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 1,333 records · Page 74
All sourcesNASA NTRSDOE OSTI
DOE OSTI2012

Moment preserving adaptive particle weights using octree velocity distributions for PIC simulations

The ratio of computational to physical particles is of primary concern to statistical particle based simulations such as DSMC and PIC. An adaptive computational particle weight algorithm is presented that conserves mass, momentum, and energy. This algorithm is then enhanced with an octree adaptive mesh in velocity space to mitigate artificial thermalization. The new octree merge is compared to a merge that randomly selects merge partners for a bi-Maxwellian velocity distribution. Results for crossing beams in a fixed potential well along with an electrostatic PIC version with and without MCC collisions based ionizing breakdown show the advantages of the merge algorithm to both fixed particle weights and randomly selected merge partners.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2025

Distributed Stochastic Optimization of a Neural Representation Network for Time-Space Tomography Reconstruction

4D time-space reconstruction of dynamic events or deforming objects using X-ray computed tomography (CT) is an important inverse problem in non-destructive evaluation. Conventional back-projection based reconstruction methods assume that the object remains static for the duration of several tens or hundreds of X-ray projection measurement images (reconstruction of consecutive limited-angle CT scans). However, this is an unrealistic assumption for many in-situ experiments that causes spurious artifacts and inaccurate morphological reconstructions of the object. To solve this problem, we propose to perform a 4D time-space reconstruction using a distributed implicit neural representation (DINR) network that is trained using a novel distributed stochastic training algorithm. Our DINR network learns to reconstruct the object at its output by iterative optimization of its network parameters such that the measured projection images best match the output of the CT forward measurement model. Here, we use a forward measurement model that is a function of the DINR outputs at a sparsely sampled set of continuous valued 4D object coordinates. Unlike previous neural representation architectures that forward and back propagate through dense voxel grids that sample the object's entire time-space coordinates, we only propagate through the DINR at a small subset of object coordinates in each iteration resulting in an order-of-magnitude reduction in memory and compute for training. DINR leverages distributed computation across several compute nodes and GPUs to produce high-fidelity 4D time-space reconstructions. We use both simulated parallel-beam and experimental cone-beam X-ray CT datasets to demonstrate the superior performance of our approach.

36 MATERIALS SCIENCE↗
DOE OSTI2024

A potential model study of the nucleon's charge and mass radius

We study the charge and mass distributions within a nucleon and compute the associated squared radii based on a potential model approach. Different constituent quark configurations such as Δ, Y, and quark-diquark are considered and compared, with model parameters calibrated by experimental measurements of the proton and neutron charge radius. The results suggest that while the charge radius is dictated by quark dynamics, the mass radius is strongly influenced by nonperturbative QCD contributions to a nucleon's mass that are not sensitive to the constituent quarks. As a result, the mass radius could become substantially different from the charge radius. The obtained nucleon mass distributions of different configurations are further used for simulations of the initial conditions in heavy ion collisions. Furthermore, the computed eccentricities ε 2 and ε 3 are found to demonstrate a considerable sensitivity to the input nucleon profiles, especially to the mass radius in the peripheral region as well as for systems with fewer participants.

Gallimore, Daniel [Indiana University, Bloomington↗
DOE OSTI2015

SU-E-T-540: MCNPX Simulation of Proton Dose Distributions in a Water Phantom

Purpose: In this study, fluence and energy deposition of proton and proton by-products and dose distributions were simulated. Lateral dose distributions were also been discussed to understand the difference between Monte Carlo simulations and pencil beam algorithm. Methods: MCNPX codes were used to build a water phantom by using “repeated structures” technique and the doses and fluences in each cell was recorded by mesh tally. This study includes, proton equilibrium and proton disequilibrium case. For the proton equilibrium case, the doses difference between proton and proton by-products were studied. A 160 MeV proton pencil beam was perpendicularly incident into a 40 × 40 × 50 cm{sup 3} water phantom and the scoring volume was 20 × 20 × 0.2 cm{sup 3}. Energy deposition and fluence were calculated from MCNPX with (1) proton only; and (2) proton and secondary particles. For the proton disequilibrium case, the dose distribution variation using different multiple Coulomb scattering were studied. A 70 MeV proton pencil beam was perpendicularly incident into a 40 × 40 × 10 cm{sup 3} water phantom and two scoring voxel sizes of 0.1 × 0.1 × 0.05 cm{sup 3} and 0.01 × 0.01 × 0.05 cm{sup 3} were used for the depth dose distribution, and 0.01 × 0.01 × 0.05 cm{sup 3} for the lateral profile distribution simulations. Results: In the water phantom, proton fluence and dose in depths beyond the Bragg peak were slightly perturbed by the choice of the simulated particle types. The dose from secondary particles was about three orders smaller, but its simulation consumed significant computing time. The depth dose distributions and lateral dose distributions of 70 MeV proton pencil beam obtained from MCNPX, GEANT4, and the pencil beam algorithm showed the significant deviations, probably caused by multiple Coulomb scattering. Conclusion: Multiple Coulomb scattering is critical when there is in proton disequilibrium.

07 ISOTOPE AND RADIATION SOURCES↗
DOE OSTI2020

Intersection-distribution-based remapping between arbitrary meshes for staggered multi-material arbitrary Lagrangian-Eulerian hydrodynamics

In this paper, we present a new intersection-distribution-based remapping method between arbitrary polygonal meshes for indirect staggered multi-material arbitrary Lagrangian-Eulerian hydrodynamics. All cell-centered material quantities are conservatively remapped using intersections between the Lagrangian (old, source) mesh and the rezoned (new, target) mesh. The new nodal masses are obtained by conservative distribution of all material masses in each new cell to the cell's corners and then collecting those corner masses at new nodes. This distribution is done using a local constrained optimization approach for each cell in the new mesh. In order to remap nodal momentum we first define cell-centered momentum for each cell in the old mesh, conservatively remap this to the new mesh and then conservatively distribute the new zonal momentum to each cell's bounding nodes, again using local constrained optimization. Our method also conserves total energy by applying a new nodal kinetic energy correction that relies on a process similar to that used for remapping nodal mass and momentum. Cell-centered kinetic energy is computed, conservatively remapped and then distributed to nodes. The discrepancy between this conservatively remapped and actual nodal kinetic energy is then conservatively distributed to the internal energies of the materials in the cells surrounding each node. Unlike conventional cell-based corrections of this type, this new nodal kinetic energy correction has not been observed to drive material internal energy negative in any of our testing. Unlike flux based remapping, our new intersection-distribution method can be applied to remapping between source and target meshes that are arbitrarily different, which provides superior flexibility in the rezoning strategy. Our method is accurate, essentially conservative and essentially bounds preserving.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2025

Simultaneous inference of equation of state parameters and unknown data errors with uncertainty quantification via hierarchical Bayesian posterior maximization

Equations of state (EOSs) are a key component in running hydrodynamic simulations as they relate the thermodynamic states for the material. The Davis reactants EOS is commonly used for modeling high explosives (HEs), and the EOS model parameters are calibrated using material specific data. The calibrations are often performed with uncertainty quantification via Bayesian inference to account for uncertainty in the data and generate ensembles of likely parameters. However, there are relatively few HE data sets to use for calibration and many are historical and lack error information. In this work, we simultaneously calibrate the Davis reactants EOS model parameters and unknown data error terms for the high explosive PBX 9501. To quantify the uncertainty in the models and the data, we use a Bayesian framework for the calibration and compute the hierarchical Bayesian posterior distribution with both a posteriori maximization approach and Markov Chain Monte Carlo. In general, we find that, given our assumptions, the two approaches result in similar calibrated parameters, posterior covariance matrices, and insights about the parameters but that the posterior maximization requires far less computational resources.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2015

The OSG open facility: A sharing ecosystem

The Open Science Grid (OSG) ties together individual experiments’ computing power, connecting their resources to create a large, robust computing grid, this computing infrastructure started primarily as a collection of sites associated with large HEP experiments such as ATLAS, CDF, CMS, and DZero. In the years since, the OSG has broadened its focus to also address the needs of other US researchers and increased delivery of Distributed High Through-put Computing (DHTC) to users from a wide variety of disciplines via the OSG Open Facility. Presently, the Open Facility delivers about 100 million computing wall hours per year to researchers who are not already associated with the owners of the computing sites, this is primarily accomplished by harvesting and organizing the temporarily unused capacity (i.e. opportunistic cycles) from the sites in the OSG. Using these methods, OSG resource providers and scientists share computing hours with researchers in many other fields to enable their science, striving to make sure that these computing power used with maximal efficiency. Furthermore, we believe that expanded access to DHTC is an essential tool for scientific innovation and work continues in expanding this service.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Stress due to electric charge density distribution in a dielectric slab

The spatial distribution of electric field due to an imposed electric charge density profile in an infinite slab of dielectric material is derived analytically by integrating Gauss’s law. Various charge density distributions are considered, including exponential and power-law forms. Here, the Maxwell stress tensor is used to compute a notional static stress in the material due to the charge density and its electric field. Characteristics of the electric field and stress distributions are computed for example cases in polyethylene, showing that field magnitudes exceeding the dielectric strength would be required in order to achieve a stress exceeding the ultimate tensile strength.

42 ENGINEERING↗
DOE OSTI2024

Gradient Coding With Iterative Block Leverage Score Sampling

Gradient coding is a method for mitigating straggling servers in a centralized computing network that uses erasure-coding techniques to distributively carry out first-order optimization methods. Randomized numerical linear algebra uses randomization to develop improved algorithms for large-scale linear algebra computations. In this study, we propose a method for distributed optimization that combines gradient coding and randomized numerical linear algebra. The proposed method uses a randomized ℓ 2 -subspace embedding and a gradient coding technique to distribute blocks of data to the computational nodes of a centralized network, and at each iteration the central server only requires a small number of computations to obtain the steepest descent update. The novelty of our approach is that the data is replicated according to importance scores, called block leverage scores, in contrast to most gradient coding approaches that uniformly replicate the data blocks. Furthermore, we do not require a decoding step at each iteration, avoiding a bottleneck in previous gradient coding schemes. We show that our approach results in a valid ℓ 2 -subspace embedding, and that our resulting approximation converges to the optimal solution.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Performance Evaluation of Python Based Data Analytics Frameworks in Summit: Early Experiences

The explosion in the volumes of data generated from ever-larger simulation campaigns and experiments or observations necessitates competent tools for data wrangling and analysis). While the Oak Ridge Leadership Computing Facility (OLCF) provides a variety of tools to perform data wrangling and data analysis tasks, Python based tools often lack scalability, or the ability to fully exploit the computational capability of OLCF’s Summit supercomputer. NVIDIA RAPIDS and Dask offer a promising solution to accelerate and distribute data analytics workloads from personal computers to heterogeneous supercomputing systems. We discuss early performance evaluation results of RAPIDS and Dask on Summit to understand their capabilities, scalability, and limitations. Our evaluation includes a subset of RAPIDS libraries, i.e., cuDF, cuML, and cuGraph, and Chainer’s CuPy, and their multi-GPU variants when available.We also draw on the observed trends from the performance evaluation results to discuss best practices for maximizing performance.

Hernandez Arreguin, Benjamin↗
DOE OSTI2023

Evaluating generative networks using Gaussian mixtures of image features

We develop a measure for evaluating the performance of generative networks given two sets of images. A popular performance measure currently used to do this is the Fréchet Inception Distance (FID). However, FID assumes that images featurized using the penultimate layer of Inception follow a Gaussian distribution. This assumption allows FID to be easily computed, since FID uses the 2-Wasserstein distance of two Gaussian distributions fitted to the featurized images. However, we show that Inception features of the ImageNet dataset are not Gaussian; in particular, each marginal is not Gaussian. To remedy this problem, we model the featurized images using Gaussian mixture models (GMMs) and compute the 2-Wasserstein distance restricted to GMMs. We define a performance measure, which we call WaM, on two sets of images by using inception (or another classifier) to featurize the images, estimate two GMMs, and use the restricted 2-Wasserstein distance to compare the GMMs. We experimentally show the advantages of WaM over FID, including how FID is more sensitive than WaM to image perturbations. By modelling the non-Gaussian features obtained from inception as GMMs and using a GMM metric, we can more accurately evaluate generative network performance.

machine learning, genrative adversarial networks↗
DOE OSTI2017

3D Magnetohydrodynamic Models of Nonthermal Photon Emission in the Binary System γ{sup 2} Velorum

Recent reports claiming an association of the massive star binary system γ{sup 2} Velorum (WR 11) with a high-energy γ-ray source observed by Fermi-LAT contrast the so far exclusive role of η Carinae as the hitherto only detected γ-ray emitter in the source class of particle-accelerating colliding-wind binary (CWB) systems. We offer support to this claim of association by providing dedicated model predictions for the nonthermal photon emission spectrum of γ{sup 2} Velorum. We use 3D magnetohydrodynamic modeling (MHD) to investigate the structure and conditions of the wind-collision region (WCR) of γ{sup 2} Velorum including the important effect of radiative braking in the stellar winds. A transport equation is then solved for the entire computational domain to study the propagation of relativistic electrons and protons. The resulting distributions of particles are subsequently used to compute nonthermal photon emission components. In agreement with observation in X-ray spectroscopy, our simulations yield a large shock-cone opening angle. We find the nonthermal γ-ray emission of γ{sup 2} Velorum to be of hadronic origin owing to the strong radiation fields in the binary system, which inhibit the acceleration of electrons to energies sufficiently high for efficient inverse-Compton radiation. We also discuss the strong dependence of a hadronic γ-ray component on the energy-dependent diffusion used in the simulations. Of two mass-loss rates for the WR star found in literature, only the higher rate is able to accommodate the observed γ-ray spectrum with reasonable values for important simulation parameters such as the injection ratio of high-energy particles within the WCR.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2018

Layered nested Markov chain Monte Carlo

A configurational sampling algorithm based on nested layerings of Markov chains (Layered Nested Markov Chain Monte Carlo or L-NMCMC) is presented for simulations of systems characterized by rugged free energy landscapes. The layerings are generated using a set of auxiliary potential energy surfaces. The implementation of the method is demonstrated in the context of a rugged, two-dimensional potential energy surface. The versatility of the algorithm is next demonstrated on a simple, many body system, namely, a canonical Lennard-Jones fluid in the liquid state. In that example, different layering schemes and auxiliary potentials are used, including variable cutoff distances and excluded volume tempering. In addition to calculating a variety of properties of the system, it is also shown that L-NMCMC, when combined with a free-energy perturbation formalism, provides a straightforward means to construct approximate free-energy surfaces at no additional computational cost using the sampling distributions of each auxiliary Markov chain. In conclusion, the proposed L-NMCMC scheme is general in that it could be complementary to any number of methods that rely on sampling from a target distribution or methods that exploit a hierarchy of time scales and/or length scales through decomposition of the potential energy.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2014

Developing a System for Testing Computational Social Models using Amazon Mechanical Turk

The US faces persistent, distributed threats from malevolent individuals, groups and organizations around the world. Computational Social Models (CSMs) help anticipate the dynamics and behaviors of these actors by modeling the behavior and interactions of individuals, groups and organizations. For strategic planners to trust the results of CSMs, they must have confidence in the validity of the models. Establishing validity before model use will enhance confidence and reduce the risk of error. One problem with validation is designing an appropriate controlled test of the model, similar to the testing of physical models. Lab experiments can do this, but are often limited to small numbers of subjects, with low subject diversity and are often in a contrived environment. Natural studies attempt to test models by gathering large-scale observational data (e.g., social media) however this loses the controlled aspect. We propose a new approach to run large-scale, controlled online experiments on diverse populations. Using Amazon Mechanical Turk, a crowdsourcing tool, we will draw large populations into controlled experiments in a manner that was not possible just a few years ago.

42 ENGINEERING↗
DOE OSTI2026

Thermodynamically consistent incorporation of the Langmuir adsorption model into compressible fluctuating hydrodynamics

For a gas–solid interfacial system where chemical species undergo reversible adsorption, we develop a mesoscopic stochastic modeling method that simulates both gas-phase hydrodynamics and surface coverage dynamics by coupling the Langmuir adsorption model with compressible fluctuating hydrodynamics. To this end, we derive a thermodynamically consistent mass–energy update scheme that accounts for how the mass and energy variables in the gas and surface subsystems should be updated according to the changes in the number of molecules of each species in each subsystem due to adsorption and desorption events. By performing a stochastic analysis for the ideal Langmuir model and the full hydrodynamic system, we analytically confirm that our mass–energy update scheme captures thermodynamic equilibrium predicted by equilibrium statistical mechanics. We find that an internal energy correction term is needed, which is attributed to the difference in the mean kinetic energy of gas molecules colliding with the surface from that computed from the Maxwell–Boltzmann distribution. By performing an equilibrium simulation study for an ideal gas mixture of CO and Ar, with CO undergoing reversible adsorption, we validate our overall simulation method and implementation.

Adsorption↗
DOE OSTI2014

Data–driven model reduction for the Bayesian solution of inverse problems

One of the prime challenges in the Bayesian solution of inverse problems governed by partial differential equations (PDEs) is the computational cost of repeatedly evaluating numerical PDE models, as required by Markov chain Monte Carlo (MCMC) methods for posterior sampling. This paper presents a data-driven projection-based model reduction technique to reduce this computational cost. The proposed technique has two distinctive features. To begin, the model reduction strategy is tailored to inverse problems: the snapshots used to construct the reduced-order model are computed adaptively from the posterior distribution. Posterior exploration and model reduction are thus pursued simultaneously. Second, to avoid repeated evaluations of the full-scale numerical model as in a standard MCMC method, we couple the full-scale model and the reduced-order model together in the MCMC algorithm. This maintains accurate inference while reducing its overall computational cost. In numerical experiments considering steady-state flow in a porous medium, the data-driven reduced-order model achieves better accuracy than a reduced-order model constructed using the classical approach. It also improves posterior sampling efficiency by several orders of magnitude compared with a standard MCMC method.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2016

Distributed Halide

Many image processing tasks are naturally expressed as a pipeline of small computational kernels known as stencils. Halide is a popular domain-specific language and compiler designed to implement image processing algorithms. Halide uses simple language constructs to express what to compute and a separate scheduling co-language for expressing when and where to perform the computation. This approach has demonstrated performance comparable to or better than hand-optimized code. Until now, however, Halide has been restricted to parallel shared memory execution, limiting its performance for memory-bandwidth-bound pipelines or large-scale image processing tasks. We present an extension to Halide to support distributed-memory parallel execution of complex stencil pipelines. These extensions compose with the existing scheduling constructs in Halide, allowing expression of complex computation and communication strategies. Existing Halide applications can be distributed with minimal changes, allowing programmers to explore the tradeoff between recomputation and communication with little effort. Approximately 10 new of lines code are needed even for a 200 line, 99 stage application. On nine image processing benchmarks, our extensions give up to a 1.4× speedup on a single node over regular multithreaded execution with the same number of cores, by mitigating the effects of non-uniform memory access. The distributed benchmarks achieve up to 18× speedup on a 16 node testing machine and up to 57× speedup on 64 nodes of the NERSC Cori supercomputer.

97 MATHEMATICS AND COMPUTING↗