Search NASA⌕ Search

SEARCH · Search NASA

Results for “Data reduction methods”

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 649 records · Page 36

A flexible class of priors for orthonormal matrices with basis function-specific structure

Statistical modeling of high-dimensional matrix-valued data motivates the use of a low-rank representation that simultaneously summarizes key characteristics of the data and enables dimension reduction. Low-rank representations commonly factor the original data into the product of orthonormal basis functions and weights, where each basis function represents an independent feature of the data. However, the basis functions in these factorizations are typically computed using algorithmic methods that cannot quantify uncertainty or account for basis function correlation structure a priori. While there exist Bayesian methods that allow for a common correlation structure across basis functions, empirical examples motivate the need for basis function-specific dependence structure. We propose a prior distribution for orthonormal matrices that can explicitly model basis function-specific structure. The prior is used within a general probabilistic model for singular value decomposition to conduct posterior inference on the basis functions while accounting for measurement error and fixed effects. We discuss how the prior specification can be used for various scenarios and demonstrate favorable model properties through synthetic data examples. Finally, we apply our method to two-meter air temperature data from the Pacific Northwest, enhancing our understanding of the Earth system’s internal variability.

97 MATHEMATICS AND COMPUTING↗

Robust Machine Learning Inference from X-ray Absorption Near Edge Spectra through Featurization

X-ray absorption spectroscopy (XAS) is a commonly employed technique for characterizing functional materials. In particular, X-ray absorption near edge spectra (XANES) encode local coordination and electronic information, and machine learning approaches to extract this information are of significant interest. To date, most ML approaches for XANES have primarily focused on using the raw spectral intensities as input, overlooking the potential benefits of incorporating spectral transformations and dimensionality reduction techniques into ML predictions. Here, in this work, we focused on systematically comparing the impact of different featurization methods on the performance of ML models for XAS analysis. We evaluated the classification and regression capabilities of these models on computed data sets and validated their performance on previously unseen experimental data sets. Our analysis revealed an intriguing discovery: the cumulative distribution function feature achieves both high prediction accuracy and exceptional transferability. This remarkably robust performance can be attributed to its tolerance to horizontal shifts in the spectra, which is crucial when validating models using experimental data. While this work exclusively focuses on XANES analysis, we anticipate that the methodology presented here will hold promise as a versatile asset to the broader spectroscopy community.

36 MATERIALS SCIENCE↗

Propagation of nonlinear dust magnetoacoustic waves in cylindrical geometry

The cylindrical dust magnetoacoustic shocks in dissipative and solitons in non-dissipative dust-ion plasmas are studied. The dust particles are assumed to be fully negatively charged so that the density of electrons in comparison with dust and ions is negligible. The cylindrical Korteweg-de Vries Burgers and cylindrical Korteweg-de Vries equations are derived for dust magnetosonic waves by employing the reductive perturbation method. It is found that dust plasma parameters such as dust charge number, dust density, and ion temperature have strong influence on the profile of dust magnetoacoustic shocks and solitons. The results are also obtained numerically by using the data from laboratory experiments on dusty plasmas.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Active sites of ligand-protected Au{sub 25} nanoparticle catalysts for CO{sub 2} electroreduction to CO

Recent experimental studies have reported the electrochemical reduction of carbon dioxide (CO{sub 2}) into CO at atomically precise negatively charged Au{sub 25}{sup −} nanoclusters. The studies showed CO{sub 2} conversion at remarkably low overpotentials, but the exact mechanisms and nature of the active sites remain unclear. We used first-principles density functional theory and continuum solvation models to examine the role of the cluster during electrochemical CO{sub 2} reduction and analyze the free energies of proposed intermediate species. Contrary to previous assumptions, our results show that the fully ligand protected cluster is not an active CO{sub 2} reduction catalyst because formation of the crucial carboxyl intermediate required very high electrochemical potentials. Instead, our calculations suggest that the reduction process likely occurs on a dethiolated gold site, and adsorbed carboxyl intermediate formation was significantly stabilized at dethiolated gold sites. These findings point to the crucial role of exposed metal sites during electrochemical CO{sub 2} reduction at gold nanocluster catalysts.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

STEMDL

The code "STEMDL" provides highly-optimized multi-gpu implementations of DL models of applicability in electron microscopy via the use of mixed-precision arithmetic, native graph construction and evaluation in Tensorflow, and adaptive scaling of learning parameters to avoid numerical instabilities due to reduced precision. STEMDL also uses low-level methods from the Horovod library for gradient communication and reduction and adds an additional optimization by allowing the user to scheduling the gradient reduction operations based on their position within the layers of a neural net. STEMDL provides optimized I/O routines for data staging and data reading/writing in both Tensorflow and Pytorch using multi-processing and asynchronous transfer between host and device. .

Laanait, Nouamane↗

Efficient Stochastic Inversion Using Adjoint Models and Kernel-PCA

Performing stochastic inversion on a computationally expensive forward simulation model with a high-dimensional uncertain parameter space (e.g. a spatial random field) is computationally prohibitive even with gradient information provided. Moreover, the ‘nonlinear’ mapping from parameters to observables generally gives rise to non-Gaussian posteriors even with Gaussian priors, thus hampering the use of efficient inversion algorithms designed for models with Gaussian assumptions. In this work, we propose a novel Bayesian stochastic inversion methodology, characterized by a tight coupling between a gradient-based Langevin Markov Chain Monte Carlo (LMCMC) method and a kernel principal component analysis (KPCA). This approach addresses the ‘curse-of-dimensionality’ via KPCA to identify a low-dimensional feature space within the high-dimensional and nonlinearly correlated spatial random field. Moreover, non-Gaussian full posterior probability distribution functions are estimated via an efficient LMCMC method on both the projected low-dimensional feature space and the recovered high-dimensional parameter space. We demonstrate this computational framework by integrating and adapting recent developments such as data-driven statistics-on-manifolds constructions and reduction-through-projection techniques to solve inverse problems in linear elasticity.

97 MATHEMATICS AND COMPUTING↗

Criticality and Radiation Shielding Impacts of ENDF/B-VIII.0 [Slides]

Several impacts of ENDF/B-VIII.0 on Radiation Shielding Applications were observed by researchers. Most reaction rates, attenuation ratios, and dose rates are similar for ENDF/B-VII.1 and ENDF/B-VIII.0. A significant reduction in reactor dosimetry results was noted. Investigation continues to reveal underlying data for discrepancies.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Modeling indicates efficient vaccine-based interventions for the elimination of hepatitis C virus among persons who inject drugs in metropolitan Chicago

Background and aims People who inject drugs (PWID) are at highest risk for acquiring and transmitting hepatitis C (HCV) infection. The recent availability of oral direct-acting antiviral (DAA) therapy with reported cure rates >90% can prevent HCV transmission, making HCV elimination an attainable goal among PWID. The World Health Organization (WHO) recently proposed a 90% reduction in HCV incidence as a key objective. Yet, given barriers to the use of DAAs in PWID, including cost, restricted access to DAAs, and risk of reinfection, combination strategies including the availability of effective vaccines are needed to eradicate HCV as a public health threat. This study hopes to model the cost and efficacy of a dual modality approach using HCV vaccines combined with DAAs to reduce HCV incidence by 90% and prevalence by 50% in PWID populations. Methods We developed a mathematical model that represents the HCV epidemic among PWID and calibrated it to empirical data from metropolitan Chicago, Illinois. Four medical interventions were considered: vaccination of HCV naive PWID, DAA treatment, DAA treatment followed by vaccination, and, a combination of vaccination and DAA treatment. Results The combination of vaccination and DAAs is the lowest cost-expensive intervention for achieving the WHO target of 90% incidence reduction. The use of DAAs without a vaccine is much less cost-effective with the additional risk of reinfection after treatment. Vaccination of naïve PWID alone, even when scaled-up to all reachable PWID, cannot achieve 90% reduction of incidence in high-prevalence populations due to infections occurring before vaccination. Similarly, the lowest cost-expensive way to halve prevalence in 15 years is through the combination of vaccination and DAAs. Conclusions The modeling results underscore the importance of developing an effective HCV vaccine and augmenting DAAs with vaccines in HCV intervention strategies in order to achieve efficient reductions in incidence and prevalence.

60 APPLIED LIFE SCIENCES↗

A Real-Time Data Dashboard for Monitoring Travel Behavior

When running a smartphone app-based travel behavior study, it is important to keep participants engaged in contributing data. Long study durations and/or large samples make it difficult to engage participants in reporting their travel behavior. We developed a dashboard web application called 'emdash' to facilitate and ease the data collection maintenance task. It allows deployers to view trip trajectories, track which participants are uploading data, and generate report ready plots. Emdash can also be useful for deployers to identify ways to best encourage a particular transportation mode shift and to provide feedback and reporting to participants. Using the dashboard, a single admin was able to support a classic travel study with 80 participants. The dashboard is currently deployed in a two-year electric bike (e-bike) pilot study called CanBikeCO, conducted in six locations across Colorado, USA. CanBikeCO targets carbon emissions reduction and equity improvement by evaluating changes to participants' travel behavior when they are offered free e-bikes. Each CanBikeCO location has access to their own version of emdash to help them monitor and support participants. After testing and gathering feedback, we made improvements to the dashboard's usability and scalability, and proposed a method to make detecting data issues easier. Improvements include configurable options and limiting the volume of data sent to the dashboard as study size and duration increase.

ADVANCED PROPULSION SYSTEMS,MATHEMATICS AND COMPUT↗

Shifting temporal dynamics of human mobility in the United States

In this paper we analyze the average hourly temporal dynamics of human mobility in the United States from 2019 to 2020. We discuss how large decreases in human mobility nonuniformly effect the daily temporal dynamics of aggregate human behavior. The data used are weekly activity patterns for POIs from 2019 to 2020 in the United States, provided by SafeGraph and made openly available to academic and research institutions. We use clustering methods to create metrics describing how human activity changes throughout the day/week at the county and national levels. In response to significant mobility reductions starting March 2020, daily temporal patterns of human activity changed nonuniformly. Morning activity started later, and evening activity started earlier in 2020 compared to 2019, and temporal behavioral patterns on weekdays began to look more similar to weekends. The changes in daily temporal behavior persisted throughout the year even as total mobility levels recovered. The results provide insights on the changes in human behavior in response covid-19 policies and illustrate influences on social systems, health, and transportation networks.

99 GENERAL AND MISCELLANEOUS↗

Energetics and the Role of Defects in Fe(II)-Catalyzed Goethite Recrystallization from Molecular Simulations

Goethite is one of the most stable and common iron (III) minerals at the Earth’s near surface. However, recent isotope- tracer studies suggest that goethite continuously recrystallizes in the presence of aqueous Fe(II) ions. Such studies often indicate the presence of two regimes of atom exchange kinetics, a rapid stage assigned to reactive defect sites initially available at particle surfaces, followed by slower continuous exchange. An autocatalytic solid-state electron conduction model coupling Fe(II) oxidative adsorption to its reductive release at spatially distinct sites has been proposed, but the thermodynamic driving force has yet to be pinpointed. Here, using a novel hybrid/reactive molecular simulation method, for goethite (110) surfaces at circumneutral pH, we rigorously tested whether surface free energy minimization, including examining the role of structural defects, is sufficient to overcome the activation energy for interfacial electron transfer and conduction. The simulations quantitatively show that: (i). on smooth stable surfaces the available thermal energy at dynamic equilibrium is sufficient to sustain the slow continuous regime of atom exchange kinetics via short intra-surface electron conduction pathways of 1-2 nm (3-5 Fe site hops); (ii). in this slower regime, the model converges to atom exchange kinetics of 10 -5 Fe s -1 cm -2 , a rate recently deduced from stochastic modeling of experimental data and linked to the reductive dissolution rate of goethite; (iii). the driving force for initially rough defective goethite surfaces to smoothen can accelerate atom exchange to an extent quantitatively consistent with that observed in the initial rapid stage, in this case accessing conduction pathways of up to 8 nm. The findings suggest that the interaction of Fe(II) with initially defective goethite surfaces can drive, by the conduction model, atom exchange that is capable of recrystallizing the interiors of nanoscale particles, and that, closer to equilibrium on smooth surfaces, slower atom exchange continues in perpetuity but likely involving only the outermost atomic layers.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

SZ3: A Modular Framework for Composing Prediction-Based Error-Bounded Lossy Compressors

Today's scientific simulations require a significant reduction of data volume because of extremely large amounts of data they produce and the limited I/O bandwidth and storage space. Here, error-bounded lossy compression has been considered one of the most effective solutions to the above problem. In practice, however, the best-fit compression method often needs to be customized or optimized in particular because of diverse characteristics in different datasets and various user requirements on the compression quality and performance. In this paper, we address this issue with a novel modular, composable compression framework named SZ3. Our contributions are four-folds. (1) We develop SZ3 which features an innovative modular abstraction for the prediction-based compression framework, such that compression modules can be plugged in easily to create new compressors based on characteristics of data and user requirements. (2) We create a new compression pipeline by SZ3 for GAMESS data, which significantly improves the compression ratios over state-of-the-art compressors. (3) We develop an adaptive compression pipeline by SZ3 for APS data with minimal efforts, which leads to the best rate-distortion among all existing error-bounded lossy compressors for any bit-rate. (4) We compare the sustainability of SZ3 with leading error-bounded prediction-based compressors, and then demonstrate the necessity of diverse pipelines by integrating and evaluating several compression pipelines on diverse scientific datasets from multiple disciplines. Experiments show that SZ3 incurs very limited overhead in compressor integration and our customized compression pipelines lead to up to 20% improvement in compression ratios under the same data distortion, when compared with the best existing approach.

97 MATHEMATICS AND COMPUTING↗

Scalable Tensor Methods for Nonuniform Hypergraphs

While multilinear algebra appears natural for studying the multiway interactions modeled by hypergraphs, tensor methods for general hypergraphs have been stymied by theoretical and practical barriers. A recently proposed adjacency tensor is applicable to nonuniform hypergraphs, but is prohibitively costly to form and analyze in practice. We develop tensor times same vector (TTSV) algorithms for this tensor which improve complexity from $O(n^r)$ to a low-degree polynomial in $r$, where $n$ is the number of vertices and $r$ is the maximum hyperedge size. Our algorithms are implicit, avoiding formation of the order $r$ adjacency tensor. Here, we demonstrate the flexibility and utility of our approach in practice by developing tensor-based hypergraph centrality and clustering algorithms. We also show these tensor measures offer complementary information to analogous graph-reduction approaches on data, and are also able to detect higher-order structure that many existing matrix-based approaches provably cannot.

97 MATHEMATICS AND COMPUTING↗

Oxidation effects on spark plasma sintering of molybdenum nanopowders

Surface-oxidized molybdenum nanopowders are compacted by spark plasma sintering (SPS). The oxide impurity behavior is analyzed under various sintering temperatures. The densification mechanism of the nanopowders with a melted oxide phase is identified in situ by regression analysis of the experimental data on the temperature-dependent porosity change and on the SPS multistep pressure dilatometry. To increase the density of the compacted pellets, the nanopowders with the oxide phase are consolidated by SPS using the two in situ oxide removal methods: carbothermic reduction and particle surface cleaning by the electric current flow through the powders. Finally, the advantages and disadvantages of these methods in terms of the density, grain size, and mechanical properties of the final products are discussed.

36 MATERIALS SCIENCE↗

Predictive Engineering Tools for Injection-Molded Long-Carbon-Fiber Thermoplastic Composites - FY 2016 First Quarterly Report

During the first quarter of FY 2016, the following technical progress has been made toward project milestones: 1) Virginia Tech completed fiber orientation (FO) measurements for the samples taken at Locations A, B, C and D (Figure 1) from the 30wt% LCF/PP and 30wt% LCF/PA66 ribbed and non-ribbed complex parts using Virginia Tech’s established procedure. Virginia Tech delivered to PNNL all the measured fiber orientation data for validating ASMI fiber orientation predictions. 2)Virginia Tech performed fiber length distribution (FLD) measurements for the samples taken at Locations A, B, C and D from these complex parts using Virginia Tech’s established procedure. Virginia also re-assessed previous data and measured fiber length distributions in the corresponding nozzle purging materials and delivered to PNNL all the measured length data for validating ASMI fiber length predictions. 3)Based on measured fiber orientation data, Autodesk identified the parameters of the anisotropic rotary diffusion reduced strain closure (ARD-RSC) model [1] and provided PNNL with the values of these parameters that were used in ASMI analyses of the complex parts. 4) Magna provided Virginia Tech with additional samples cut out from the 30wt% LCF/PP and 30wt% LCF/PA66 ribbed parts (Figure 1) for fiber length and orientation measurements. 5) In discussion with Autodesk, PNNL performed 3D ASMI analyses of the 30wt% LCF/PP and 30wt% LCF/PA66 ribbed and non-ribbed complex parts to predict fiber orientations and length distributions in these parts. The issues observed through the analyses regarding fiber orientation distributions profiles and abnormal length distributions were reported to Autodesk. Autodesk is working to resolve these issues. 6) PNNL completed 3D ASMI analyses of the complex parts and compared predicted fiber orientation results at Locations A, B, and C on the non-ribbed parts, and at Locations A, B, C and D on the ribbed parts with the corresponding measured data. PNNL also evaluated the within-15%-agreement criterion using the principal tensile and flexural moduli computed based on predicted vs. measured fiber orientation results. 7) PNNL developed and discussed with Toyota, Magna and PlastiComp a method to perform weight and cost reduction for making the 30wt% LCF/PA66 ribbed part through comparative three-point bending simulations of this part and of similar parts in steel.

36 MATERIALS SCIENCE↗

Final Report, DOE Early Career Award: Predictive modeling of complex physical systems: new tools for statistical inference, uncertainty quantification, and experimental design

Predictive simulation of complex physical systems increasingly rests on the interplay of experimental observations with computational models. Key inputs, parameters, or structural aspects of models may be incomplete or unknown, and must be developed from indirect and limited observations. At the same time, quantified uncertainties are needed to qualify computational predictions in the support of design and decision-making. In this context, Bayesian statistics provides a foundation for inference from noisy and limited data, but at prohibitive computional expense. This project intends to make rigorous predictive modeling *feasible* in complex physical systems, via accelerated and scalable tools for uncertainty quantification, Bayesian inference, and experimental design. Specific objectives are as follows: 1. Develop adaptive posterior approximations and dimensionality reduction approaches for Bayesian inference in high-dimensional nonlinear systems. 2. Extend accelerated Bayesian methodologies to large-scale {\em sequential} data assimilation, fully treating nonlinear models and non-Gaussian state and parameter distributions. 3. Devise efficient surrogate-based methods for Bayesian model selection and the learning of model structure. 4. Develop scalable simulation/optimization approaches to nonlinear Bayesian experimental design, for both parameter inference and model selection. 5. Demonstrate these inferential tools on chemical kinetic models in reacting flow, constructing and refining thermochemical and electrochemical models from limited data. Demonstrate Bayesian filtering on canonical stochastic PDEs and in the dynamic estimation of inhomogeneous subsurface properties and flow fields.

97 MATHEMATICS AND COMPUTING↗

Performing an allreduce operation on a plurality of compute nodes of a parallel computer

Methods, apparatus, and products are disclosed for performing an allreduce operation on a plurality of compute nodes of a parallel computer, each node including at least two processing cores, that include: performing, for each node, a local reduction operation using allreduce contribution data for the cores of that node, yielding, for each node, a local reduction result for one or more representative cores for that node; establishing one or more logical rings among the nodes, each logical ring including only one of the representative cores from each node; performing, for each logical ring, a global allreduce operation using the local reduction result for the representative cores included in that logical ring, yielding a global allreduce result for each representative core included in that logical ring; and performing, for each node, a local broadcast operation using the global allreduce results for each representative core on that node.

97 MATHEMATICS AND COMPUTING↗

Clinical Results of Image-Guided Deep Inspiration Breath Hold Breast Irradiation

Purpose: To evaluate the feasibility, cardiac dose reduction, and the influence of the setup error on the delivered dose for fluoroscopy-guided deep inspiration breath hold (DIBH) irradiation using a cone-beam CT for irradiation of left-sided breast cancer patients. Methods and Materials: Nineteen patients treated according to the DIBH protocol were evaluated regarding dose to the ipsilateral breast (or thoracic wall), heart, (left ventricle [LV]and left anterior descending artery [LAD]), and lung. The DIBH treatment plan was compared to the free-breathing (FB) treatment planning and to the dose data in which setup error was taken into account (i.e., actual delivered dose). Results: The largest setup variability was observed in the direction perpendicular to the RT field ({mu} = -0.8 mm, {Sigma} = 2.9 mm, {sigma} = 2.0 mm). The mean (D{sub mean}) and maximum (D{sub max}) doses of the DIBH treatment plan was significantly lower compared with the FB treatment plan for the heart (34% and 25%, p < 0.001), LV (71% and 28%, p < 0.001), and LAD (52% and 39.8%, p < 0.001). For some patients, large differences were observed between the heart D{sub max} according to the DIBH treatment plan and the actual delivered dose (up to 71%), although D{sub max} was always smaller than the planned FB dose (mean group reduction = 29%, p < 0.001). Conclusion: The image-guided DIBH treatment protocol is a feasible irradiation method with small setup variability that significantly reduces the dose to the heart, LV, and LAD.

62 RADIOLOGY AND NUCLEAR MEDICINE↗