Search NASA⌕ Search

SEARCH · Search NASA

Results for “data dependencies”

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 1,315 records · Page 73

A High Performance Sparse Tensor Algebra Compiler in MLIR

Sparse tensor algebra is widely used in many applications, including scientific computing, machine learning, and data analytics. The performance of sparse tensor algebra kernels strongly depends on the intrinsic characteristics of the input tensors, hence many storage formats are designed for tensors to achieve optimal performance for particular applications/architectures, which makes it challenging to implement and optimize every tensor operation of interest on a given architecture. We propose a tensor algebra domain-specific language (DSL) and compiler framework to automatically generate kernels for mixed sparse-dense tensor algebra operations. The proposed DSL provides high-level programming abstractions that resemble the familiar Einstein notation to represent tensor algebra operations. The compiler introduces a new Sparse Tensor Algebra dialect built on top of LLVM's extensible MLIR compiler infrastructure for efficient code generation while covering a wide range of tensor storage formats. Our compiler also leverages input-dependent code optimization to enhance data locality for better performance. Our results show that the performance of automatically generated kernels outperforms the state-of-the-art sparse tensor algebra compiler, with up to 20.92x, 6.39x, and 13.9x performance improvement over state-of-the-art tensor algebra compilers, for parallel SpMV, SpMM, and TTM, respectively.

Tian, Ruiqin↗

Hierarchical Task-Data Parallelism using Kokkos and Qthreads

This report describes a new capability for hierarchical task-data parallelism using Sandia's Kokkos and Qthreads, and evaluation of this capability with sparse matrix Cholesky factorization and social network triangle enumeration mini-applications. Hierarchical task-data parallelism consists of a collection of tasks with executes-after dependences where each task contains data parallel operations performed on a team of hardware threads. The collection of tasks and dependences form a directed acyclic graph of tasks - a task DAG. Major challenges of this research and development effort include: portability and performance across multicore CPU; manycore Intel Xeon Phi, and NVIDIA GPU architectures; scalability with respect to hardware concurrency and size of the task DAG; and usability of the application programmer interface (API).

97 MATHEMATICS AND COMPUTING↗

Magnetism in Solid Oxygen Studied by High-Pressure Neutron Diffraction

This article reviews progress achieved over the last ~ 15 years in our understanding of magnetism in solid oxygen under high pressure with a particular emphasis on the contribution of neutron diffraction in the multi-GPa range. The paper highlights the unexpected complexity of magnetic structures in the δ phase at 5–8 GPa, presents data on the pressure dependence of diffuse scattering in β-O{sub 2} and discusses potential magnetism in ε-O{sub 2}. High-resolution diffraction data of all three solid phases at ambient pressure are presented.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Trade-off between morphology, extended defects, and compositional fluctuation induced carrier localization in high In-content InGaN films

We elucidate the role of growth parameters (III/N flux ratio, temperature T{sub G}) on the morphological and structural properties, as well as compositional homogeneity and carrier localization effects of high In-content (x(In) > 0.75) In–polar InGaN films grown by plasma–assisted molecular beam epitaxy (PAMBE). Variations in III/N flux ratio evidence that higher excess of In yields higher threading dislocation densities as well as larger compositional inhomogeneity as measured by x-ray diffraction. Most interestingly, by variation of growth temperature T{sub G} we find a significant trade-off between improved morphological quality and compositional homogeneity at low–T{sub G} (∼450–550 °C) versus improved threading dislocation densities at high–T{sub G} (∼600–630 °C), as exemplified for InGaN films with x(In) = 0.9. The enhanced compositional homogeneity mediated by low–T{sub G} growth is confirmed by systematic temperature-dependent photoluminescence (PL) spectroscopy data, such as lower PL peakwidths, >5× higher PL efficiency (less temperature-induced quenching) and a distinctly different temperature-dependent S-shape behavior of the PL peak energy. From these, we find that the carrier localization energy is as low as ∼20 meV for low–T{sub G} grown films (T{sub G} = 550 °C), while it rises to ∼70 meV for high–T{sub G} grown films (T{sub G} = 630 °C) right below the onset of In–N dissociation. These findings point out that for the kinetically limited metal-rich PAMBE growth of high In-content InGaN a III/N flux ratio of ∼1 and low-to-intermediate T{sub G} are required to realize optically more efficient materials.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Analysis of 83m Kr prompt scintillation signals in the PIXeY detector

Prompt scintillation signals from 83m Kr calibration sources are a useful metric to calibrate the spatial variation of light collection efficiency and electric field magnitude of a two phase liquid-gas xenon time projection chamber. Because 83m Kr decays in two steps, there are two prompt scintillation pulses for each calibration event, denoted S1a and S1b. We study the ratio of S1b to S1a signal sizes in the Particle Identification in Xenon at Yale (PIXeY) experiment and its dependence on the time separation between the two signals (Δ t), notably its increase at low Δ t. Finally, in PIXeY data, the Δ t dependence of S1b/S1a is observed to exhibit two exponential components: one with a time constant of 0.05 ± 0.02 μ s, which can be attributed to processing effects and pulse overlap and one with a time constant of 10.2 ± 2.2 μs that increases in amplitude with electric drift field, the origin of which is not yet understood.

47 OTHER INSTRUMENTATION↗

Growth of (101) Faces of Tetragonal Lysozyme Crystals: Measured Growth Rate Trends

Earlier extensive measurements of the growth rates of the (110) face of tetragonal lysozyme crystals had shown unexpected dependencies on the supersaturation. In this study similar growth rate measurements were done for the (101) faces of the crystals. The data show a similar dependence on the supersaturation, becoming appreciable only at high supersaturations, reaching a maximum value and then decreasing. As reported in earlier studies, the (101) growth rates are larger at low supersaturations than the (110) growth rates at the same conditions, and smaller at high supersaturations. These trends suggest that the growth mechanism of the (101) is similar to that of the (110) face, involving the addition of lysozyme aggregates formed in solution, but with a growth unit smaller than that of the (110) face.

Forsythe, Elizabeth L.↗

Nuclear Dependence of Charm Production

With data taken by SELEX, which accumulated data during the 1996-1997 fixed target run at Fermilab, we study the production of charmed hadrons on copper and carbon targets with {Sigma}{sup -}, p, {pi}{sup -}, and {pi}{sup +} beams. Parameterizing the production cross section {infinity} A{sup {alpha}}, A being the atomic number, we determine {alpha} for D{sup +}, D{sup 0}, D{sub s}{sup +}, D{sup +}(2010), {Lambda}{sub c}{sup +}, and their respective anti-particles, as a function of their transverse momentum p{sub t} and scaled longitudinal momentum x{sub F}. Within our statistics there is no dependence of {alpha} on x{sub F} for any charm species for the interval 0.1 < x{sub F} < 1.0. The average value of {alpha} for charm production by pion beams is {alpha}{sub meson} = 0.850 {+-} 0.028. This is somewhat larger than the corresponding average {alpha}{sub baryon} = 0.755 {+-} 0.016 for charm production by baryon beams ({Sigma}{sup -}, p).

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Data fusion for a vision-aided radiological detection system: Calibration algorithm performance

In order to improve the ability to detect, locate, track and identify nuclear/radiological threats, the University of Florida nuclear detection community has teamed up with the 3D vision community to collaborate on a low cost data fusion system. The key is to develop an algorithm to fuse the data from multiple radiological and 3D vision sensors as one system. The system under development at the University of Florida is being assessed with various types of radiological detectors and widely available visual sensors. A series of experiments were devised utilizing two EJ-309 liquid organic scintillation detectors (one primary and one secondary), a Microsoft Kinect for Windows v2 sensor and a Velodyne HDL-32E High Definition LiDAR Sensor which is a highly sensitive vision sensor primarily used to generate data for self-driving cars. Each experiment consisted of 27 static measurements of a source arranged in a cube with three different distances in each dimension. The source used was Cf-252. The calibration algorithm developed is utilized to calibrate the relative 3D-location of the two different types of sensors without need to measure it by hand; thus, preventing operator manipulation and human errors. The algorithm can also account for the facility dependent deviation from ideal data fusion correlation. Use of the vision sensor to determine the location of a sensor would also limit the possible locations and it does not allow for room dependence (facility dependent deviation) to generate a detector pseudo-location to be used for data analysis later. Using manually measured source location data, our algorithm-predicted the offset detector location within an average of 20 cm calibration-difference to its actual location. Calibration-difference is the Euclidean distance from the algorithm predicted detector location to the measured detector location. The Kinect vision sensor data produced an average calibration-difference of 35 cm and the HDL-32E produced an average calibration-difference of 22 cm. Using NaI and He-3 detectors in place of the EJ-309, the calibration-difference was 52 cm for NaI and 75 cm for He-3. In conclusion, the algorithm is not detector dependent; however, from these results it was determined that detector dependent adjustments are required.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Cosmological constraints on induced gravity dark energy models

We study induced gravity dark energy models coupled with a simple monomial potential ∝ σ{sup n} and a positive exponent n . These simple potentials lead to viable dark energy models with a weak dependence on the exponent, which characterizes the accelerated expansion in the asymptotic attractor, when ordinary matter becomes negligible. We use recent cosmological data to constrain the coupling γ to the Ricci curvature, under the assumptions that the scalar field starts at rest deep in the radiation era and that the gravitational constant in the Einstein equations is compatible with the one measured in a Cavendish-like experiment. By using Planck 2015 data only, we obtain the 95 % CL bound γ < 0.0017 for n =4, which is further tightened to γ < 0.00075 by adding Baryonic Acoustic Oscillations (BAO) data. This latter bound improves by ∼ 30 % the limit obtained with the Planck 2013 data and the same compilation of BAO data. We discuss the dependence of the γ and Ġ{sub N}/G{sub N} (z = 0) on n.

79 ASTRONOMY AND ASTROPHYSICS↗

Analysis of Wind Tunnel Longitudinal Static and Oscillatory Data of the F-16XL Aircraft

Static and oscillatory wind tunnel data are presented for a 10-percent-scale model of an F-16XL aircraft. Static data include the effect of angle of attack, sideslip angle, and control surface deflections on aerodynamic coefficients. Dynamic data from small-amplitude oscillatory tests are presented at nominal values of angle of attack between 20 and 60 degrees. Model oscillations were performed at five frequencies from 0.6 to 2.9 Hz and one amplitude of 5 degrees. A simple harmonic analysis of the oscillatory data provided Fourier coefficients associated with the in-phase and out-of-phase components of the aerodynamic coefficients. A strong dependence of the oscillatory data on frequency led to the development of models with unsteady terms in the form of indicial functions. Two models expressing the variation of the in-phase and out-of-phase components with angle of attack and frequency were proposed and their parameters estimated from measured data.

Klein, Vladislav↗

Growth of (101) faces of tetragonal lysozyme crystals: measured growth-rate trends

Previous extensive measurements of the growth rates of the (110) face of tetragonal lysozyme crystals have shown unexpected dependencies on the supersaturation. In this study, similar growth-rate measurements were performed for the (101) faces of the crystals. The data show a similar dependence on the supersaturation, becoming appreciable only at high supersaturations, reaching a maximum value and then decreasing. The (101) growth rates are larger at low supersaturations than the (110) growth rates under the same conditions and are smaller at high supersaturations. These trends suggest that the growth mechanism of the (101) face is similar to that of the (110) face: both processes involve the addition of multimeric growth units formed in solution, but the average size of the units for the (101) face is likely to be smaller than for the (110) face.

Egg Proteins/chemistry↗

Determining the Optimal Number of Clusters with the Clustergram

Cluster analysis aids research in many different fields, from business to biology to aerospace. It consists of using statistical techniques to group objects in large sets of data into meaningful classes. However, this process of ordering data points presents much uncertainty because it involves several steps, many of which are subject to researcher judgment as well as inconsistencies depending on the specific data type and research goals. These steps include the method used to cluster the data, the variables on which the cluster analysis will be operating, the number of resulting clusters, and parts of the interpretation process. In most cases, the number of clusters must be guessed or estimated before employing the clustering method. Many remedies have been proposed, but none is unassailable and certainly not for all data types. Thus, the aim of current research for better techniques of determining the number of clusters is generally confined to demonstrating that the new technique excels other methods in performance for several disparate data types. Our research makes use of a new cluster-number-determination technique based on the clustergram: a graph that shows how the number of objects in the cluster and the cluster mean (the ordinate) change with the number of clusters (the abscissa). We use the features of the clustergram to make the best determination of the cluster-number.

Fluegemann, Joseph K.↗

Interplanetary magnetic field dependency of stable Sun-aligned polar cap arcs

This is the first analysis, using a statistically significant data set, of the morphological dependence of the presence, orientation, and motion of stable sun-aligned polar cap arcs upon the vector interplanetary magnetic field (IMF). For the one winter season analyzed we had 1392 all-sky 630.0-nm images of 2-min resolution containing a total of 150 polar cap arcs, all with corresponding values of the IMF as measured by International Monitoring Platform (IMP) 8 or International Sun Earth Explorer (ISEE) 2. After demonstrating an unbiased data set with smooth normal distributions of events versus the dimensions of time, space, and IMF component, we examine IMF dependencies of the properties of the optical arcs. A well-defined dependence for B(sub z) is found for the presence/absence of stable Sun-aligned polar cap arcs. Consistent with previous statistical studies, the probability of observing polar cap aurora steadily increases for larger positive values of B(sub z), and linearly decreases when B(sub z) becomes more negative. The probability of observing Sun-aligned arcs within the polar cap is determined to vary sharply as a function of the arc location; arcs were observed 40% of the time on the dawnside and only 10% on the duskside. This implies an overall probability of at least 40% for the whole polar cap. 20% of the arcs were observed during 'southward IMF conditions,' but in fact under closer inspection were found to have been formed under northward IMF conditions; these 'residual' positive B(sub z) arcs ha d a delayed residence time in the polar cap of about what would be expected after a north to south transition of B(sub z). A firm dependence on B(sub y) is also found for both the orientation and the dawn-dusk direction of motion of the arcs. All the arcs are Sun-aligned to a first approximation, but present deviations from this orientation, depending primarily upon the location of the arc in corrected geomagnetic (CG) coordinates. The arcs populating the 06-12 and the 12-18 quadrants of the CG coordinate system point toward the cusp. The B(sub y) dependency of the arc alignment is consistent with a cusp displacement in local time according to the sign of B(sub y). We found that the arc direction of motion depended both on B(sub y) and the arc location within the polar cap. For a given value of B(sub y) two well-defined regions (or cells) exist. Within each cell the arcs move in the same direction toward the boundary between the cells. The arcs located in the duskside move dawnward; those in the dawnside move duskward. The relative size of these dusk and dawn regions (or cells) are controlled by the magnitude of B(sub y). This persistent dusk-dawn motion fo the polar cap arcs is interpreted in terms of newly open flux tubes entering the polar cap and exerting a displacement of the convective cells and the polar cap arcs that are embedded within them.

Valladares, C. E.↗

Reduced-Order Modeling for Flutter/LCO Using Recurrent Artificial Neural Network

The present study demonstrates the efficacy of a recurrent artificial neural network to provide a high fidelity time-dependent nonlinear reduced-order model (ROM) for flutter/limit-cycle oscillation (LCO) modeling. An artificial neural network is a relatively straightforward nonlinear method for modeling an input-output relationship from a set of known data, for which we use the radial basis function (RBF) with its parameters determined through a training process. The resulting RBF neural network, however, is only static and is not yet adequate for an application to problems of dynamic nature. The recurrent neural network method [1] is applied to construct a reduced order model resulting from a series of high-fidelity time-dependent data of aero-elastic simulations. Once the RBF neural network ROM is constructed properly, an accurate approximate solution can be obtained at a fraction of the cost of a full-order computation. The method derived during the study has been validated for predicting nonlinear aerodynamic forces in transonic flow and is capable of accurate flutter/LCO simulations. The obtained results indicate that the present recurrent RBF neural network is accurate and efficient for nonlinear aero-elastic system analysis

Yao, Weigang↗

Communication-avoiding optimization methods for distributed massive-scale sparse inverse covariance estimation

Across a variety of scientific disciplines, sparse inverse covariance estimation is a popular tool for capturing the underlying dependency relationships in multivariate data. Unfortunately, most estimators are not scalable enough to handle the sizes of modern high-dimensional data sets (often on the order of terabytes), and assume Gaussian samples. To address these deficiencies, we introduce HP-CONCORD, a highly scalable optimization method for estimating a sparse inverse covariance matrix based on a regularized pseudolikelihood framework, without assuming Gaussianity. Our parallel proximal gradient method uses a novel communication-avoiding linear algebra algorithm and runs across a multi-node cluster with up to 1k nodes (24k cores), achieving parallel scalability on problems with up to ≈819 billion parameters (1.28 million dimensions); even on a single node, HP-CONCORD demonstrates scalability, outperforming a state-of-the-art method. We also use HP-CONCORD to estimate the underlying dependency structure of the brain from fMRI data, and use the result to identify functional regions automatically. The results show good agreement with a clustering from the neuroscience literature.

97 MATHEMATICS AND COMPUTING↗

Communication-Avoiding Optimization Methods for Distributed Massive-Scale Sparse Inverse Covariance Estimation.

Across a variety of scientific disciplines, sparse inverse covariance estimation is a popular tool for capturing the underlying dependency relationships in multivariate data. Unfortunately, most estimators are not scalable enough to handle the sizes of modern high-dimensional data sets (often on the order of terabytes), and assume Gaussian samples. To address these deficiencies, we introduce HP-CONCORD, a highly scalable optimization method for estimating a sparse inverse covariance matrix based on a regularized pseudolikelihood framework, without assuming Gaussianity. Our parallel proximal gradient method uses a novel communication-avoiding linear algebra algorithm and runs across a multi-node cluster with up to 1k nodes (24k cores), achieving parallel scalability on problems with up to ~819 billion parameters (1.28 million dimensions); even on a single node, HP-CONCORD demonstrates scalability, outperforming a state-of-the-art method. We also use HP-CONCORD to estimate the underlying dependency structure of the brain from fMRI data, and use the result to identify functional regions automatically. The results show good agreement with a clustering from the neuroscience literature.

Koanantakool, Penporn↗

Kinetic characterization of Vibrio cholerae ApbE: Substrate specificity and regulatory mechanisms

ApbE is a member of a novel family of flavin transferases that incorporates flavin mononucleotide (FMN) to subunits of diverse respiratory complexes, which fulfill important homeostatic functions. In this work a detailed characterization of Vibrio cholerae ApbE physiologic activity, substrate specificity and pH dependency was carried out. The data obtained show novel characteristics of the regulation and function of this family. For instance, our experiments indicate that divalent cations are essential for ApbE function, and that the selectivity depends largely on size and the coordination sphere of the cation. Our data also show that ApbE regulation by pH, ADP and potassium is an important mechanism that enhances the adaptation, survival and colonization of V. cholerae in the small intestine. Moreover, studies of the pH-dependency of the activity show that the reaction is favored under alkaline conditions, with a pKa of 8.4. These studies, together with sequence and structure analysis allowed us to identify His257, which is absolutely conserved in the family, as a candidate for the residue whose deprotonation controls the activity. Remarkably, the mutant H257G abolished the flavin transfer activity, strongly indicating that this residue plays an important role in the catalytic mechanism of ApbE.

59 BASIC BIOLOGICAL SCIENCES↗

Contrastive learning for robust representations of neutrino data

In neutrino physics, analyses often depend on large simulated datasets, making it essential for models to generalize effectively to real-world detector data. Contrastive learning, a well-established technique in deep learning, offers a promising solution to this challenge. By applying controlled data augmentations to simulated data, contrastive learning enables the extraction of robust and transferable features. This improves the ability of models trained on simulations to adapt to real experimental data distributions. In this paper, we investigate the application of contrastive learning methods in the context of neutrino physics. Through a combination of empirical evaluations and theoretical insights, we demonstrate how contrastive learning enhances model performance and adaptability. Additionally, we compare it to other domain adaptation techniques, highlighting the unique advantages of contrastive learning for this field. Published by the American Physical Society 2025

Wilkinson, Alex (ORCID:0000000253404506)↗