Search NASA⌕ Search

SEARCH · Search NASA

Results for “Accelerators, Computing”

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 1,351 records · Page 75

GPU-acceleration of the ELPA2 distributed eigensolver for dense symmetric and hermitian eigenproblems

The solution of eigenproblems is often a key computational bottleneck that limits the tractable system size of numerical algorithms, among them electronic structure theory in chemistry and in condensed matter physics. Large eigenproblems can easily exceed the capacity of a single compute node, thus must be solved on distributed-memory parallel computers. We here present GPU-oriented optimizations of the ELPA two-stage tridiagonalization eigensolver (ELPA2). On top of cuBLAS-based GPU offloading, we add a CUDA kernel to speed up the back-transformation of eigenvectors, which can be the computationally most expensive part of the two-stage tridiagonalization algorithm. Furthermore, we benchmark the performance of this GPU-accelerated eigensolver on two hybrid CPU–GPU architectures, namely a compute cluster based on Intel Xeon Gold CPUs and NVIDIA Volta GPUs, and the Summit supercomputer based on IBM POWER9 CPUs and NVIDIA Volta GPUs. Consistent with previous benchmarks on CPU-only architectures, the GPU-accelerated two-stage solver exhibits a parallel performance superior to the one-stage counterpart. Finally, we demonstrate the performance of the GPU-accelerated eigensolver developed in this work for routine semi-local KS-DFT calculations comprising thousands of atoms.

97 MATHEMATICS AND COMPUTING↗

Language model-accelerated deep symbolic optimization

Symbolic optimization methods have been used to solve varied challenging and relevant problems such as symbolic regression and neural architecture search. However, the current state of the art typically learns each problem from scratch and is unable to leverage pre-existing knowledge and datasets that are available for many applications. Here, inspired by the similarity between sequence representations learned in natural language processing and the formulation of symbolic optimization as a discrete sequence optimization problem, we propose language model-accelerated deep symbolic optimization (LA-DSO), a method that leverages language models to learn symbolic optimization solutions more efficiently. We demonstrate LA-DSO in two tasks: symbolic regression, which allows us to perform extensive experimentation due to its low computation requirements, and computational antibody optimization, which shows that our proposal accelerates learning in challenging real-world problems.

97 MATHEMATICS AND COMPUTING↗

Residual loss due to magnetic flux trapping in niobium superconducting microwave cavities

One of the significant source of radio frequency loss in superconducting microwave cavities is the trapped residual magnetic field during the cooldown due to the incomplete Messier effect. The sensitivity of flux trapping depends on the distribution of impurities, network of dislocations and defects which pin vortices. Recent advancement in cavity processing techniques resulted in very high quality factors with impurity doping, however the rf loss related to the flux trapping increases. High quality factor in SRF cavities is not only limited to particle accelerators but also emerging as an application to quantum computing and quantum information science. In this presentation, we present the results of measurements of flux trapping sensitivity on several 1.3 GHz elliptical accelerating cavities with different grain size, impurity content and surface preparations.

Dhakal, P.↗

Modified dispersion relations and the response of the rotating Unruh-DeWitt detector

We study the response of a rotating monopole detector that is coupled to a massless scalar field which is described by a nonlinear dispersion relation in flat spacetime. Since it does not seem to be possible to evaluate the response of the rotating detector analytically, we resort to numerical computations. Interestingly, unlike the case of the uniformly accelerated detector that has been considered recently, we find that defining the transition probability rate of the rotating detector poses no difficulties. Further, we show that the response of the rotating detector can be computed exactly (albeit, numerically) even when it is coupled to a field that is governed by a nonlinear dispersion relation. We also discuss the response of the rotating detector in the presence of a cylindrical boundary on which the scalar field is constrained to vanish. While superluminal dispersion relations hardly affect the standard results, we find that subluminal dispersion relations can lead to relatively large modifications.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

GTF Transverse and Longitudinal Emittance Data Analysis Technique

The SSRL Gun Test Facility (GTF) was built to develop a high brightness electron injector for the LCLS and has been operational since 1996. Measurements at the GTF include quadrupole scan transverse emittance measurements and linac phase scan longitudinal emittance measurements. Typically the beam size is measured on a screen as a function of a quadrupole current or linac phase and the beam matrix is then fit to the measured data. Often the emittance which is the final result of the measurement is the only number reported. However, the method used to reduce the data to the final emittance value can have a significant effect on the result. This paper describes in painful detail the methods used to analyze the transverse and longitudinal emittance data collected at the GTF.

43 PARTICLE ACCELERATORS↗

QCOR; A Language Extension Specification for the Heterogeneous Quantum-Classical Model of Computation

Quantum computing (QC) is an emerging computational paradigm that leverages the laws of quantum mechanics to perform elementary logic operations. Existing programming models for QC were designed with fault-tolerant hardware in mind, envisioning stand-alone applications. However, the susceptibility of near-term quantum computers to noise limits their stand-alone utility. To better leverage limited computational strengths of noisy quantum devices, hybrid algorithms have been suggested whereby quantum computers are used in tandem with their classical counterparts in a heterogeneous fashion. This modus operandi calls out for a programming model and a high-level programming language that natively and seamlessly supports heterogeneous quantum-classical hardware architectures in a single-source-code paradigm. Motivated by the lack of such a model, we introduce a language extension specification, called QCOR, which enables single-source quantum-classical programming. Programs written using the QCOR library–based language extensions can be compiled to produce functional hybrid binary executables. After defining QCOR’s programming model, memory model, and execution model, we discuss how QCOR enables variational, iterative, and feed-forward QC. Additionally, QCOR approaches quantum-classical computation in a hardware-agnostic heterogeneous fashion and strives to build on best practices of high-performance computing. The high level of abstraction in the language extension is intended to accelerate the adoption of QC by researchers familiar with classical high-performance computing.

97 MATHEMATICS AND COMPUTING↗

Exploring Acceleration Schemes for Nonlinear Heat Equations in Fusion Reactor Modeling

Plasma modeling inside a tokamak reactor poses a challenging and often times computationally expensive endeavor. The process of running these simulations has the potential to increase the predictive capabilities of these models. Here we investigate the effect particular acceleration schemes, mainly Anderson Acceleration, have on a nonlinear transport equation solver, Tango, developed by Parker et al. We show that Anderson acceleration offers a method for decreasing the computational cost through a test problem developed by Shestakov et al, which shows promise towards utilizing Anderson Acceleration in complex transport simulations

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

THE SPECIFIC ACCELERATION RATE IN LOOP-STRUCTURED SOLAR FLARES-IMPLICATIONS FOR ELECTRON ACCELERATION MODELS

We analyze electron flux maps based on RHESSI hard X-ray imaging spectroscopy data for a number of extended coronal-loop flare events. For each event, we determine the variation of the characteristic loop length L with electron energy E, and we fit this observed behavior with models that incorporate an extended acceleration region and an exterior 'propagation' region, and which may include collisional modification of the accelerated electron spectrum inside the acceleration region. The models are characterized by two parameters: the plasma density n in, and the longitudinal extent L{sub 0} of, the acceleration region. Determination of the best-fit values of these parameters permits inference of the volume that encompasses the acceleration region and of the total number of particles within it. It is then straightforward to compute values for the emission filling factor and for the specific acceleration rate (electrons s{sup -1} per ambient electron above a chosen reference energy). For the 24 events studied, the range of inferred filling factors is consistent with a value of unity. The inferred mean value of the specific acceleration rate above E{sub 0} = 20 keV is {approx}10{sup -2} s{sup -1}, with a 1{sigma} spread of about a half-order-of-magnitude above and below this value. We compare these values with the predictions of several models, including acceleration by large-scale, weak (sub-Dreicer) fields, by strong (super-Dreicer) electric fields in a reconnecting current sheet, and by stochastic acceleration processes.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Curved Mesh Correction And Adaptation Tool to Improve COMPASS Electromagnetic Analyses

SLAC performs large-scale simulations for the next-generation accelerator design using higher-order finite elements. This method requires using valid curved meshes and adaptive mesh refinement in complex 3D curved domains to achieve its fast rate of convergence. ITAPS has developed a procedure to address those mesh requirements to enable petascale electromagnetic accelerator simulations by SLAC. The results demonstrate that those correct valid curvilinear meshes can not only make the simulation more reliable but also improve computational efficiency up to 30%. This paper presents a procedure to track moving adaptive mesh refinement in curved domains. The procedure is capable of generating suitable curvilinear meshes to enable large-scale accelerator simulations. The procedure can generate valid curved meshes with substantially fewer elements to improve the computational efficiency and reliability of the COMPASS electromagnetic analyses. Future work will focus on the scalable parallelization of all steps for petascale simulations.

43 PARTICLE ACCELERATORS↗

Progress Towards Accelerating the Unified Model on Hybrid Multi-Core Systems

The cloud microphysics scheme, CASIM, and the radiation scheme, SOCRATES, are two computationally intensive parts within the Met Office's Unified Model (UM). This study enables CASIM and SOCRATES to use accelerated multi-core systems for optimal computational performance of the UM. Using profiling to guide our efforts, we refactored the code for optimal threading and kernel arrangement and implemented OpenACC directives manually or through the CLAW source-to-source translator. Initial porting results achieved 10.02x and 9.25x speedup in CASIM and SOCRATES respectively on 1 GPU compared with 1 CPU core. A granular performance analysis of the strategy and bottlenecks are discussed. These improvements will enable UM to run on heterogeneous computers and a path forward for further improvements is provided.

Zhang, Wei↗

Compact tunable Compton x-ray source from laser-plasma accelerator and plasma mirror

We present an in-depth experimental-computational study of the parameters necessary to optimize a tunable, quasi-monoenergetic, efficient, low-background Compton backscattering (CBS) x-ray source that is based on the self-aligned combination of a laser-plasma accelerator (LPA) and a plasma mirror (PM). The main findings are (1) an LPA driven in the blowout regime by 30 TW, 30 fs laser pulses produce not only a high-quality, tunable, quasi-monoenergetic electron beam, but also a high-quality, relativistically intense (a{sub 0} ∼ 1) spent drive pulse that remains stable in profile and intensity over the LPA tuning range. (2) A thin plastic film near the gas jet exit retro-reflects the spent drive pulse efficiently into oncoming electrons to produce CBS x-rays without detectable bremsstrahlung background. Meanwhile, anomalous far-field divergence of the retro-reflected light demonstrates relativistic “denting” of the PM. Exploiting these optimized LPA and PM conditions, we demonstrate quasi-monoenergetic (50% FWHM energy spread), tunable (75–200 KeV) CBS x-rays, characteristics previously achieved only on more powerful laser systems by CBS of a split-off, counter-propagating pulse. Moreover, laser-to-x-ray photon conversion efficiency (∼6 × 10{sup −12}) exceeds that of any previous LPA-based quasi-monoenergetic Compton source. Particle-in-cell simulations agree well with the measurements.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Nonlinear Dynamics and Phase Space Manipulations of High-Brightness Beams

The project consisted of two main tasks. Each task had several specific aims. The three main objectives of Task 1 were: 1. Investigation of the flat beam technique to produce high transverse emittance ratio longitudinally-compressed bunches, 2. Exploration of a new chicane-like transverse-to-longitudinal phase space exchanger for precise pulse shaping and emittance exchange, 3. Applications of flat-beam with shaped current profile(s) to dielectric wakefield acceleration in an asymmetric waveguide. The outcome of our research was to performed details simulations of flat-beam generation and compression. Some preliminary experiment on flat-beam generation and compression were conducted at the FAST-IOTA injector in parallel to commissioning activities. Likewise, the longitudinal phase-space transformation was not fully explored; we participated in experiments conducted at the Argonne Wakefield Accelerator and especially introduced a simple transverse-shaping technique.

43 PARTICLE ACCELERATORS↗

A GPU Accelerated Mixed‐Precision Finite Difference Informed Random Walker (FDiRW) Solver for Strongly Inhomogeneous Diffusion Problems

In nature, many complex multi‐physics coupling problems exhibit significant diffusivity inhomogeneity, where one process occurs several orders of magnitude faster than others temporally. Simulating rapid diffusion alongside slower processes demands intensive computational resources due to the necessity for small time steps. To address these computational challenges, we have developed an efficient numerical solver named Finite Difference informed Random Walker (FDiRW). In this study, we propose a GPU‐accelerated, mixed‐precision configuration for the FDiRW solver to maximize efficiency through GPU multi‐threaded parallel computation and lower precision computation. Numerical evaluation results reveal that the proposed GPU‐accelerated mixed‐precision FDiRW solver can achieve a 117× speedup over the CPU baseline, while an additional 1.75× speedup is achieved by employing lower precision GPU computation. Notably, for large model sizes, the GPU‐accelerated mixed‐precision FDiRW solver demonstrates strong scaling with the number of nodes used in simulation. When simulating radionuclide absorption processes by porous wasteform particles with a medium‐sized model of 192 × 192 × 192, this approach reduces the total computational time to 10 min, enabling the simulation of larger systems with strongly inhomogeneous diffusivity.

97 MATHEMATICS AND COMPUTING↗

HDBind: encoding of molecular structure with hyperdimensional binary representations

Traditional methods for identifying “hit” molecules from a large collection of potential drug-like candidates rely on biophysical theory to compute approximations to the Gibbs free energy of the binding interaction between the drug and its protein target. These approaches have a significant limitation in that they require exceptional computing capabilities for even relatively small collections of molecules. Increasingly large and complex state-of-the-art deep learning approaches have gained popularity with the promise to improve the productivity of drug design, notorious for its numerous failures. However, as deep learning models increase in their size and complexity, their acceleration at the hardware level becomes more challenging. Hyperdimensional Computing (HDC) has recently gained attention in the computer hardware community due to its algorithmic simplicity relative to deep learning approaches. The HDC learning paradigm, which represents data with high-dimension binary vectors, allows the use of low-precision binary vector arithmetic to create models of the data that can be learned without the need for the gradient-based optimization required in many conventional machine learning and deep learning methods. This algorithmic simplicity allows for acceleration in hardware that has been previously demonstrated in a range of application areas (computer vision, bioinformatics, mass spectrometery, remote sensing, edge devices, etc.). To the best of our knowledge, our work is the first to consider HDC for the task of fast and efficient screening of modern drug-like compound libraries. We also propose the first HDC graph-based encoding methods for molecular data, demonstrating consistent and substantial improvement over previous work. We compare our approaches to alternative approaches on the well-studied MoleculeNet dataset and the recently proposed LIT-PCBA dataset derived from high quality PubChem assays. We demonstrate our methods on multiple target hardware platforms, including Graphics Processing Units (GPUs) and Field Programmable Gate Arrays (FPGAs), showing at least an order of magnitude improvement in energy efficiency versus even our smallest neural network baseline model with a single hidden layer. Our work thus motivates further investigation into molecular representation learning to develop ultra-efficient pre-screening tools. We make our code publicly available at https://github.com/LLNL/hdbind.

59 BASIC BIOLOGICAL SCIENCES↗

Advanced Simulation Tools for Muon-Based Accelerators

The project developed new software tools incorporating the most accurate theoretical calculations and experimental data available for crucial and not-yet-considered physics processes specific to muon accelerators. These new tools were applied to modeling muon ionization cooling channels for a muon collider or a neutrino factory, the facilities that have the potential to discover and explore new exciting fundamental physics but require the development of demanding technologies and innovative concepts. G4beamline, COSY Infinity, and WARP computer codes are particularly well-suited for addressing muon accelerator modeling problems. G4beamline was chosen for its flexibility and modularity, and for being one of the de facto standard codes used by the muon accelerator community. COSY was selected for speed, ability to take into account high-order nonlinear effects essential for the large beams under consideration, and availability of built-in optimization tools. WARP was chosen for its space charge calculation capabilities, speed and flexibility of implementing new lattice elements such as RF cavities. This final technical report summarizes the efforts on implementing advanced new tools in these computer codes addressing the shortcomings in modeling muon physics processes and using these tools to model muon ionization cooling channels.

43 PARTICLE ACCELERATORS↗

Impact response of US Army and National Football League helmet pad systems

Lawrence Livermore National Laboratory [LLNL] was tasked to compare the impact response of NFL helmet pad systems and U.S. Army pad systems compatible with an Advanced Combat Helmet [ACH] at impact velocities up to 20 ft/s. This was a one-year study funded by the U.S. Army and JIEDDO. The Army/JIEDDO point of contact is COL R. Todd Dombroski, DO, JIEDDO Surgeon. LLNL was chosen by committee to perform the research based on prior published computational studies of the mechanical response of helmets and skulls to blast. Our collaborators include the U.S. Army Aeromedical Research Laboratory [USAARL] (a DoD laboratory responsible for impact testing helmets), Team Wendy and Oregon Aero (current and former ACH pad manufacturers), Riddell and Xenith (NFL pad manufacturers), and d3o (general purpose sports pad manufacturer). The manufacturer-supplied pad systems that were studied are shown in the figure below. The first two are the Army systems, which are bilayer foam pads with both hard and soft foam and a water-resistant airtight wrapper (Team Wendy) or a water-resistant airtight coating (Oregon Aero). The next two are NFL pad systems. The Xenith system consists of a thin foam pad and a hollow air-filled cylinder that elastically buckles under load. The Riddell system is a bilayer foam pad that is encased in an inflatable airbag with relief channels to neighboring pads in the helmet. The inflatable airbag is for comfort and provides no enhancement to impact mitigation. The d3o system consists of a rate-sensitive homogeneous dense foam. LLNL performed experiments to characterize the material properties of the individual foam materials and the response of the complete pad systems, to obtain parameters needed for the simulations. LLNL also performed X-ray CT scans of an ACH helmet shell that were used to construct a geometrically accurate computational model of the helmet. Two complementary sets of simulations were performed. The first set of simulations reproduced the experimental helmet impact certification tests performed by USAARL, who provided data for comparison. The goal of this set of simulations was to demonstrate the overall validity of LLNL's computational analyses and methods and understand the general physics of helmet impacts. In these tests and the corresponding simulations, an inverted ACH containing pads and a head-form are dropped onto a hemispherical anvil, at 10 and 14.14 ft/s impact velocities. The simulations predicted peak accelerations (the metric used by USAARL for comparing the performance of pad systems), rebound velocities, and impact durations consistent with the experimental data, thus demonstrating the validity and relevance of the simulation methods. Because the NFL pad systems are approximately double the thickness of the U.S. Army pads, they do not fit into the ACH. As a result, the NFL pads could not be simply placed into an ACH shell in either a simulation or an experiment without modifying their size and shape. Since impact mitigation depends critically on the available stopping distance and the area over which the stopping force is applied, it is important to consider identically shaped pads in order to compare their performance in a fair and meaningful manner. Consequently, the second set of simulations utilized a simplified simulation geometry consisting of a 5 kg cylindrical impactor (equal in mass to a head) striking equally sized pads from each manufacturer. The simulated bilayer foam pads had the same proportions of hard and soft foam as the actual pad systems, while the Xenith pads were simulated as a bilayer foam pad with material properties adjusted to give the same response as the actual Xenith pads. The effects of trapped air were included in the simulations of the Team Wendy and Oregon Aero pads. All simulations used material properties derived from the experiments conducted at LLNL. The acceleration history of the center of mass of the impactor was used to calculate the Head Injury Criterion (HIC) for each simulation, to assess the pad performance. The HIC is a well-established metric that combines both acceleration and duration of impact to assess the danger of injury, and is a more robust measure than peak acceleration. Our key findings are: (1) The performance of a pad depends on the range of impact velocities. At lower impact velocity, softer pads perform better. At higher impact velocity, harder pads perform better; (2) Thicker pads perform better at all velocities, but especially at high velocities; and (3) For comparable thicknesses, neither the NFL systems nor the Oregon Aero pads outperform the Team Wendy pads currently used in the ACH system in militarily-relevant impact scenarios (impact speeds less than 20 ft/s). The second finding suggests a commercial off-the-shelf solution for mitigating impact-related traumatic brain injury to soldiers.

42 ENGINEERING↗

Meta-optic accelerators for object classifiers

Rapid advances in deep learning have led to paradigm shifts in a number of fields, from medical image analysis to autonomous systems. These advances, however, have resulted in digital neural networks with large computational requirements, resulting in high energy consumption and limitations in real-time decision-making when computation resources are limited. Here, we demonstrate a meta-optic–based neural network accelerator that can off-load computationally expensive convolution operations into high-speed and low-power optics. In this architecture, metasurfaces enable both spatial multiplexing and additional information channels, such as polarization, in object classification. End-to-end design is used to co-optimize the optical and digital systems, resulting in a robust classifier that achieves 93.1% accurate classification of handwriting digits and 93.8% accuracy in classifying both the digit and its polarization state. This approach could enable compact, high-speed, and low-power image and information processing systems for a wide range of applications in machine vision and artificial intelligence.

42 ENGINEERING↗