Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Distributed computation”

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 1,567 records · Page 87

Load‐balanced and locality‐aware scheduling for data‐intensive workloads at extreme scales

Summary Data‐driven programming models such as many‐task computing (MTC) have been prevalent for running data‐intensive scientific applications. MTC applies over‐decomposition to enable distributed scheduling. To achieve extreme scalability, MTC proposes a fully distributed task scheduling architecture that employs as many schedulers as the compute nodes to make scheduling decisions. Achieving distributed load balancing and best exploiting data locality are two important goals for the best performance of distributed scheduling of data‐intensive applications. Our previous research proposed a data‐aware work‐stealing technique to optimize both load balancing and data locality by using both dedicated and shared task ready queues in each scheduler. Tasks were organized in queues based on the input data size and location. Distributed key‐value store was applied to manage task metadata. We implemented the technique in MATRIX, a distributed MTC task execution framework. In this work, we devise an analytical suboptimal upper bound of the proposed technique, compare MATRIX with other scheduling systems, and explore the scalability of the technique at extreme scales. Results show that the technique is not only scalable but can achieve performance within 15% of the suboptimal solution. Copyright © 2015 John Wiley & Sons, Ltd.

Wang, Ke↗

Investigation of the Window Variance Noise Component of Multicomponent Signals

Signal partitioning or signal segmentation allows to perform data classification, prediction of signals’ behavior, and profound interpretation of obtained data. In accordance with the signal model some distribution characteristics of the window variance noise component are investigated. It was shown that when all true underlying signal components remain unchanged, the window variance noise component is gamma distributed. Applying window variance to multicomponent signal (i.e., pharmacokinetic curve), it was shown that the window variance allows to split visually the signal record into phases due to prevalent processes.

97 MATHEMATICS AND COMPUTING↗

Resilience of the Electric Grid Through Trustable IoT-Coordinated Assets

The electricity grid has evolved from a physical system to a cyberphysical system with digital devices that perform measurement, control, communication, computation, and actuation. The increased penetration of distributed energy resources (DERs) including renewable generation, flexible loads, and storage provides extraordinary opportunities for improvements in efficiency and sustainability. However, they can introduce new vulnerabilities in the form of cyberattacks, which can cause significant challenges in ensuring grid resilience. We propose a framework in this paper for achieving grid resilience through suitably coordinated assets including a network of Internet of Things devices. A local electricity market is proposed to identify trustable assets and carry out this coordination. Situational Awareness (SA) of locally available DERs with the ability to inject power or reduce consumption is enabled by the market, together with a monitoring procedure for their trustability and commitment. With this SA, we show that a variety of cyberattacks can be mitigated using local trustable resources without stressing the bulk grid. Multiple demonstrations are carried out using a high-fidelity cosimulation platform, real-time hardware-in-the-loop validation, and a utility-friendly simulator.

distributed energy resources↗

EQSIM—A multidisciplinary framework for fault-to-structure earthquake simulations on exascale computers, part II: Regional simulations of building response

The existing observational database of the regional-scale distribution of strong ground motions and measured building response for major earthquakes continues to be quite sparse. As a result, details of the regional variability and spatial distribution of ground motions, and the corresponding distribution of risk to buildings and other infrastructure, are not comprehensively understood. Utilizing high-performance computing platforms, emerging high-resolution, physics-based ground motion simulations can now resolve frequencies of engineering interest and provide detailed synthetic ground motions at high spatial density. This provides an opportunity for new insight into the distribution of infrastructure seismic demands and risk. In the work presented herein, the EQSIM fault-to-structure computational framework described in a companion paper, McCallen et al., is employed to investigate the regional-scale response of buildings to large earthquakes. A representative M = 7.0 strike-slip event is used to explore the distribution and amplitude of building demand, and comparisons are made between building response computed with fault-to-structure simulations and building response computed with existing measured near-fault earthquake records. New information on the distribution and variability of building response from high-performance parallel simulations is described and analyzed, and favorable first comparisons between building response predicted with both fault-to-structure simulations and real ground motions records are presented.

58 GEOSCIENCES↗

Primal-Dual Differentiable Programming for Distribution System Critical Load Restoration: Preprint

Swift and reliable critical load restoration (CLR) can help make a distribution system resilient towards extreme events. To optimally achieve that, alongside practical concerns such as limiting online computational burden, some studies leverage model-free reinforcement learning (RL) to train control policies. Despite the advantages provided by RL algorithms, these approaches suffer from two issues: 1) the lack of a proper mechanism for constraint enforcement, and 2) poor sample efficiency. Therefore, in this paper, a primal-dual differentiable programming (PDDP) method is developed for guiding the training leading to a constraint-satisfying policy. Additionally, the model-based nature of the proposed method aims at improving sample efficiency. The experiment on a CLR problem demonstrates that PDDP can effectively train a control policy that both achieves desirable performance and satisfies required constraints.

differentiable programming↗

Max-Compound Cox Processes. I

Extreme values are considered in samples with random size that have a mixed Poisson distribution that is generated by a doubly stochastic Poisson process. Some inequalities are proved relating the distributions and moments of extrema with those of the leading process (the mixing distribution). Limit theorems are proved for the distributions of max-compound Cox processes, and limit distributions are described. An important particular case of the negative binomial distribution of a sample size corresponding to the case where the Cox process is led by a gamma Lévy process is considered, explaining a possible genesis of tempered asymptotic models.

97 MATHEMATICS AND COMPUTING↗

Designing a parallel Feel-the-Way clustering algorithm on HPC systems

This paper introduces a new parallel clustering algorithm, named Feel-the-Way clustering algorithm, that provides better or equivalent convergence rate than the traditional clustering methods by optimizing the synchronization and communication costs. Our algorithm design centers on how to optimize three factors simultaneously: reduced synchronizations, improved convergence rate, and retained same or comparable optimization cost. To compare the optimization cost, we use the Sum of Square Error (SSE) cost as the metric, which is the sum of the square distance between each data point and its assigned clusters. Compared with the traditional MPI k-means algorithm, the new Feel-the-Way algorithm requires less communications among participating processes. As for the convergence rate, the new algorithm requires fewer number of iterations to converge. As for the optimization cost, it obtains the SSE costs that are close to the k-means algorithm. In the paper, we first design the full-step Feel-the-Way k-means clustering algorithm that can significantly reduce the number of iterations that are required by the original k-means clustering method. Next, we improve the performance of the full-step algorithm by adopting an optimized sampling-based approach, named reassignment-history-aware sampling. Our experimental results show that the optimized sampling-based Feel-the-Way method is significantly faster than the widely used k-means clustering method, and can provide comparable optimization costs. More extensive experiments with several synthetic datasets and real-world datasets (e.g., MNIST, CIFAR-10, ENRON, and PLACES-2) show that the new parallel algorithm can outperform the open source MPI k-means library by up to 110% on a high-performance computing system using 4,096 CPU cores. In addition, the new algorithm can take up to 51% fewer iterations to converge than the k-means clustering algorithm.

97 MATHEMATICS AND COMPUTING↗

GADRAS-DRF Version 18 User's Manual

The Gamma Detector Response and Analysis Software Detector Response Function (GADRAS-DRF) application computes the response of gamma-ray and neutron detectors to incoming radiation and provides analysis on measured spectra. This manual provides step-by-step procedures to acquaint new users with the use of the application. The capabilities include characterization of detector response parameters, plotting and viewing measured and computed spectra, analyzing spectra to identify isotopes, estimating source energy distributions from measured spectra, and creating inject data. GADRAS-DRF can compute and provide detector responses quickly and accurately, giving users the ability to obtain usable results in a timely manner (a matter of seconds or minutes).

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

A Stochastic Intracellular Model of Anthrax Infection With Spore Germination Heterogeneity

We present a stochastic mathematical model of the intracellular infection dynamics of Bacillus anthracis in macrophages. Following inhalation of B. anthracis spores, these are ingested by alveolar phagocytes. Ingested spores then begin to germinate and divide intracellularly. This can lead to the eventual death of the host cell and the extracellular release of bacterial progeny. Some macrophages successfully eliminate the intracellular bacteria and will recover. Here, a stochastic birth-and-death process with catastrophe is proposed, which includes the mechanism of spore germination and maturation of B. anthracis . The resulting model is used to explore the potential for heterogeneity in the spore germination rate, with the consideration of two extreme cases for the rate distribution: continuous Gaussian and discrete Bernoulli. We make use of approximate Bayesian computation to calibrate our model using experimental measurements from in vitro infection of murine peritoneal macrophages with spores of the Sterne 34F2 strain of B. anthracis . The calibrated stochastic model allows us to compute the probability of rupture, mean time to rupture, and rupture size distribution, of a macrophage that has been infected with one spore. We also obtain the mean spore and bacterial loads over time for a population of cells, each assumed to be initially infected with a single spore. Our results support the existence of significant heterogeneity in the germination rate, with a subset of spores expected to germinate much later than the majority. Furthermore, in agreement with experimental evidence, our results suggest that most of the spores taken up by macrophages are likely to be eliminated by the host cell, but a few germinated spores may survive phagocytosis and lead to the death of the infected cell. Finally, we discuss how this stochastic modelling approach, together with dose-response data, allows us to quantify and predict individual infection risk following exposure.

59 BASIC BIOLOGICAL SCIENCES↗

A probabilistic graphical model approach to stochastic multiscale partial differential equations

We develop a probabilistic graphical model based methodology to efficiently perform uncertainty quantification in the presence of both stochastic input and multiple scales. Both the stochastic input and model responses are treated as random variables in this framework. Their relationships are modeled by graphical models which give explicit factorization of a high-dimensional joint probability distribution. The hyperparameters in the probabilistic model are learned using sequential Monte Carlo (SMC) method, which is superior to standard Markov chain Monte Carlo (MCMC) methods for multi-modal distributions. Finally, we make predictions from the probabilistic graphical model using the belief propagation algorithm. Numerical examples are presented to show the accuracy and efficiency of the predictive capability of the developed graphical model.

97 MATHEMATICS AND COMPUTING↗

Gray and multigroup radiation transport through 3D binary stochastic media with different sphere radii distributions

Here, gray and multigroup radiation is transported through 3D media consisting of spheres randomly placed in a uniform background. Comparisons are made between using constant radii spheres and three different distributions of sphere radii. Because of the computational cost of 3D calculations, only the lowest angle order, n=1, is tested. If the mean chord length is held constant, using different radii distributions makes little difference. This is true for both gray and multigroup solutions. 3D transport solutions are compared to 2D and 1D solutions with the same mean chord lengths. 2D disk and 3D sphere media give solutions that are nearly identical while 1D slab solutions are fundamentally different.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Structure of the Pion and Kaon from First-Principles Calculation in Lattice QCD

I outline recent developments in lattice QCD that enable the computation of the x-dependent Parton Distribution Functions, and then present recent calculations of the valence PDF of the pion by the HadStruct collaboration. I describe the challenges and opportunities to compute the gluon and sea-quark contributions to meson structure, and how such lattice calculations can form a key part of the physics goals of the EIC.

Richards, David↗

Light-Front Holographic QCD

The relation between the hadronic short-distance constituent quark and gluon particle limit and the long-range confining domain is yet one of the most challenging aspects of particle physics due to the strong coupling nature of Quantum Chromodynamics, the fundamental theory of the strong interactions. The central question is how one can compute hadronic properties from first principles; i.e., directly from the QCD Lagrangian. The most successful theoretical approach thus far has been to quantize QCD on discrete lattices in Euclidean space-time. Lattice numerical results follow from computation of frame-dependent moments of distributions in Euclidean space and dynamical observables in Minkowski spacetime, such as the time-like hadronic form factors, are not amenable to Euclidean lattice computations. The Dyson-Schwinger methods have led to many important insights, such as the infrared fixed point behavior of the strong coupling constant, but in practice, the analyses are limited to ladder approximation in Landau gauge. Baryon spectroscopy and the excitation dynamics of nucleon resonances encoded in the nucleon transition form factors can provide fundamental insight into the strong-coupling dynamics of QCD. New theoretical tools are thus of primary interest for the interpretation of the results expected at the new mass scale and kinematic regions accessible to the JLab 12 GeV Upgrade Project. The AdS/CFT correspondence between gravity or string theory on a higher-dimensional anti-de Sitter (AdS) space and conformal field theories in physical space-time has led to a semiclassical approximation for strongly-coupled QCD, which provides physical insights into its nonperturbative dynamics. The correspondence is holographic in the sense that it determines a duality between theories in different number of space-time dimensions. This geometric approach leads in fact to a simple analytical and phenomenologically compelling nonperturbative approximation to the full light-front QCD Hamiltonian 'Light-Front Holography'. Light-Front Holography is in fact one of the most remarkable features of the AdS/CFT correspondence. The Hamiltonian equation of motion in the light-front (LF) is frame independent and has a structure similar to eigenmode equations in AdS space. This makes a direct connection of QCD with AdS/CFT methods possible. Remarkably, the AdS equations correspond to the kinetic energy terms of the partons inside a hadron, whereas the interaction terms build confinement and correspond to the truncation of AdS space in an effective dual gravity approximation. One can also study the gauge/gravity duality starting from the bound-state structure of hadrons in QCD quantized in the light-front. The LF Lorentz-invariant Hamiltonian equation for the relativistic bound-state system is P{sub {mu}}P{sup {mu}}|{psi}(P)> = (P{sup +}P{sup -} - P{sub {perpendicular}}{sup 2})|{psi}(P)> = M{sup 2}|{psi}(P)>, P{sup {+-}} = P{sup 0} {+-} P{sup 3}, where the LF time evolution operator P{sup -} is determined canonically from the QCD Lagrangian. To a first semiclassical approximation, where quantum loops and quark masses are not included, this leads to a LF Hamiltonian equation which describes the bound-state dynamics of light hadrons in terms of an invariant impact variable {zeta} which measures the separation of the partons within the hadron at equal light-front time {tau} = x{sup 0} + x{sup 3}. This allows us to identify the holographic variable z in AdS space with an impact variable {zeta}. The resulting Lorentz-invariant Schroedinger equation for general spin incorporates color confinement and is systematically improvable. Light-front holographic methods were originally introduced by matching the electromagnetic current matrix elements in AdS space with the corresponding expression using LF theory in physical space time. It was also shown that one obtains identical holographic mapping using the matrix elements of the energy-momentum tensor by perturbing the AdS metric around its static solution. A gravity dual to QCD is not known, but the mechanisms of confinement can be incorporated in the gauge/gravity correspondence by modifying the AdS geometry in the large infrared (IR) domain z {approx} 1 = {Lambda}{sub QCD}, which also sets the scale of the strong interactions. In this simplified approach we consider the propagation of hadronic modes in a fixed effective gravitational background asymptotic to AdS space, which encodes salient properties of the QCD dual theory, such as the ultraviolet (UV) conformal limit at the AdS boundary, as well as modifications of the background geometry in the large z IR region to describe confinement. The modified theory generates the point-like hard behavior expected from QCD, instead of the soft behavior characteristic of extended objects.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Per-page control of physical address space distribution among memory modules

Systems, apparatuses, and methods for implementing per-page control of physical address space distribution among memory modules are disclosed. A computing system includes a plurality of processing units coupled to a plurality of memory modules. A determination is made as to which physical address space distribution granularity to implement for physical memory pages allocated for a first data structure. The determination can be made on a per-data-structure basis (e.g., file, page, block, etc.) or on a per-application-basis. A physical address space distribution granularity is encoded as a property of each physical memory page allocated for the first data structure, and physical memory pages of the first data structure distributed across the plurality of memory modules based on a selected physical address space distribution granularity. Page table entries (PTEs) may be annotated with the selected physical address space distribution granularity, using an addressing mapping granularity (AMG) field of a page table entry, where the granularity may be, for example, a fine-grain distribution granularity or a coarse-grain distribution granularity.

Jayasena, Nuwan S.↗

Stochastic Constrained Extended System Dynamics for Solving Charge Equilibration Models

In this work, we present a new stochastic extended Lagrangian molecular dynamics solution to charge equilibration that eliminates self-consistent field (SCF) calculations, thus eliminating the computational bottleneck in solving the charge distribution with standard SCF solvers. By formulating both charges and chemical potential as latent variables and introducing a holonomic constraint that satisfies charge conservation, the SC-XLMD method accurately reproduces thermodynamic, dynamic, and structural properties within the framework of ReaxFF for a bulk water system and highly reactive RDX molecules simulated at high temperature. The SC-XLMD method shows excellent computational performance and is available in the publicly available LAMMPS package.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Methods and systems for evaluating and improving distribution-grid observability

This document describes systems and techniques for evaluating and improving distribution-grid observability. These systems and techniques allow engineers to quantify the observability of a distribution grid, which represents an ability to combine actual measurements and various types of computations (e.g., analytics, estimators, forecasters), from a system model. Distribution engineers can also identify islands of observability where operating parameters, including voltages, currents, and power flows, can be determined from available sensor readings. By exclusion, distribution engineers can similarly identify areas of the distribution grid with observability deficiencies that may require additional instrumentation to maintain proper operation. Distribution engineers, using an iterative or automated process, can determine the observability of the system model with new or relocated sensors to generate a sensor allocation plan. The sensor allocation plan can indicate the number and location of sensors to either maximize observability for a fixed sensor cost or minimize sensor cost for predetermined observability.

Reiman, Anthony P.↗

Methods and systems for evaluating data transportability in distribution grids

This document describes systems and techniques for evaluating and improving distribution-grid data transportability. These systems and techniques allow engineers to quantify the data transportability of a communication system within or connected to a distribution grid, which represents an ability to transport in real-time telemetry from source locations (e.g., sensors in the distribution grid) to control mechanisms. Distribution engineers can use the sensor readings to perform grid analytics, control operating parameters, and operate protection systems. Distribution engineers can also use the transportability of the communication system to evaluate the observability of the distribution grid, which represents an ability to combine actual measurements and various types of computations (e.g., analytics, estimators, forecasters) from a system model. Distribution engineers can then generate a sensor allocation plan that indicates the number and location of sensors to maximize observability for a fixed sensor cost and/or minimize sensor cost for predetermined observability.

Taft, Jeffrey D.↗

Electron and photon structure functions at two loops

We present a fully analytic computation of the complete electron and photon structure functions, or QED lepton parton distribution functions (PDFs) up to two-loop order. Our computation is performed using modern techniques of reduction to Master Integrals and solving them with the differential equation method. We obtain explicit expressions for the electron-in-electron, positron-in-electron, photon-in-electron, electron-in-photon, and photon-in-photon distributions at next-to-next-to-leading order (NNLO). Cross-checks against one-loop results, existing two-loop calculations, and a recent soft-collinear effective theory (SCET) analysis of the electron structure functions are presented.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗