SEARCH · Engineering Papers
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.
Generating graphics to faithfully represent information can be a computationally intensive task. A way of using the Massively Parallel Processor to generate images by ray tracing is presented. This technique uses sort computation, a method of performing generalized routing interspersed with computation on a single-instruction-multiple-data (SIMD) computer.
In 1967 Amdahl expressed doubts about the ultimate utility of multiprocessors. The formulation, now called Amdahl's law, became part of the computing folklore and has inspired much skepticism about the ability of the current generation of massively parallel processors to efficiently deliver all their computing power to programs. The widely publicized recent results of a group at Sandia National Laboratory, which showed speedup on a 1024 node hypercube of over 500 for three fixed size problems and over 1000 for three scalable problems, have convincingly challenged this bit of folklore and have given new impetus to parallel scientific computing.
The simulation of the computational plasticity on a complex structure remains a formidable computational task, especially when a highly nonlinear, complex material model was used. It appears that the computational requirements for a such problem can only be satisfied by massively parallel architectures. In order to effectively harness the tremendous computational power provided by such architectures, it is imperative to investigate and to study the algorithmic and implementation issues pertaining to dynamic load balancing for computational plasticity on a highly parallel, distributed-memory, multiple-instruction, multiple-data computers. This paper will measure the effectiveness of the algorithms developed in handling the dynamic load balancing.
During the past two decades, there has been significant progress in the field of numerical simulation of unsteady compressible viscous flows. At present, a variety of solution techniques exist such as the transonic small disturbance analyses (TSD), transonic full potential equation-based methods, unsteady Euler solvers, and unsteady Navier-Stokes solvers. These advances have been made possible by developments in three areas: (1) improved numerical algorithms; (2) automation of body-fitted grid generation schemes; and (3) advanced computer architectures with vector processing and massively parallel processing features. In this work, the GMRES scheme has been considered as a candidate for acceleration of a Newton iteration time marching scheme for unsteady 2-D and 3-D compressible viscous flow calculation; from preliminary calculations, this will provide up to a 65 percent reduction in the computer time requirements over the existing class of explicit and implicit time marching schemes. The proposed method has ben tested on structured grids, but is flexible enough for extension to unstructured grids. The described scheme has been tested only on the current generation of vector processor architecture of the Cray Y/MP class, but should be suitable for adaptation to massively parallel machines.
Scientific applications at exascale generate and analyze massive amounts of data. A critical requirement of these applications is the capability to access and manage this data efficiently on exascale systems. As such, parallel I/O, the key technology enables moving data between compute nodes and storage, faces monumental challenges from new applications, memory, and storage architectures considered in the designs of exascale systems. As the storage hierarchy is expanding to include node-local persistent memory, burst buffers, etc., as well as disk-based storage, data movement among these layers must be efficient. Parallel I/O libraries of the future should be capable of handling file sizes of many terabytes and beyond. In this paper, we describe new capabilities we have developed in Hierarchical Data Format version 5 (HDF5), the most popular parallel I/O library for scientific applications. HDF5 is one of the most used libraries at the leadership computing facilities for performing parallel I/O on existing HPC systems. The state-of-the-art features we describe include: Virtual Object Layer (VOL), Data Elevator, asynchronous I/O, full-featured single-writer and multiple-reader (Full SWMR), and parallel querying. In this paper, we introduce these features, their implementations, and the performance and feature benefits to applications and other libraries.
Inspired by the in-memory computing architectures of biological systems, neuromorphic computing using crossbar arrays of artificial synapses based on non-volatile memory (NVM) devices with variable conductances has emerged as a new paradigm to enable massively parallel and ultra-low power computing hardware for data centric applications. Although inference has been demonstrated successfully using crossbars based on a variety of NMV technologies, efficient learning and scaling to large arrays (>10 6 elements) remains a challenge due to the synaptic elements' non-ideal electrical characteristics which degrades ANN accuracy. A further challenge is that in the conductive state memristors draw large currents >μA resulting in significant voltage drops in the interconnect wires and increased probability of failure in scaled arrays. We suggest the organic polymer redox transistor (RT) is an alternate approach that could solve many of these challenges, enabling both inference and parallel outer product updates, as recently demonstrated by Fuller et al. An RT consists of redox-active channel and gate electrodes in contact with a liquid or solid electrolyte. lon insertion through the electrolyte controls the channel electronic conductivity, while electron transfer through an external circuit maintains overall charge neutrality. Unlike a rechargeable battery, in the RT the voltage built-up across the electrolyte is kept to a minimum (typically <100 mV) by using the same material for the gate and channel. Elimination of the voltage offset simplifies integration of the RT into programmable arrays by enabling the use of various selectors. RTs based on inorganic and organic materials have been recently demonstrated with conductance tuning occurring at potentials of just a few mV and hundreds to thousands of linearly and symmetrically programmable conductance states, enabling near ideal accuracy in neural network simulations. Introduced in the 1980's, redox transistors with metallic gate electrodes and organic channel materials, also known as organic electrochemical transistors (OECTs), have been explored for a variety of applications such as chem- and bio-sensing, neural interfaces, and low cost printed circuits. A typical channel material for OECTs is the conducting polymer poly(3,4-ethylenedioxythiophene) doped with poly(styrene sulfonate) (PEDOT:PSS). PEDOT is a p-type semiconducting polymer with mobile positively charged polarons that hop chain-to-chain.
Altimetric data from the TOPEX/POSEIDON mission will be used for studies of global ocean circulation and marine geophysics. However, it is first necessary to remove the ocean tides, which are aliased in the raw data. The tides are constrained by the two distinct types of information: the hydrodynamic equations which the tidal fields of elevations and velocities must satisfy, and direct observational data from tide gauges and satellite altimetry. Here we develop and apply a generalized inverse method, which allows us to combine rationally all of this information into global tidal fields best fitting both the data and the dynamics, in a least squares sense. The resulting inverse solution is a sum of the direct solution to the astronomically forced Laplace tidal equations and a linear combination of the representers for the data functionals. The representer functions (one for each datum) are determined by the dynamical equations, and by our prior estimates of the statistics or errors in these equations. Our major task is a direct numerical calculation of these representers. This task is computationally intensive, but well suited to massively parallel processing. By calculating the representers we reduce the full (infinite dimensional) problem to a relatively low-dimensional problem at the outset, allowing full control over the conditioning and hence the stability of the inverse solution. With the representers calculated we can easily update our model as additional TOPEX/POSEIDON data become available. As an initial illustration we invert harmonic constants from a set of 80 open-ocean tide gauges. We then present a practical scheme for direct inversion of TOPEX/POSEIDON crossover data. We apply this method to 38 cycles of geophysical data records (GDR) data, computing preliminary global estimates of the four principal tidal constituents, M(sub 2), S(sub 2), K(sub 1) and O(sub 1). The inverse solution yields tidal fields which are simultaneously smoother, and in better agreement with altimetric and ground truth data, than previously proposed tidal models. Relative to the 'default' tidal corrections provided with the TOPEX/POSEIDON GDR, the inverse solution reduces crossover difference variances significantly (approximately 20-30%), even though only a small number of free parameters (approximately equal to 1000) are actually fit to the crossover data.
This paper describes algorithms for efficiently computing inventory estimates from satellite based images. The algorithms incorporate a one dimensional feature extraction which optimizes the pairwise sum of Fisher distances. Biases are eliminated with a premultiplication by the inverse of the analytically derived error matrix. The technique is demonstrated with a numerical example using statistics obtained from an actual Landsat scene. Attention was given to implementation of the Massively Parallel processor (MPP). A timing analysis demonstrates that the inventory estimation can be performed an order of magnitude faster on the MPP than on a conventional serial machine.
This dataset provides three formats of Road Elevation Model (REM) data: 3D road line/polygon GeoPackage (GPKG), road lidar LAZ and COPC LAZ, and road digital surface model (DSM) GeoTIFF. Data are produced from the ~50TB TxGIO (formerly TNRIS) state lidar collections. This dataset is currently organized by maintenance section in each TxDOT district. Computation is done on GPU computing resources at Oak Ridge National Laboratory (ORNL), through a Strategic Partnership Project with UT Austin and an NSF ACCESS computing allocation award that enables fast massive data movement between TACC Corral and ORNL CADES/OLCF using Globus. In addition to this release from ORNL, a copy of this dataset can also be downloaded at https://web.corral.tacc.utexas.edu/nfiedata/road3d/.
We explore the potential of pulsar timing arrays (PTAs) such as NANOGrav, EPTA, and PPTA to detect the stochastic gravitational wave background in theories of massive gravity. In general relativity, the function describing the dependence of the correlation between the arrival times of signals from two pulsars on the angle between them is known as the Hellings-Downs curve. We compute the analogous overlap reduction function for massive gravity, including the additional polarization states and the correction due to the mass of the graviton, and compare the result with the Hellings-Downs curve. The primary result is a complete analytical form for the analog Hellings-Downs curve, providing a starting point for future numerical studies aimed at a detailed comparison between PTA data and the predictions of massive gravity. We study both the massless limit and the stationary limit as checks on our calculation, and discuss how our formalism also allows us to study the impact of massive spin-2 dark matter candidates on data from PTAs.
Abstract Floods drive dynamic and deeply uncertain risks for people and infrastructures. Uncertainty characterization is a crucial step in improving the predictive understanding of multi‐sector dynamics and the design of risk‐management strategies. Current approaches to estimate flood hazards often sample only a relatively small subset of the known unknowns, for example, the uncertainties surrounding the model parameters. This approach neglects the impacts of key uncertainties on hazards and system dynamics. Here we mainstream a recently developed method for Bayesian inference to calibrate a computationally expensive distributed hydrologic model. We compare three different calibration approaches: (a) stepwise line search, (b) precalibration or screening, and (c) the Fast Model Calibrations (FaMoS) approach. FaMoS deploys a particle‐based approach that takes advantage of the massive parallelization afforded by modern high‐performance computing systems. We quantify how neglecting parametric uncertainty and data discrepancy can drastically underestimate extreme flood events and risks. Precalibration improves prediction skill score over a stepwise line search. The Bayesian calibration improves the uncertainty characterization of model parameters and flood risk projections.
The analysis and design of the Space Shuttle's main engine (SSME) power head involved extended finite difference computations conducted on a supercomputer. Certain problems arise, however, in connection with such a computer, and it is pointed out that the massive output from such analyses can actually inhibit insight into physical behavior. It is, therefore, also important to develop effective techniques for computer-user data transfer which match the computer's computational speed. In order to satisfy such requirements, powerful engineering workstations and three-dimensional color computer graphics codes were employed in the computational fluid dynamics (CFD) analyses of the Shuttle's engine. Consequently, the complex physical characteristics of the SSME's hot-gas manifold could be quickly visualized, interpreted, and compared with alternative designs.
Balanced correlation method and the Maximum Entropy Method (MEM) were implemented to reconstruct a laboratory X-ray source as imaged by a Uniformly Redundant Array (URA) system. Although the MEM method has advantages over the balanced correlation method, it is computationally time consuming because of the iterative nature of its solution. Massively Parallel Processing, with its parallel array structure is ideally suited for such computations. These preliminary results indicate that it is possible to use the MEM method in future coded-aperture experiments with the help of the MPP.
Atomistic Molecular Dynamics (MD) simulations provide researchers the ability to model biomolecular structures such as proteins and their interactions with drug-like small molecules with greater spatiotemporal resolution than is otherwise possible using experimental methods. MD simulations are notoriously expensive computational endeavors that have traditionally required massive investment in specialized hardware to access biologically relevant spatiotemporal scales. Our goal is to summarize the fundamental algorithms that are employed in the literature to then highlight the challenges that have affected accelerator implementations in practice. We consider three broad categories of accelerators: Graphics Processing Units (GPUs), Field-Programmable Gate Arrays (FPGAs), and Application Specific Integrated Circuits (ASICs). These categories are comparatively studied to facilitate discussion of their relative trade-offs and to gain context for the current state of the art. We conclude by providing insights into the potential of emerging hardware platforms and algorithms for MD.
Coulomb collisions in plasmas are typically modeled using the Boltzmann collision operator, or its variants, which apply to weakly magnetized plasmas in which the typical gyroradius of particles significantly exceeds the Debye length. Conversely, O’Neil has developed a kinetic theory to treat plasmas that are so strongly magnetized that the typical gyroradius of particles is much smaller than the distance of the closest approach in a binary collision. Here, we develop a generalized collision operator that applies across the full range of magnetization strength. Since there is no closed-form solution for the scattering cross section when plasma is strongly magnetized, the input to the collision operator is obtained by numerically calculating particle trajectories. To demonstrate novel physics associated with strong magnetization, it is used to compute the friction force on a massive test charge. In addition to the traditional stopping power component, this is found to exhibit a transverse component that is perpendicular to both the velocity and Lorentz force vectors in the strongly magnetized regime, as was predicted recently using linear response theory. As a result, good agreement is found between the collision theory and linear response theory in the regime in which both apply, but the new collision theory also applies to stronger magnetization strength regimes than the linear response theory is expected to apply in.
Deep Neural Networks (DNNs) have become increasingly capable of performing tasks ranging from image recognition to content generation. The training and inference of DNNs heavily rely on GPUs, as GPUs' massively parallel architecture delivers extremely high computing capability. With the growing complexity of DNNs and the size of training datasets, training DNNs with a large number of GPUs is becoming a prevalent strategy. Researchers have been exploring how to design software and hardware systems for GPU farms to achieve the best utilization, efficiency, and DNN accuracy during training or inference. However, when designing and deploying such systems, designers usually rely on testing on physical hardware platforms equipped with many GPUs, incurring high costs that are almost prohibitive for system designers to test different configurations and designs, even for highly resourceful companies. While an alternative solution is to test on GPU simulators, they are often too slow for these l
The development of reduced models provide efficient methods that can be used to perform short term experimental data analysis or narrow down the parametric range of more sophisticated numerical approaches. Reduced models are derived by simplifying the physics description with the goal of retaining only the essential ingredients required to reproduce the phenomena under study. This is the role of the gyro-fluid code FAR3d, dedicated to analyze the linear and nonlinear stability of Alfvén Eigenmodes (AE), Energetic Particle Modes (EPM) and magnetic-hydrodynamic modes as pressure gradient driven mode (PGDM) and current driven modes (CDM) in nuclear fusion devices. Such analysis is valuable for improving the plasma heating efficiency and confinement; this can enhance the overall device performance. The present review is dedicated to a description of the most important contributions of the FAR3d code in the field of energetic particles (EP) and AE/EPM stability. FAR3d is used to model and characterize the AE/EPM activity measured in fusion devices as LHD, JET, DIII-D, EAST, TJ-II and Heliotron J. In addition, the computational efficiency of FAR3d facilitates performing massive parametric studies leading to the identification of optimization trends with respect to the AE/EPM stability. This can aid in identifying operational regimes where AE/EPM activity is avoided or minimized. This technique is applied to the analysis of optimized configurations with respect to the thermal plasma parameters, magnetic field configuration, external actuators and the effect of multiple EP populations. In addition, the AE/EPM saturation phase is analyzed, taking into account both steady-state phases and bursting activity observed in LHD and DIII-D devices. The nonlinear calculations provide: the induced EP transport, the generation of zonal structures as well as the energy transfer towards the thermal plasma and between different toroidal/helical families. Finally, FAR3d is used to forecast the AE/EPM stability in operational scenarios of future devices as ITER, CFETR, JT60SA and CFQS as well as possible approaches to optimization with respect to variations in the most important plasma parameters.
An analysis for viscous radiating shock layer flow over a blunt body with massive blowing is presented. Detailed equilibrium computations for the thermodynamics and transport properties are included in the solution. Molecular band and atomic line, as well as continuum radiation are included. For a typical Jovian entry problem, a 40% reduction in total radiative flux at stagnation point, due to the injection of carbon gas, was found. The reduction increases to about 70% at about four body nose radii downstream. The increase of flux reduction indicates that most of the radiative energy is absorbed by the injected carbon gas and dumped into the wake. An examination of the spectral distribution of the surface radiative flux reveals that most of the flux reduction occurs in molecular bands of C2.