Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “sparse factorization”

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 379 records · Page 21
All sourcesNASA NTRSDOE OSTI
DOE OSTI2016

The sensitivity of core-collapse supernovae to nuclear electron capture

A weak-rate library aimed at investigating the sensitivity of astrophysical environments to variations of electron-capture rates on medium-heavy nuclei has been developed. With this library, the sensitivity of the core-collapse and early post-bounce phases of core-collapse supernovae to nuclear electron capture is examined. The rates are adjusted by factors consistent with uncertainties indicated by comparing theoretical rates to those deduced from charge-exchange and β-decay measurements. To ensure a model-independent assessment, sensitivity studies across a comprehensive set of progenitors and equations of state are performed. We find a +16/−4% range in the mass of the inner core at shock formation and a ±20% range of peak ν{sub e} luminosity during the deleptonization burst. These ranges are five times as large as those seen from a separate progenitor study, where we evaluate the sensitivity of these parameters to 32 presupernova models. Additionally, the simulations are found to be more sensitive to a reduction in electron-capture rates than an enhancement, and specifically to the reduction in rates for neutron-rich nuclei near the N = 50 closed neutron shell. As measurements for medium-heavy (A>65) and neutron-rich nuclei are sparse, and because accurate theoretical models that account for nuclear structure considerations of individual nuclei are not readily available, rates for these nuclei may be overestimated. If more accurate estimates confirm this, results from this study indicate that significant changes to the core-collapse trajectory are expected. For this reason, experimental and theoretical efforts should focus on this region of the nuclear chart.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2022

Two-Stage Gauss-Seidel Preconditioners and Smoothers for Krylov Solvers on a GPU Cluster: Preprint

Gauss-Seidel (GS) relaxation is often employed as a preconditioner for a Krylov solver or as a smoother for Algebraic Multigrid (AMG). However, the requisite sparse triangular solve is difficult to parallelize on many-core architectures such as graphics processing units (GPUs). In the present study, the performance of the sequential GS relaxation based on a triangular solve is compared with two-stage variants, replacing the direct triangular solve with a fixed number of inner Jacobi-Richardson (JR) iterations. When a small number of inner iterations is sufficient to maintain the Krylov convergence rate, the two-stage GS (GS2) often outperforms the sequential algorithm on many-core architectures. The GS2 algorithm is also compared with JR. When they perform the same number of ops for SpMV (e.g. three JR sweeps compared to two GS sweeps with one inner JR sweep), the GS2 iterations, and the Krylov solver preconditioned with GS2, may converge faster than the JR iterations. Moreover, for some problems (e.g. elasticity), it was found that JR may diverge with a damping factor of one, whereas two-stage GS may improve the convergence with more inner iterations. Finally, to study the performance of the two-stage smoother and preconditioner for a practical problem, these were applied to incompressible uid ow simulations on GPUs.

algebraic multigrid↗
DOE OSTI2022

A dictionary learning algorithm for compression and reconstruction of streaming data in preset order

There has been an emerging interest in developing and applying dictionary learning (DL) to process massive datasets in the last decade. Many of these efforts, however, focus on employing DL to compress and extract a set of important features from data, while considering restoring the original data from this set a secondary goal. On the other hand, although several methods are able to process streaming data by updating the dictionary incrementally as new snapshots pass by, most of those algorithms are designed for the setting where the snapshots are randomly drawn from a probability distribution. In this paper, we present a new DL approach to compress and denoise massive dataset in real time, in which the data are streamed through in a preset order (instances are videos and temporal experimental data), so at any time, we can only observe a biased sample set of the whole data. Here, our approach incrementally builds up the dictionary in a relatively simple manner: if the new snapshot is adequately explained by the current dictionary, we perform a sparse coding to find its sparse representation; otherwise, we add the new snapshot to the dictionary, with a Gram-Schmidt process to maintain the orthogonality. To compress and denoise noisy datasets, we apply the denoising to the snapshot directly before sparse coding, which deviates from traditional dictionary learning approach that achieves denoising via sparse coding. Compared to full-batch matrix decomposition methods, where the whole data is kept in memory, and other mini-batch approaches, where unbiased sampling is often assumed, our approach has minimal requirement in data sampling and storage: i) each snapshot is only seen once then discarded, and ii) the snapshots are drawn in a preset order, so can be highly biased. Through experiments on climate simulations and scanning transmission electron microscopy (STEM) data, we demonstrate that the proposed approach performs competitively to those methods in data reconstruction and denoising.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2016

Revisiting the Lyman continuum escape crisis: predictions for z > 6 from local galaxies

The intrinsic escape fraction of ionizing Lyman continuum photons (f{sub esc}) is crucial to understanding whether galaxies are capable of reionizing the neutral hydrogen in the early universe at z > 6. Unfortunately, it is not possible to access f{sub esc} at z > 4 with direct observations, and the handful of measurements from low-redshift galaxies consistently find f{sub esc} < 10%, while at least f{sub esc} ∼ 10% is necessary for galaxies to dominate reionization. Here, we present the first empirical prediction of f{sub esc} at z > 6 by combining the (sparsely populated) relation between [O III]/[O II] and f{sub esc} with the redshift evolution of [O III]/[O II] as predicted from local high-z analogs selected by their Hα equivalent width. We find f{sub esc}=5.7{sub −3.3}{sup +8.3}% at z = 6 and f{sub esc}=10.4{sub −6.3}{sup +15.5}% at z = 9 for galaxies with log(M/M{sub ⊙})∼9.0 (errors given as 1σ). However, there is a negative correlation with stellar mass and we find up to 50% larger f{sub esc} per 0.5 dex decrease in stellar mass. The population-averaged escape fraction increases according to f{sub esc}=f{sub esc,0}((1+z)/3){sup α}, with f {sub esc,0} = (2.3 ± 0.05)% and α = 1.17 ± 0.02 at z > 2 for log(M/M{sub ⊙})∼9.0. With our empirical prediction of f{sub esc} (thus fixing an important, previously unknown variable) and further reasonable assumptions on clumping factor and the production efficiency of Lyman continuum photons, we conclude that the average population of galaxies is just capable of reionizing the universe by z ∼ 6.

79 ASTRONOMY AND ASTROPHYSICS↗
NASA NTRS2011

Isolation Effects on the Moon: High Topographic Slope Observations from the LRO and LOLA Instruments

The extremely low temperatures in the Moon's polar permanent shadow regions (PSR) has long been considered a unique factor necessary for entrapping volatile Hydrogen (H). However, recent discoveries indicate some H concentrations lie outside PSR, suggesting other geophysical factors may also influence H distributions. In this study we consider insolation and its resulting thermal effects as a loss/redistribution process influencing the Moon's near-surface < 1m volatile H budget. To isolate regional (5deg latitude band) insolation effects we correlate two data sets collected from the ongoing, 1.5 year long mapping mission of the Lunar Reconnaissance Orbiter (LRO). Epithermal neutron mapping data from the Lunar Exploration Neutron Detector (LEND) is registered and analyzed in the context of slope derivations from Lunar topography maps produced by the Lunar Observing Laser Altimeter (LOLA). Lunar epithermal neutrons are inferred to be direct geochemical evidence for near-surface H due to the correlated suppression of surface leakage fluxes of epithermal neutrons with increased H concentration. Regional suppressions of neutrons seen in LEND maps are considered localized evidence of H concentration increase in the upper 1 m of the Lunar surface. To quantify spatially localized insolation effects, LEND data are averaged from sparsely distributed pixels, classed as a function of the LOLA slope derivations.

McClanahan, T. P.↗
DOE OSTI2016

Sparse maps—A systematic infrastructure for reduced-scaling electronic structure methods. II. Linear scaling domain based pair natural orbital coupled cluster theory

Domain based local pair natural orbital coupled cluster theory with single-, double-, and perturbative triple excitations (DLPNO-CCSD(T)) is a highly efficient local correlation method. It is known to be accurate and robust and can be used in a black box fashion in order to obtain coupled cluster quality total energies for large molecules with several hundred atoms. While previous implementations showed near linear scaling up to a few hundred atoms, several nonlinear scaling steps limited the applicability of the method for very large systems. In this work, these limitations are overcome and a linear scaling DLPNO-CCSD(T) method for closed shell systems is reported. The new implementation is based on the concept of sparse maps that was introduced in Part I of this series [P. Pinski, C. Riplinger, E. F. Valeev, and F. Neese, J. Chem. Phys. 143, 034108 (2015)]. Using the sparse map infrastructure, all essential computational steps (integral transformation and storage, initial guess, pair natural orbital construction, amplitude iterations, triples correction) are achieved in a linear scaling fashion. In addition, a number of additional algorithmic improvements are reported that lead to significant speedups of the method. The new, linear-scaling DLPNO-CCSD(T) implementation typically is 7 times faster than the previous implementation and consumes 4 times less disk space for large three-dimensional systems. For linear systems, the performance gains and memory savings are substantially larger. Calculations with more than 20 000 basis functions and 1000 atoms are reported in this work. In all cases, the time required for the coupled cluster step is comparable to or lower than for the preceding Hartree-Fock calculation, even if this is carried out with the efficient resolution-of-the-identity and chain-of-spheres approximations. The new implementation even reduces the error in absolute correlation energies by about a factor of two, compared to the already accurate previous implementation.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2015

WE-EF-207-07: Dual Energy CT with One Full Scan and a Second Sparse-View Scan Using Structure Preserving Iterative Reconstruction (SPIR)

Purpose: Conventional dual energy CT (DECT) reconstructs CT and basis material images from two full-size projection datasets with different energy spectra. To relax the data requirement, we propose an iterative DECT reconstruction algorithm using one full scan and a second sparse-view scan by utilizing redundant structural information of the same object acquired at two different energies. Methods: We first reconstruct a full-scan CT image using filtered-backprojection (FBP) algorithm. The material similarities of each pixel with other pixels are calculated by an exponential function about pixel value differences. We assume that the material similarities of pixels remains in the second CT scan, although pixel values may vary. An iterative method is designed to reconstruct the second CT image from reduced projections. Under the data fidelity constraint, the algorithm minimizes the L2 norm of the difference between pixel value and its estimation, which is the average of other pixel values weighted by their similarities. The proposed algorithm, referred to as structure preserving iterative reconstruction (SPIR), is evaluated on physical phantoms. Results: On the Catphan600 phantom, SPIR-based DECT method with a second 10-view scan reduces the noise standard deviation of a full-scan FBP CT reconstruction by a factor of 4 with well-maintained spatial resolution, while iterative reconstruction using total-variation regularization (TVR) degrades the spatial resolution at the same noise level. The proposed method achieves less than 1% measurement difference on electron density map compared with the conventional two-full-scan DECT. On an anthropomorphic pediatric phantom, our method successfully reconstructs the complicated vertebra structures and decomposes bone and soft tissue. Conclusion: We develop an effective method to reduce the number of views and therefore data acquisition in DECT. We show that SPIR-based DECT using one full scan and a second 10-view scan can provide high-quality DECT images and accurate electron density maps as conventional two-full-scan DECT.

60 APPLIED LIFE SCIENCES↗
NASA NTRS2004

Meridional Variations of C2H2 and C2H6 in Jupiter's Atmosphere from Cassini CIRS Infrared Spectra

Hydrocarbons such as acetylene (C2H2) and ethane (C2H6) are important tracers in Jupiter's atmosphere, constraining our models of the chemical and dynamical processes. However, our knowledge of the vertical and meridional variations of their abundances has remained sparse. During the flyby of the Cassini spacecraft in December 2000, the Composite Infrared Spectrometer (CIRS) instrument was used to map the spatial variation of emissions from 10-1400 cm(sup -1) (1000-7 microns). In this paper we analyze a zonally-averaged set of CIRS spectra taken at the highest (0.5 cm(sup -1)) resolution, to infer atmospheric temperatures in the stratosphere at 0.5-20 mbar via the v4 band of CH4, and in the troposphere at 150-400 mbar, via the H2 absorption at 600-800 cm(sup -1). Simultaneously, we retrieve the abundances of C2H2 and C2H6 via the v5 and vg bands respectively. Tropospheric absorption and stratospheric emission are highly anti-correlated at the CIRS resolution, introducing a non-uniqueness into the retrievals, such that vertical gradient and column abundance cannot both be found without additional constraints. Assuming profile gradients from photochemical calculations, we show that the column abundance of C2H2 decreases sharply towards the poles by a factor approximately 4, while C2H6 is unchanged in the north and increasing in the south, by a factor approximately 1.8. An explanation for the meridional trends is proposed in terms of a combination of photochemistry and dynamics. Poleward, the decreasing UV flux is predicted to decrease the abundances of C2H2 and C2H6 by factors 2.7 and 3.5 respectively at a latitude 70 deg. However, the lifetime of C2H6 in the stratosphere (5 x 10(exp 9)) is much longer than the dynamical timescale for meridional motions inferred from SL-9 debris (5 x 10(exp 8 s)), and therefore the constant or rising abundance towards high latitudes likely indicates that meridional mixing dominates over photochemical effects. For C2H2, the opposite occurs, with the relatively short photochemical lifetime (3 x 10(exp 7 s)), compared to meridional mixing times, ensuring that the expected photochemical trends are visible.

Nixon, C. A.↗
DOE OSTI2018

False alarms and the IMS infrasound network: Understanding the factors influencing the creation of false events

The International Monitoring System (IMS) infrasound network has been designed to acquire the necessary data to detect and locate explosions in the atmosphere with a yield equivalent to 1 kiloton of TNT anywhere on Earth. A major associated challenge is the task of automatically processing data from all IMS infrasound stations to identify possible nuclear tests for subsequent review by analysts. This paper is the first attempt to quantify the false alarm rate (FAR) of the IMS network, and in particular to assess how the FAR is affected by the numbers and distributions of detections at each infrasound station. To ensure that the results are sufficiently general, and not dependent entirely on one detection algorithm, the assessment is based on two detection algorithms that can be thought of as end members in their approach to the trade-off between missed detections and false alarms. The results show that the FAR for events formed at only two arrays is extremely high (ranging from 10’s to 100’s of false events per day across the IMS network, depending on the detector tuning). It is further shown that the FAR for events formed at three or more IMS arrays is driven by ocean-generated waves (microbaroms), despite efforts within both detection algorithms for avoiding these signals, indicating that further research into this issue is merited. Overall, the results highlight the challenge of processing data from a globally sparse network of stations to detect and form events. On the basis of these results, it is shown that further work on understanding signal and noise characteristics is needed before attempting to automatically form events detected at only two arrays on the global network. In particular, more work is required to reduce false alarms caused by the detection of microbarom signals.

47 OTHER INSTRUMENTATION↗
DOE OSTI2020

Sparse thermal data for cellular automata modeling of grain structure in additive manufacturing

Grain growth in the wake of the melt pool formed during alloy-based additive manufacturing (AM) is complex and multifaceted, depending on parameters governing heat transport, fluid flow, and solidification itself. Cellular automata (CA) models have proven effective in providing computationally efficient and physically sound predictions of grain structure for several AM problems, but their efficiency is tied to the performance of heat transport models. CA models use only a small portion of the problem's temperature data (near the moving melt pool boundary), and much of the CA calculations do not affect the final result due to re-melting of material. Coupling of and communication between heat transport and solidification models, and eliminating operations irrelevant towards final grain structure prediction, will be necessary for using these methods for efficient simulation of large parts. In this work, we introduce a procedure of decoupling the CA from temperature field simulation, using files of relevant temperature data written by the heat transport model. This approach is validated against the standard coupling approach using data obtained through the computational fluid dynamics software OpenFOAM. Negligible differences are seen in grain size, volume, and texture distributions for multilayer simulation of test problems, while the quantity of temperature data for these test problems was reduced by four orders of magnitude (from 100s of GB to 10s of MB) and the code performance sped up by a factor of around 50. Furthermore, variability in microstructure as a function of cell size, substrate, time step, and nucleation parameters is studied, and it is found that cell sizes less than or equal to 1.67 μm and sufficiently small time steps yield Αstatistically equivalent microstructures. Finally, a potential use case for this CA approach—the layer-wise convergence in grain structure starting from extremes in initial grain size—is examined. This approach's ability to simulate expected trends in nucleation and epitaxial grain growth for large regions of microstructure, simulated independently of heat transport models themselves, should prove useful for investigation of various microstructure uncertainties and prediction of part-scale experimental results.

36 MATERIALS SCIENCE↗
DOE OSTI2025

RatXcan: A framework for cross-species integration of genome-wide association and gene expression data

Genome-wide association studies (GWAS) have implicated specific alleles and genes as risk factors for numerous complex traits. However, translating GWAS results into biologically and therapeutically meaningful discoveries remains extremely challenging. Most GWAS results identify noncoding regions of the genome, suggesting that differences in gene regulation are the major driver of trait variability. To better integrate GWAS results with gene regulatory polymorphisms, we previously developed PrediXcan (also known as “transcriptome-wide association studies” orTWAS), which maps SNPs to predicted gene expression using GWAS data. In this study, we developed RatXcan, a framework that extends this methodology to outbred heterogeneous stock (HS) rats. RatXcan accounts for the close familial relationships among HS rats by modeling the relatedness with a random effect that encodes the genetic relatedness. RatXcan also corrects for polygenic-driven inflation because of the equivalence between a relatedness random effect and the infinitesimal polygenic model. To develop RatXcan, we trained transcript predictors for 8,934 genes using reference genotype and expression data from five rat brain regions. We found that the cis genetic architecture of gene expression in both rats and humans was sparse and similar across brain tissues. We tested the association between predicted expression in rats and two example traits (body length and BMI) using phenotype and genotype data from 5,401 densely genotyped HS rats and identified a significant enrichment between the genes associated with rat and human body length and BMI. Thus, RatXcan represents a valuable tool for identifying the relationship between gene expression and phenotypes across species and paves the way to explore shared biological mechanisms of complex traits.

Genetics & Heredity↗
NASA NTRS2016

Overview of Global Monitoring of Terrestrial Chlorophyll Fluorescence from Space

Despite the critical importance of photosynthesis for the Earth system, understanding how it is influenced by factors such as climate variability, disturbance history, and water or nutrient availability remains a challenge because of the complex interactions and the lack of GPP measurements at various temporal and spatial scales. Space observations of the sun-induced chlorophyll fluorescence (SIF) electromagnetic signal emitted by plants in the 650-850nm spectral range hold the promise of providing a new view of vegetation photosynthesis on a global basis. Global retrievals of SIF from space have recently been achieved from a number of spaceborne spectrometers originally intended for atmospheric research. Despite not having been designed for land applications, such instruments have turned out to provide the necessary spectral and radiometric sensitivity for SIF retrieval from space. The first global measurements of SIF were achieved in 2011 from spectra acquired by the Japanese GOSAT mission launched in 2009. The retrieval takes advantage of the high spectral resolution provided by GOSATs Fourier Transform Spectrometer (FTS) which allows the evaluation of the in-filling of solar Fraunhofer lines by SIF. Unfortunately, GOSAT only provides a sparse spatial sampling with individual soundings separated by several hundred kilometers. Complementary, the Global Ozone Monitoring Experiment-2 (GOME-2) instruments onboard MetOp-A and MetOp-B enable SIF retrievals since 2007 with a continuous and global spatial coverage. GOME-2 measures in the red and near-infrared (NIR) spectral regions with a spectral resolution of 0.5 nm and a pixel size of up to 40x40 km2. Most recently, another global and spatially continuous data set of SIF retrievals at 740 nm spanning the 2003-2012 time frame has been produced from ENVISATSCIAMACHY. This observational scenario has been completed by the first fluorescence data from the NASA-JPL OCO-2 mission (launched in July 2014) and the upcoming Copernicus' Sentinel 5-Precursor to be launched in early 2016. OCO-2 and TROPOMI offer the possibility of monitoring SIF globally with a 100-fold improvement in spatial and temporal resolution with respect to the current measurements from the GOSAT, GOME-2 and SCIAMACHY missions. In this contribution, we will provide an overview of existing global SIF data sets derived from space-based atmospheric spectrometers and will demonstrate the potential of such data to improve our knowledge of vegetation photosynthesis and gross primary production at the synoptic scale. We will show examples of ongoing research exploiting SIF data for an improved monitoring of photosynthetic activity in different ecosystems, including large crop belts worldwide, the Amazon rainforest and boreal evergreen forests.

vegetation agriculture satellite↗
DOE OSTI2020

Modeling the smoky troposphere of the southeast Atlantic: a comparison to ORACLES airborne observations from September of 2016

The southeast Atlantic is home to well-defined smoke outflow from Africa coinciding vertically with extensive marine boundary-layer cloud decks, both reaching their climatological maxima in spatial extent around September. A framework is put forth for evaluating the performance of a range of global and regional aerosol models against observations made during the NASA ORACLES (ObseRvations of Aerosols above CLouds and their intEractionS) airborne mission in September 2016. The sparse airborne observations are first aggregated into 2o grid boxes and into three vertical layers: the cloud-topped marine boundary layer (MBL), the layer from cloud top to 3 km, and the 3-6 km layer. Aerosol extensive properties simulated for the entire study region for all September suggest that the 2016 ORACLES observations are reasonably representative of the regional monthly average, with systematic deviations of 30% or less. All six models typically place the bottom of the smoke layer at lower altitudes than do the airborne lidar observations by 300-1400 m, whereas model aerosol top heights are within 0-500 m of the observations. All but one of the models that report carbonaceous aerosol masses underestimate the ratio of particulate extinction to the masses, a proxy for mass extinction efficiency, in 3-6 km. Notable findings on individual models include that WRF-CAM5 predicts the mass of black carbon and organic aerosols with minor (~10% or less) biases. GEOS-5 overestimates the carbonaceous particle masses in the MBL by a factor of 3-6. Extinction coefficients in the free troposphere (FT) and above-cloud aerosol optical depth (ACAOD) are 10-30% lower in WRF-CAM5, 30-50% lower in GEOS-5, 10-40% higher in GEOS-Chem, 10-20% higher in EAM-E3SM except for the practically unbiased 3-6 km extinction, and 20-70% lower in the Unified Model, than the airborne in situ, lidar and sunphotometer measurements. ALADIN-Climate also underestimates the ACAOD, by 30%. GEOS-5 and GEOS-Chem predict carbon monoxide in the MBL with small (10% or less) negative biases, despite their overestimates of carbonaceous aerosol masses. Overall, this study highlights a new approach to utilizing airborne aerosol measurements for model diagnosis.

Shinozuka, Yohei↗
DOE OSTI2020

Postglacial migration shaped the genomic diversity and global distribution of the wild ancestor of lager-brewing hybrids

The wild, cold-adapted parent of hybrid lager-brewing yeasts, Saccharomyces eubayanus, has a complex and understudied natural history. The exploration of this diversity can be used both to develop new brewing applications and to enlighten our understanding of the dynamics of yeast evolution in the wild. Here, we integrate whole genome sequence and phenotypic data of 200 S. eubayanus strains, the largest collection known to date. S. eubayanus has a multilayered population structure, consisting of two major populations that are further structured into six subpopulations. Four of these subpopulations are found exclusively in the Patagonian region of South America; one is found predominantly in Patagonia and sparsely in Oceania and North America; and one is specific to the Holarctic ecozone. Plant host associations differed between subpopulations and between S. eubayanus and its sister species, Saccharomyces uvarum. S. eubayanus is most abundant and genetically diverse in northern Patagonia, where some locations harbor more genetic diversity than is found outside of South America, suggesting that northern Patagonia east of the Andes was a glacial refugium for this species. All but one subpopulation shows isolation-by-distance, and gene flow between subpopulations is low. However, there are strong signals of ancient and recent outcrossing, including two admixed lineages, one that is sympatric with and one that is mostly isolated from its parental populations. Using our extensive biogeographical data, we build a robust model that predicts all known and a handful of additional regions of the globe that are climatically suitable for S. eubayanus, including Europe where host accessibility and competitive exclusion by other Saccharomyces species may explain its continued elusiveness. We conclude that this industrially relevant species has rich natural diversity with many factors contributing to its complex distribution and natural history.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2016

Quantifying and reducing model-form uncertainties in Reynolds-averaged Navier–Stokes simulations: A data-driven, physics-informed Bayesian approach

Despite their well-known limitations, Reynolds-Averaged Navier–Stokes (RANS) models are still the workhorse tools for turbulent flow simulations in today's engineering analysis, design and optimization. While the predictive capability of RANS models depends on many factors, for many practical flows the turbulence models are by far the largest source of uncertainty. As RANS models are used in the design and safety evaluation of many mission-critical systems such as airplanes and nuclear power plants, quantifying their model-form uncertainties has significant implications in enabling risk-informed decision-making. In this work we develop a data-driven, physics-informed Bayesian framework for quantifying model-form uncertainties in RANS simulations. Uncertainties are introduced directly to the Reynolds stresses and are represented with compact parameterization accounting for empirical prior knowledge and physical constraints (e.g., realizability, smoothness, and symmetry). An iterative ensemble Kalman method is used to assimilate the prior knowledge and observation data in a Bayesian framework, and to propagate them to posterior distributions of velocities and other Quantities of Interest (QoIs). We use two representative cases, the flow over periodic hills and the flow in a square duct, to evaluate the performance of the proposed framework. Both cases are challenging for standard RANS turbulence models. Simulation results suggest that, even with very sparse observations, the obtained posterior mean velocities and other QoIs have significantly better agreement with the benchmark data compared to the baseline results. At most locations the posterior distribution adequately captures the true model error within the developed model form uncertainty bounds. The framework is a major improvement over existing black-box, physics-neutral methods for model-form uncertainty quantification, where prior knowledge and details of the models are not exploited. This approach has potential implications in many fields in which the governing equations are well understood but the model uncertainty comes from unresolved physical processes. - Highlights: • Proposed a physics–informed framework to quantify uncertainty in RANS simulations. • Framework incorporates physical prior knowledge and observation data. • Based on a rigorous Bayesian framework yet fully utilizes physical model. • Applicable for many complex physical systems beyond turbulent flows.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2018

High-Performance Sparse Matrix-Matrix Products on Intel KNL and Multicore Architectures

Sparse matrix-matrix multiplication (SpGEMM) is a computational primitive that is vastly used in areas ranging from traditional numerical applications to recent big data analysis and machine learning. While many SpGEMM algorithms have been proposed, hardware specific optimizations for multi- and many-core processors are lacking and a detailed analysis of their performance under various use cases and matrices is not available. We firstly identify and mitigate multiple bottlenecks with memory management and thread scheduling on Intel Xeon Phi (Knights Landing or KNL). Specifically targeting multi- and many-core processors, we develop a hash-table-based algorithm and optimize a heap-based shared-memory SpGEMM algorithm. We investigate their performance together with other publicly available codes. Different from the literature, our evaluation also includes use cases that are representative of real graph algorithms, such as multi-source breadth-first search or triangle counting. Our hash-table and heap-based algorithms are showing significant speedups from libraries in the majority of the cases while different algorithms dominate the other scenarios with different matrix size, sparsity, compression factor and operation type. We wrap up in-depth evaluation results and make a recipe to give the best SpGEMM algorithm for target scenario. In conclusion, a critical finding is that hash-table-based SpGEMM gets a significant performance boost if the nonzeros are not required to be sorted within each row of the output matrix.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS2011

A Study on the Potential Applications of Satellite Data in Air Quality Monitoring and Forecasting

In this study we explore the potential applications of MODIS (Moderate Resolution Imaging Spectroradiometer) -like satellite sensors in air quality research for some Asian regions. The MODIS aerosol optical thickness (AOT), NCEP global reanalysis meteorological data, and daily surface PM(sub 10) concentrations over China and Thailand from 2001 to 2009 were analyzed using simple and multiple regression models. The AOT-PM(sub 10) correlation demonstrates substantial seasonal and regional difference, likely reflecting variations in aerosol composition and atmospheric conditions, Meteorological factors, particularly relative humidity, were found to influence the AOT-PM(sub 10) relationship. Their inclusion in regression models leads to more accurate assessment of PM(sub 10) from space borne observations. We further introduced a simple method for employing the satellite data to empirically forecast surface particulate pollution, In general, AOT from the previous day (day 0) is used as a predicator variable, along with the forecasted meteorology for the following day (day 1), to predict the PM(sub 10) level for day 1. The contribution of regional transport is represented by backward trajectories combined with AOT. This method was evaluated through PM(sub 10) hindcasts for 2008-2009, using ohservations from 2005 to 2007 as a training data set to obtain model coefficients. For five big Chinese cities, over 50% of the hindcasts have percentage error less than or equal to 30%. Similar performance was achieved for cities in northern Thailand. The MODIS AOT data are responsible for at least part of the demonstrated forecasting skill. This method can be easily adapted for other regions, but is probably most useful for those having sparse ground monitoring networks or no access to sophisticated deterministic models. We also highlight several existing issues, including some inherent to a regression-based approach as exemplified by a case study for Beijing, Further studies will be necessa1Y before satellite data can see more extensive applications in the operational air quality monitoring and forecasting.

Li, Can↗
DOE OSTI2014

SU-E-I-20: Dead Time Count Loss Compensation in SPECT/CT: Projection Versus Global Correction

Purpose: To compare projection-based versus global correction that compensate for deadtime count loss in SPECT/CT images. Methods: SPECT/CT images of an IEC phantom (2.3GBq 99mTc) with ∼10% deadtime loss containing the 37mm (uptake 3), 28 and 22mm (uptake 6) spheres were acquired using a 2 detector SPECT/CT system with 64 projections/detector and 15 s/projection. The deadtime, Ti and the true count rate, Ni at each projection, i was calculated using the monitor-source method. Deadtime corrected SPECT were reconstructed twice: (1) with projections that were individually-corrected for deadtime-losses; and (2) with original projections with losses and then correcting the reconstructed SPECT images using a scaling factor equal to the inverse of the average fractional loss for 5 projections/detector. For both cases, the SPECT images were reconstructed using OSEM with attenuation and scatter corrections. The two SPECT datasets were assessed by comparing line profiles in xyplane and z-axis, evaluating the count recoveries, and comparing ROI statistics. Higher deadtime losses (up to 50%) were also simulated to the individually corrected projections by multiplying each projection i by exp(-a*Ni*Ti), where a is a scalar. Additionally, deadtime corrections in phantoms with different geometries and deadtime losses were also explored. The same two correction methods were carried for all these data sets. Results: Averaging the deadtime losses in 5 projections/detector suffices to recover >99% of the loss counts in most clinical cases. The line profiles (xyplane and z-axis) and the statistics in the ROIs drawn in the SPECT images corrected using both methods showed agreement within the statistical noise. The count-loss recoveries in the two methods also agree within >99%. Conclusion: The projection-based and the global correction yield visually indistinguishable SPECT images. The global correction based on sparse sampling of projections losses allows for accurate SPECT deadtime loss correction while keeping the study duration reasonable.

60 APPLIED LIFE SCIENCES↗