Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Vector Processing”

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 631 records · Page 35
All sourcesNASA NTRSDOE OSTI
DOE OSTI2023

An SU(5) × U(1)' SUSY GUT with a “vector-like chiral” fourth family to fit all low energy data, including the muon g – 2

An additional generation of quarks and leptons and their SUSY counterparts, which are vector-like under the Standard Model gauge group but are chiral with respect to the new U(1) 3–4 gauge symmetry, are added to the Minimal Supersymmetric Standard Model (MSSM). We show that this model is a GUT and unifies the three SM gauge couplings and also the additional U(1) 3–4 coupling at a GUT scale of ≈ 5 × 10 16 GeV and explains the experimentally observed deviation of the muon g – 2. We also fit the quark flavor changing processes consistent with the latest experimental data and look at the effect of the new particles on the W boson mass without obviously conflicting with the observed masses of particles, CKM matrix elements, neutrino mixing angles, their mass differences, and the lepton-flavor violating bounds. This model predicts sparticle masses less than 25 TeV, with a gluino mass ≈ 2.3 – 3 TeV consistent with constraints, and one of the neutralinos as the LSP with a mass of ≈ 480 – 580 GeV, which is a potential dark matter candidate. The model is string theory motivated and predicts the VL quarks, leptons, a massive Z' and two Dirac neutrinos at the TeV scale and the branching ratios of μ → eγ, τ → μγ and τ → 3μ with BR(μ → eγ) within reach of future experiments.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2019

Beyond the effectiveness factor: Multi-step reactions with intraparticle diffusion limitations

Though common in many industrial chemical processes, multi-step reactions occurring within porous catalyst particles provide significant challenges to previously established modeling frameworks. In this work, a system of reaction-diffusion equations is solved analytically via eigenanalysis. The intraparticle concentration profiles are integrated over the particle volume to obtain effectiveness factors for each reacting species, referred to herein as a multi-step effectiveness vector (MEV). The MEV solution is tested for the case of uniform flow past a stationary catalyst particle with a state-of-the-art, multi-step reaction scheme for fluid catalytic cracking (FCC). Outputs from particle-resolved direct numerical simulation (DNS) are compared to outputs from ordinary differentials with effective reaction closures coming from the MEV solution, the classic single-step effectiveness factor (SEF) (R. Aris, 'The mathematical theory of diffusion and reaction in permeable catalysts: The theory of the steady state,' Oxford University Press, 1975), and utilizing the intrinsic rates as-is. When compared to DNS, the MEV solution shows excellent agreement for the formation of all species and quantitatively captures the intraparticle concentration profiles. By contrast, the intrinsic treatment shows an over-prediction for product formation and the SEF treatment shows mixed levels of accuracy that depend upon the reaction coupling. Specifically, the SEF treatment accurately characterizes product formation if the kinetic scheme is weakly coupled but shows significant departure if stronger coupling is present. The computational cost associated with the MEV solution is associated with eigendecomposition of the Thiele matrix, an N x N matrix where N is the number of reacting components. Overall, the method is shown to be relatively easy to implement and may significantly enhance the fidelity of chemical reactor simulations that employ multi-step reaction schemes.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2023

When less is more: How increasing the complexity of machine learning strategies for geothermal energy assessments may not lead toward better estimates

Previous moderate- and high-temperature geothermal resource assessments of the western United States utilized data-driven methods and expert decisions to estimate resource favorability. Although expert decisions can add confidence to the modeling process by ensuring reasonable models are employed, expert decisions also introduce human and, thereby, model bias. This bias can present a source of error that reduces the predictive performance of the models and confidence in the resulting resource estimates. Our study aims to develop robust data-driven methods with the goals of reducing bias and improving predictive ability. We present and compare nine favorability maps for geothermal resources in the western United States using data from the U.S. Geological Survey's 2008 geothermal resource assessment. Two favorability maps are created using the expert decision-dependent methods from the 2008 assessment (i.e., weight-of-evidence and logistic regression). With the same data, we then create six different favorability maps using logistic regression (without underlying expert decisions), XGBoost, and support-vector machines paired with two training strategies. The training strategies are customized to address the inherent challenges of applying machine learning to the geothermal training data, which have no negative examples and severe class imbalance. We also create another favorability map using an artificial neural network. We demonstrate that modern machine learning approaches can improve upon systems built with expert decisions. We also find that XGBoost, a non-linear algorithm, produces greater agreement with the 2008 results than linear logistic regression without expert decisions, because the expert decisions in the 2008 assessment rendered the otherwise linear approaches non-linear despite the fact that the 2008 assessment used only linear methods. The F1 scores for all approaches appear low (F1 score < 0.10), do not improve with increasing model complexity, and, therefore, indicate the fundamental limitations of the input features (i.e., training data). Until improved feature data are incorporated into the assessment process, simple non-linear algorithms (e.g., XGBoost) perform equally well or better than more complex methods (e.g., artificial neural networks) and remain easier to interpret.

15 GEOTHERMAL ENERGY↗
DOE OSTI2022

Cross section measurements of the $e^+e^–$ → $D^{*+}D^{*–}$ and $e^+e^–$ → $D^{*+}D^{–}$ processes at center-of-mass energies from 4.085 to 4.600 GeV

The Born cross sections of the $e^+e^–$ → $D^{*+}D^{*–}$ and $e^+e^–$ → $D^{*+}D^{–}$ processes are measured using $e^+e^–$ collision data collected with the BESIII experiment at center-of-mass energies from 4.085 to 4.600 GeV, corresponding to an integrated luminosity of 15.7 fb -1 . The results are consistent with and more precise than the previous measurements by the Belle, Babar and CLEO collaborations. The measurements are essential for understanding the nature of vector charmonium and charmonium-like states.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2024

Light-driven nanoscale vectorial currents

Controlled charge flows are fundamental to many areas of science and technology, serving as carriers of energy and information, as probes of material properties and dynamics and as a means of revealing or even inducing broken symmetries. Emerging methods for light-based current control offer particularly promising routes beyond the speed and adaptability limitations of conventional voltage-driven systems. However, optical generation and manipulation of currents at nanometre spatial scales remains a basic challenge and a crucial step towards scalable optoelectronic systems for microelectronics and information science. Here we introduce vectorial optoelectronic metasurfaces in which ultrafast light pulses induce local directional charge flows around symmetry-broken plasmonic nanostructures, with tunable responses and arbitrary patterning down to subdiffractive nanometre scales. Local symmetries and vectorial currents are revealed by polarization-dependent and wavelength-sensitive electrical readout and terahertz (THz) emission, whereas spatially tailored global currents are demonstrated in the direct generation of elusive broadband THz vector beams 17 . We show that, in graphene, a detailed interplay between electrodynamic, thermodynamic and hydrodynamic degrees of freedom gives rise to rapidly evolving nanoscale driving forces and charge flows under the extremely spatially and temporally localized excitation. These results set the stage for versatile patterning and optical control over nanoscale currents in materials diagnostics, THz spectroscopies, nanomagnetism and ultrafast information processing.

36 MATERIALS SCIENCE↗
DOE OSTI2022

Effect of Foot-and-Mouth Disease Virus 2B Viroporin on Expression and Extraction of Mammalian Cell Culture Produced Foot-and-Mouth Disease Virus-like Particles

To improve the production of foot-and-mouth disease (FMD) molecular vaccines, we sought to understand the effects of the FMD virus (FMDV) 2B viroporin in an experimental, plasmid-based, virus-like particle (VLP) vaccine. Inclusion of the FMDV viroporin 2B into the human Adenovirus 5 vectored FMD vaccine enhanced transgene expression despite independent 2B expression negatively affecting cell viability. Evaluating both wildtype 2B and mutants with disrupted viroporin activity, we confirmed that viroporin activity is detrimental to overall transgene expression when expressed independently. However, the incorporation of 2B into an FMD molecular vaccine construct containing a wildtype FMDV 3C protease, a viral encoded protease responsible for processing structural proteins, resulted in enhancement of transgene expression, validating previous observations. This benefit to transgene expression was negated when using the FMDV 3CL127P mutant, which has reduced processing of host cellular proteins, a reversion resulting from 2B viroporin activity. Inclusion of 2B into VLP production constructs also adversely impacted antigen extraction, a possible side effect of 2B-dependent rearrangement of cellular membranes. These results demonstrate that inclusion of 2B enhanced transgene expression when a wildtype 3C protease is present but was detrimental to transgene expression with the 3CL127P mutant. This has implications for future molecular FMD vaccine constructs, which may utilize mutant FMDV 3C proteases.

2B↗
DOE OSTI2019

First M87 Event Horizon Telescope Results. V. Physical Origin of the Asymmetric Ring

The Event Horizon Telescope (EHT) has mapped the central compact radio source of the elliptical galaxy M87 at 1.3 mm with unprecedented angular resolution. Here we consider the physical implications of the asymmetric ring seen in the 2017 EHT data. To this end, we construct a large library of models based on general relativistic magnetohydrodynamic (GRMHD) simulations and synthetic images produced by general relativistic ray tracing. We compare the observed visibilities with this library and confirm that the asymmetric ring is consistent with earlier predictions of strong gravitational lensing of synchrotron emission from a hot plasma orbiting near the black hole event horizon. The ring radius and ring asymmetry depend on black hole mass and spin, respectively, and both are therefore expected to be stable when observed in future EHT campaigns. Overall, the observed image is consistent with expectations for the shadow of a spinning Kerr black hole as predicted by general relativity. If the black hole spin and M87's large scale jet are aligned, then the black hole spin vector is pointed away from Earth. Models in our library of non-spinning black holes are inconsistent with the observations as they do not produce sufficiently powerful jets. At the same time, in those models that produce a sufficiently powerful jet, the latter is powered by extraction of black hole spin energy through mechanisms akin to the Blandford-Znajek process. We briefly consider alternatives to a black hole for the central compact object. Analysis of existing EHT polarization data and data taken simultaneously at other wavelengths will soon enable new tests of the GRMHD models, as will future EHT campaigns at 230 and 345 GHz.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2018

A Hierarchical Multi-Agent Based Anomaly Detection for Wide-Area Protection in Smart Grid

Future smart grid capabilities provide assurance to expand the advanced information and communication technologies to evolve into densely interconnected cyber physical system. Remedial Action Scheme (RAS), widely used for wide-area protection, relies on the interconnected networks and data sharing devices, which are exposed to the multitude of vulnerabilities. This paper presents our proposed approach to developing multi-agent based RAS scheme against the system-aware stealthy cyber-attacks. Specifically, we propose the two-level hierarchical architecture which consists of distributed local RAS controllers (RAScs) as local agents, operating at different zones/ areas, which are constantly monitored by an overseer, the central agent. The local controllers receive local and randomly changing outside zonal measurements and cyclically forwards to the overseer. The overseer identifies the corrupted controller using the anomaly detection algorithm which processes the measurements coming from the local controllers, compute measurement errors using local and outside zonal measurements, perform validation checks, and finally detect anomalies based on the two-step verification. Next, as a proof of concept, we have implemented and validated the proposed methodology in cyber physical environment at Iowa State’s PowerCyber testbed. We have also implemented the coordinated attack vectors which involve corrupting the local controller and later performing stealthy attacks on the system’s generator. We have evaluated its performance during the online testing in terms of detection rate and Iatency. The experimental results show that it is efficient in detecting different classes of attacks, including ramp and pulse attacks.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2017

Dark Energy Survey Year 1 Results: Multi-Probe Methodology and Simulated Likelihood Analyses

We present the methodology for and detail the implementation of the Dark Energy Survey (DES) 3x2pt DES Year 1 (Y1) analysis, which combines configuration-space two-point statistics from three different cosmological probes: cosmic shear, galaxy-galaxy lensing, and galaxy clustering, using data from the first year of DES observations. We have developed two independent modeling pipelines and describe the code validation process. We derive expressions for analytical real-space multi-probe covariances, and describe their validation with numerical simulations. We stress-test the inference pipelines in simulated likelihood analyses that vary 6-7 cosmology parameters plus 20 nuisance parameters and precisely resemble the analysis to be presented in the DES 3x2pt analysis paper, using a variety of simulated input data vectors with varying assumptions. We find that any disagreement between pipelines leads to changes in assigned likelihood $\Delta \chi^2 \le 0.045$ with respect to the statistical error of the DES Y1 data vector. We also find that angular binning and survey mask do not impact our analytic covariance at a significant level. We determine lower bounds on scales used for analysis of galaxy clustering (8 Mpc$~h^{-1}$) and galaxy-galaxy lensing (12 Mpc$~h^{-1}$) such that the impact of modeling uncertainties in the non-linear regime is well below statistical errors, and show that our analysis choices are robust against a variety of systematics. These tests demonstrate that we have a robust analysis pipeline that yields unbiased cosmological parameter inferences for the flagship 3x2pt DES Y1 analysis. We emphasize that the level of independent code development and subsequent code comparison as demonstrated in this paper is necessary to produce credible constraints from increasingly complex multi-probe analyses of current data.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2017

CUDA Computation of the Feynman Distribution

In 2006, NVIDIA Corporation introduced the Computer Unified Device Architecture (CUDA) for C, C++, and Fortran programming languages. The CUDA parallel computing platform uses the Graphics Processing Unit (GPU) of NVIDIA graphics card, and it is designed for parallel computations. In these computations, the same algorithms are executed on the data in parallel. The GPU contains many Streaming Multiprocessors (SM) that can independently execute blocks of threads. A thread is a sequence of serial programmed instructions. The SM executes threads in groups of 32 and a group of 32 threads is called warp. The threads within a block can be synchronized and have access to a shared memory that is visible only to them. Conversely, blocks of threads cannot be synchronized and have access only to the global memory of the graphics card. This memory is called device memory and is independent of the host memory controlled by the Central Processing Unit (CPU). A single thread has access to a local memory (register) that is not visible to any other thread. The part of the program that is executed on the graphics card is called kernel. Each thread of the kernel has a unique identification index that is accessible by the in-built variable threadIdx; the latter is a 3-component vector. A threads block can contain up to 1024 threads. Blocks of threads have a unique identification index that is accessible by the in-built variable blockIdx; the latter is a 3-component vector. The in-built variable blockDim contains the number of threads per block; this variable is a 3-component vector. Threads and blocks represent different levels of parallelism; a kernel is executed as a grid of blocks of threads. Figure 1 illustrates a grid of thread blocks for a GPU with 2 SM. A SM can execute multiple blocks of threads in parallel; a block of threads is executed on a single SM. A GPU can execute multiple kernels concurrently. This work has used the Quadro M5000 graphics card with the following technical specifications: - 2048 cores with 861 MHz frequency, - 16 streaming multiprocessors, - 5.2 trillion transistors, - 8.2 Gb device memory, - 64 kb register memory (per thread), - 48 kb shared memory (per block) - 211.6 GB/s memory bandwidth, and - 3.5 GFLOPS floating point performance.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2017

miR-429 inhibits metastasis by targeting KIAA0101 in Soft Tissue Sarcoma

Soft tissue sarcomas (STS) are a heterogeneous group of rare tumors with high metastatic potential. There being only a handful of publication on metastasis of STS, we investigated the miRNA mediated target gene regulations in modulating the metastatic processes in this cancer. In this study, we amalgamated gene and miRNA expression profiles of high-grade STS samples with miRNA target predictions and identified miR-429 targeting KIAA0101 as a novel pair, which remain unexplored in STS metastasis. We validated their expression in metastatic fibrosarcoma cell line, HT1080 and performed several functional assays using miRNA mimics and KIAA0101 over-expression vector to confirm their role in metastasis. We observed miR-429 is downregulated in HT1080 cells and acting as an anti-metastatic miRNA that inhibited proliferation, migration, anchorage independent growth and invasion by de-repressing KIAA0101. Moreover, the renilla luciferase reporter assay confirmed that miR-429 targets KIAA0101 by binding to its 3{sup /}UTR and influence its expression. Taken together, our work demonstrated miR-429 mediates deregulation of KIAA0101 by acting as an anti-metastatic miRNA that targets KIAA0101 pro-metastatic gene during metastasis of STS. - Highlights: • miR-429 inhibits metastasis by targeting KIAA0101 in Soft Tissue Sarcoma. • miR-429 suppresses proliferation, migration, anchorage independent growth and invasion in HT1080. • We validated targeting of miR-429 and KIAA0101 by reporter assay. • KIAA0101 overexpression is involved in decreased survival of STS patients.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2019

Effect of a magnetic field on polarisation of light in an optical fibre with a random distribution of linear birefringence

A numerical model is proposed to describe the evolution of polarisation of a light wave propagating through a telecommunication fibre with random linear birefringence in a magnetic field. As a result of statistical processing of a set of numerical simulation results, a convenient phenomenological formula is obtained for the first time for the dependence of the average value of the polarisation rotation angle on the magnetic field, the fibre parameters and its length. It is found that the average value of the polarisation rotation angle in a long telecommunication fibre in the representation of the Stokes vectors linearly depends on the applied longitudinal magnetic field (as in the classical Faraday effect for an isotropic medium) but is proportional to the root of the fibre length. It is theoretically shown and experimentally confirmed that the polarisation rotation angle for an extended segment of a telecommunication fibre (50 km) is two orders of magnitude less than that for an isotropic fibre of the same length and material. (optical fibres)

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2022

Magnesium- and intermetallic alloys-based hydrides for energy storage: modelling, synthesis and properties

Hydrides based on magnesium and intermetallic compounds provide a viable solution to the challenge of energy storage from renewable sources, thanks to their ability to absorb and desorb hydrogen in a reversible way with a proper tuning of pressure and temperature conditions. Therefore, they are expected to play an important role in the clean energy transition and in the deployment of hydrogen as an efficient energy vector. This review, by experts of Task 40 ‘Energy Storage and Conversion based on Hydrogen’ of the Hydrogen Technology Collaboration Programme of the International Energy Agency, reports on the latest activities of the working group ‘Magnesium- and Intermetallic alloys-based Hydrides for Energy Storage’. The following topics are covered by the review: multiscale modelling of hydrides and hydrogen sorption mechanisms; synthesis and processing techniques; catalysts for hydrogen sorption in Mg; Mg-based nanostructures and new compounds; hydrides based on intermetallic TiFe alloys, high entropy alloys, Laves phases, and Pd-containing alloys. Finally, an outlook is presented on current worldwide investments and future research directions for hydrogen-based energy storage.

Energy storage↗
DOE OSTI2023

Imaging Nanoscale Energy Transport and Conversion with Ultrafast Electron Microscopy (Final Technical Report)

Light-matter interactions are ubiquitous in nature and reside at the heart of innumerable technologies. The cascade of processes that occur when a material absorbs a photon of light are exceedingly complex and are interwoven in both space and time, rendering precise determination of the atomic-scale and ultrafast mechanisms immensely challenging. The advent of methods for generating short pulses of light several decades ago led to major advances in understanding the initial moments of light absorption and the resultant effects, though directly interrogating the response of the atoms within the material continued to prove challenging. More recently, methods for generating ultrashort pulses of X-rays and fast electrons have opened the way to probing photoinduced structural dynamics of a wide range of matter in multiple phases. An especially promising laboratory-scale method is ultrafast electron microscopy (UEM), wherein the modalities of conventional transmission electron microscopes (imaging, diffraction, spectroscopy) are extended into the femtosecond temporal regime. Here, the ultrafast imaging and diffraction modalities of UEM were used to study the transient structural aspects of photoexcitation of semiconducting materials (e.g., spatially-resolved electron-phonon coupling, excitation and emission of acoustic phonons, and discrete nanoscale scattering processes). The project had three main objectives: (1) determination of the excitation mechanisms of dense, hypersonic charge-carrier waves and the spatially mediated means by which they couple to the lattice via coherent phonon emission, especially with UEM imaging, wherein effects of nanoscale structural and morphological features on the coupling and relaxation dynamics are expected to dictate nucleation sites and preferred wave vectors, (2) elucidation of photoinduced acoustic-phonon seeding, emergence, propagation, and decay over nanoscale crystal regions and especially with respect to local strain fields and atomic-scale disruptions in lattice order, and (3) interwoven with the first two were efforts aimed at realizing combined angstrom-femtosecond spatiotemporal imaging with UEM. The outcomes and impacts of this project were the generation of new knowledge with respect to fundamental light-matter interactions and, especially, the spatially-mediated excitation and evolution of the structural response of materials following coherent photoexcitation. Importantly, the spatial and temporal resolutions of the UEM imaging modalities used are well-suited for such studies and enable spatially-resolved mechanisms to be determined as a function of atomic order, structural features, and morphology. In addition, ultrafast crystallographic measurements were used to correlate real- and reciprocal-space dynamics in order to determine atomic-scale preferential wave vectors and ultrafast scattering mechanisms, especially as dictated by specimen boundary conditions. The obtained results, as detailed in peer-reviewed publications and presentations, illustrate the importance of ultrafast, angstrom-scale real-space imaging for developing a comprehensive understanding of energy transport and conversion in materials.

47 OTHER INSTRUMENTATION↗
DOE OSTI2018

The Structure and Dynamics of the Subparsec Jet in M87 Based on 50 VLBA Observations over 17 Years at 43 GHz

The central radio source in M87 provides the best opportunity to study jet formation because it has a large angular size for the gravitational radius of the black hole and has a bright jet that is well resolved by very long baseline interferometry observations. We present intensive monitoring observations from 2007 and 2008, plus roughly annual observations that span 17 years, all made with the the Very Long Baseline Array at 43 GHz with a resolution of about 30 by 60 R S . Our high dynamic range images clearly show the wide opening angle structure and the counterjet. The jet and counterjet are nearly symmetric in the inner 1.5 mas (0.12 pc in projection), with both being edge brightened. Both show deviations from parabolic shape in the form of an initial rapid expansion and subsequent contraction followed by further rapid expansion and, beyond the visible counterjet, subsequent collimation. Proper motions and counterjet/jet intensity ratios both indicate acceleration from apparent speeds of ≲ 0.5c to ≳ 2c in the inner ~2 mas and suggest a helical flow. The jet displays a sideways shift with an approximately 8–10 yr quasi-periodicity. Here, the shift propagates outward nonballistically and significantly more slowly than the flow speed revealed by the fastest-moving components. Polarization data show a systematic structure with magnetic field vectors that suggest a toroidal field close to the core.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2025

NEML2: An efficient and modular multiphysics constitutive modeling library for hybrid computing environments

This paper presents NEML2, an open-source, high-performance library developed for constitutive material modeling, designed to support the flexible and modular development of models for complex material behavior. Building on the foundational structure of its predecessor, NEML, the NEML2 library introduces significant improvements, including enhanced vectorization, automatic differentiation, and seamless integration with PyTorch, facilitating the application of machine learning techniques in material simulations. NEML2 provides a C++ backend with Python bindings, enabling users to create custom material models that can be executed efficiently on both CPU and GPU platforms. The library also supports coupling with Multiphysics simulation frameworks like MOOSE, making it suitable for realistic simulations involving coupled physical processes. Rigorous quality assurance through unit and regression testing ensures the reliability of results, while the extensible, user-friendly design encourages collaboration and reproducibility across the scientific community. This paper provides an overview of NEML2’s architecture, core features, and applications, highlighting its impact on accelerating material qualification and advancing computational methods in materials science.

GPU↗
DOE OSTI2025

A two-level GPU-accelerated incomplete LU preconditioner for general sparse linear systems

This paper presents a parallel preconditioning approach based on incomplete LU (ILU) factorizations in the framework of Domain Decomposition (DD) for general sparse linear systems. We focus on distributed memory parallel architectures, specifically, those that are equipped with graphic processing units (GPUs). In addition to block-Jacobi, we present general purpose two-level ILU Schur complement-based approaches, where different strategies are presented to solve the coarse-level reduced system. These strategies are combined with modified ILU methods in the construction of the coarse-level operator, in order to effectively remove smooth errors by targeting an algebraically smooth vector. We leverage available GPU-based sparse matrix kernels to accelerate the setup and the solve phases of the proposed ILU preconditioner. We evaluate the efficiency of the proposed methods as a smoother for algebraic multigrid (AMG) and as a preconditioner for Krylov subspace methods on challenging anisotropic diffusion problems and a collection of general sparse matrices.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

CrossSim Inference Manual v2.0

Neural networks are largely based on matrix computations. During forward inference, the most heavily used compute kernel is the matrix-vector multiplication (MVM): $W \vec{x} $. Inference is a first frontier for the deployment of next-generation hardware for neural network applications, as it is more readily deployed in edge devices, such as mobile devices or embedded processors with size, weight, and power constraints. Inference is also easier to implement in analog systems than training, which has more stringent device requirements. The main processing kernel used during inference is the MVM.

97 MATHEMATICS AND COMPUTING↗