. However, it is now recognized that continuous, minute improvements to state-of-the-art do not necessarily translate to advances in real-world applications. We argue that a promising pathway for the development of ML methods is via the route of domain-specific deployable algorithms in areas such as electron and scanning probe microscopy and chemical imaging. This will benefit both fundamental physical studies and serve as a test bed for more complex autonomous systems such as robotics and manufacturing. Favorable environment characteristics of scanning and electron microscopy include low risk, extensive availability of domain-specific priors and rewards, relatively small effects of exogenous variables, and often the presence of both upstream first principles as well as downstream learnable physical models for both statics and dynamics. Recent developments in programmable interfaces, edge computing, and access to application programming interfaces (APIs) facilitating microscope control, all render the deployment of ML codes on operational microscopes straightforward. We discuss these considerations and hope that these arguments will lead to create novel set of development targets for the ML community by accelerating both real world ML applications and scientific progress.

47 OTHER INSTRUMENTATION↗
DOE OSTI2025

Ensuring Safe, Effective, and Reliable Use of Artificial Intelligence-Based Applications for Nuclear Electricity Generation through a Systems Approach

There is a significant potential to reduce operating and maintenance cost at nuclear power plants using artificial intelligence (AI) and machine learning (ML). For instance, AI/ML has the potential to significant improve work management processes, condition reporting, and plant surveillance activities. However, the nuclear industry has been slow in adopting AI/ML due to several multifaceted barriers discussed in this paper. This work presents such multidisciplinary approach intended to 1) accelerate industry adoption of AI/ML-based applications at nuclear power plants and 2) ensure their safe, reliable, and effective use. This approach is discussed further in this work and will be used to address overarching challenges associated with AI deployment at scale to provide industry guidance that support accelerated adoption of AI/ML technologies throughout the industry.

99 - GENERAL AND MISCELLANEOUS↗
DOE OSTI2020

Application of radiochromic gel dosimetry to commissioning of a megavoltage research linear accelerator for small–field animal irradiation studies

To develop and implement an efficient and accurate commissioning procedure for small-field static beam animal irradiation studies on an MV research linear accelerator (Linatron-M9) using radiochromic gel dosimetry. The research linear accelerator (Linatron-M9) is a 9 MV linac with a static fixed collimator opening of 5.08 cm diameter. Lead collimators were manually placed to create smaller fields of 2 × 2 cm 2 , 1 × 1 cm 2 , and 0.5 × 0.5 cm 2 . Relative dosimetry measurements were performed, including profiles, percent depth dose (PDD) curves, beam divergence, and relative output factors using various dosimetry tools, including a small volume ionization chamber (A14), GAFCHROMIC™ EBT3 film, and Clearview gel dosimeters. The gel dosimeter was used to provide a 3D volumetric reference of the irradiated fields. The Linatron profiles and relative output factors were extracted at a reference depth of 2 cm with the output factor measured relative to the 2 × 2 cm 2 reference field. Absolute dosimetry was performed using A14 ionization chamber measurements, which were verified using a national standards laboratory remote dosimetry service. Absolute dosimetry measurements were confirmed within 1.4% (k = 2, 95% confidence = 5%). The relative output factor of the small fields measured with films and gels agreed with a maximum relative percent error difference between the two methods of 1.1 % for the 1 × 1 cm 2 field and 4.3 % for the 0.5 × 0.5 cm 2 field. These relative errors were primarily due to the variability in the collimator positioning. The measured beam profiles demonstrated excellent agreement for beam size (measured as FWHM), within approximately 0.8 mm (or less). Film measurements were more accurate in the penumbra region due to the film’s finer resolution compared with the gel dosimeter. Following the van Dyk criteria, the PDD values of the film and gel measurements agree within 11% in the buildup region starting from 0.5 cm depth and within 2.6 % beyond maximum dose and into the fall-off region for depths up to 5 cm. The 2D beam profile isodose lines agree within 0.5 mm in all regions for the 0.5 × 0.5 cm 2 and the 1 × 1 cm 2 fields and within 1 mm for the larger field of 2 × 2 cm 2 . The 2D PDD curves agree within approximately 2% of the maximum in the typical therapy region (1–4 cm) for the 1 × 1 cm 2 and 2 × 2 cm 2 and within 5% for the 0.5 × 0.5 cm 2 field. This work provides a commissioning process to measure the beam characteristics of a fixed beam MV accelerator with detailed dosimetric evaluation for its implementation in megavoltage small animal irradiation studies. Radiochromic gel dosimeters are efficient small-field relative dosimetry tools providing 3D dose measurements allowing for full representation of dose, dosimeter misalignment corrections and high reproducibility with low inter-dosimeter variability. Altogether, radiochromic gels are valuable for fast, full relative dosimetry commissioning in comparison to films for application in high-energy small-field animal irradiation studies.

62 RADIOLOGY AND NUCLEAR MEDICINE↗
DOE OSTI2023

Advanced Photon Acceleration Schemes for Tunable XUV/Soft X-Ray Sources

We report on the advances made under the purview of DOE award number DE-SC0019135 that was active during the period: 09/01/2018–08/31/2021. The grant investigated the application of a “flying focus” to the problem of photon acceleration—in which a dynamic refractive index gradient is used to continuously upshift a probe beam’s frequency. Codes were written to describe the creation of ionization waves of arbitrary velocity (IWAVs) for use as a photon-accelerating medium, as well as the behavior of a witness pulse residing in said medium. Experiments first verified the spatiotemporal control over laser intensity provided by a chromatic flying focus, then used that ability to produce small-diameter IWAVs in the far-field with the expected dynamics, and finally demonstrated even further flexibility by producing large-diameter IWAVs in the laser quasi-far-field that maintained the beneficial dynamics. Multiple innovative diagnostics—spectrally resolved Schlieren and spectrally resolved interferometry—were pioneered in order to diagnose the IWAVs. For IWAV production in the laboratory, however, beam quality was identified as a key limitation in the quasi-far-field. Since the original chromatic flying focus was found to result in relatively long (ps duration) intensity peaks, which could limit some applications including photon acceleration, additional techniques were invented to provide similar spatiotemporal control while also retaining ultrashort intensity peaks. While simulations have identified several interesting regimes for photon acceleration—first predicting the upshift of a counterpropagating witness pulse from the optical to the extreme ultraviolet in less than 1 cm, and later obtaining similar shifts in less than 100 μm in a simpler self-seeded configuration—experimental demonstration is left for future work.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2020

Toward fusion plasma scenario planning for NSTX-U using machine-learning-accelerated models

One of the most promising devices for realizing power production through nuclear fusion is the tokamak. To maximize performance, it is preferable that tokamak reactors achieve advanced operating scenarios characterized by good plasma confinement, improved magnetohydrodynamic (MHD) stability, and a largely non-inductively driven plasma current. Such scenarios could enable steady-state reactor operation with high \emph{fusion gain} --- the ratio of produced fusion power to the external power provided through the plasma boundary. Precise and robust control of the evolution of the plasma boundary shape as well as the spatial distribution of the plasma current, density, temperature, and rotation will be essential to achieving and maintaining such scenarios. The complexity of the evolution of tokamak plasmas, arising due to nonlinearities and coupling between various parameters, motivates the use of model-based control algorithms that can account for the system dynamics. In this work, a learning-based accelerated model trained on data from the National Spherical Torus Experiment Upgrade (NSTX-U) is employed to develop planning and control strategies for regulating the density and temperature profile evolution around desired trajectories. The proposed model combines empirical scaling laws developed across multiple devices with neural networks trained on empirical data from NSTX-U and a database of first-principles-based computationally intensive simulations. The reduced execution time of the accelerated model will enable practical application of optimization algorithms and reinforcement learning approaches for scenario planning and control development. An initial demonstration of applying optimization approaches to the learning-based model is presented, including a strategy for mitigating the effect of leaving the finite validity range of the accelerated model. The approach shows promise for actuator planning between experiments and in real-time.

machine learning↗
DOE OSTI2025

Beyond Like-for-Like: A User-centered Approach to Modernizing Legacy Applications

When modernizing a legacy application, it is easy to fall back on a like-for-like replica with new tools and updated design stylings, but this is an opportunity to explore making a more intuitive application that supports user tasks and efficiency. Rather than having a blank canvas-unburdened by legacy tech debt-to create a new application, you are working with an existing application that is integral to accelerator operations and one that expert users are already familiar with. Due to this, you might assume people will prefer the like-for-like, but you could be carrying forward the pain points, processes that are inefficient, and ultimately wind up with an application that no one wants to use because it doesn't solve existing problems. Getting users involved can make all the difference in your approach to modernizing a legacy application that caters to both newer and expert users. It also can bridge the gap between like-for-like and introducing new GUI design. Having a legacy application doesn't have to make the modernized one difficult to develop, as the existing application is a tool in how you move forward with the new application. It provides insight into areas that an application with a clean slate doesn't give you.

Polzin, M. [Fermilab]↗
DOE OSTI2024

Secondary Beams at Jefferson Laboratory

This thesis explores the possibility of producing secondary beams at the Thomas Jefferson National Accelerator Facility (TJNAF). Their possible applications in future experiments are studied using advanced Monte Carlo simulations with state-of-the-art software. In Jefferson Lab’s Hall-A, the interaction of the high-current O(100) µA, medium-energy O(11) GeV electron beam produced by the Continuous Electron Beams Accelerator Facility (CEBAF) with the Beam-Dump (BD) generates a substantial shower of standard model particles. This interaction creates a high-intensity secondary muon and neutrino beam and hypothetical Light Dark Matter (LDM) particles. Additionally, with the proposed Compact Photon Source (CPS), Hall-D could offer the possibility of producing an intense muon beam alongside a photon beam.

Fulci, Antonino↗