Search NASA⌕ Search

SEARCH · Search NASA

Results for “distribution shift”

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 73 records · Page 4
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Enabling dynamic 3D coherent diffraction imaging via adaptive latent space tuning of generative autoencoders

Abstract Coherent diffraction imaging (CDI) is an advanced non-destructive 3D X-ray imaging technique for measuring a sample’s electron density. The main challenge of CDI is loss of phase information in diffraction intensity measurements, resulting in lengthy iterative reconstruction processes that can return non-unique solutions, which pose challenges for experiments attempting to track dynamic sample evolution through multiple states. As the increased brightness of fourth-generation light sources enables faster sample measurements and drives operando experiments with Bragg CDI, there is a growing need for faster reconstruction techniques that can keep pace. We have developed an adaptive generative autoencoder approach for uniquely tracking a sample’s electron density as it dynamically evolves. Our approach adaptively tunes the low-dimensional latent embedding of a generative autoencoder, enabling a computationally efficient manner to account for time-varying shifting distributions in real-time. Analytic proof of convergence is provided as well as numerical demonstration of sample tracking with noisy measurements.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2014

Quasiclassical trajectory study of the effect of antisymmetric stretch mode excitation on the O({sup 3}P) + CH{sub 4}(ν{sub 3} = 1) → OH + CH{sub 3} reaction on an analytical potential energy surface. Comparison with experiment

Motivated by a recent crossed-beam experiment on the title reaction reported by Pan and Liu [J. Chem. Phys. 140, 191101 (2014)], a detailed dynamics study was performed at three collision energies using quasiclassical trajectory (QCT) calculations based on a full-dimensional potential energy surface recently developed by our group (PES-2014). Although theory/experiment agreement is not yet quantitative, in general the theoretical results reproduce the experimental evidence: the vibrational branching ratio of OH(v = 1)/OH(v = 0) is ∼0.8/0.2, excitation of the antisymmetric CH stretching mode in methane increases reactivity by factor 2.28–1.50, although an equivalent amount as translational energy is more efficient in promoting the reaction and, finally, product angular distribution shifts from backward in the CH{sub 4}(ν = 0) ground-state to sideways when the antisymmetric CH stretching mode is excited. These results give confidence to the PES-2014 surface, depend on the quantization procedure used, are comparable with recent QCT calculations or improve previous theoretical studies using a different surface, and demonstrate the utility of the theory/experiment collaboration.

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

Effect of electrode spacing on the density distributions of electrons, ions, and metastable and radical molecules in SiH{sub 4}/NH{sub 3}/N{sub 2}/He capacitively coupled plasmas

Semiconductor fabrication often requires the deposition of hydrogenated silicon nitride (SiN{sub x}H{sub y}) film using SiH{sub 4}/NH{sub 3}/N{sub 2}/He capacitively coupled plasma (CCP) discharge. As analysis of the discharge geometry is essential to understanding CCP deposition, the effect of electrode spacing on the two-dimensional distributions of electrons, ions, and metastable and radical molecules was analyzed numerically using a fluid model. The simulation shows that the spatial variations in the ionization rates near the sheath become more obvious as the electrode spacing increases. In addition, as molecule-molecule gas-phase reactions are significantly affected by the local residence time, large electrode spacings are associated with significant volumetric losses for positive ions. Consequently, an increase of the electrode spacing leads axial density profiles of ions to change from bell shaped to double humped. However, NH{sub 4}{sup +} persistently maintains a bell-shaped axial density profile regardless of the degree of electrode spacing. We set the mole fraction of NH{sub 3} to only 1% of the total flow at the inlet, but NH{sub 4}{sup +} is the most abundant positive ion at the large electrode spacings. As the gas flow can transport the radicals around the space between the electrodes, we found that radical density distribution shifts toward the grounded electrode. The shift becomes pronounced as the electrode spacing increases. Finally, to validate our model, we compared the calculated deposition rate profile with the experimental data obtained along the wafer radius. According to our numerical results, the SiN{sub x}H{sub y} deposition rate decreases by approximately 16% when the electrode spacing increases from 9 to 20 mm.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2015

Simulation study of high-frequency energetic particle driven geodesic acoustic mode

High-frequency energetic particle driven geodesic acoustic modes (EGAM) observed in the large helical device plasmas are investigated using a hybrid simulation code for energetic particles and magnetohydrodynamics (MHD). Energetic particle inertia is incorporated in the MHD momentum equation for the simulation where the beam ion density is comparable to the bulk plasma density. Bump-on-tail type beam ion velocity distribution created by slowing down and charge exchange is considered. It is demonstrated that EGAMs have frequencies higher than the geodesic acoustic modes and the dependence on bulk plasma temperature is weak if (1) energetic particle density is comparable to the bulk plasma density and (2) charge exchange time (τ{sub cx}) is sufficiently shorter than the slowing down time (τ{sub s}) to create a bump-on-tail type distribution. The frequency of high-frequency EGAM rises as the energetic particle pressure increases under the condition of high energetic particle pressure. The frequency also increases as the energetic particle pitch angle distribution shifts to higher transit frequency. It is found that there are two kinds of particles resonant with EGAM: (1) trapped particles and (2) passing particles with transit frequency close to the mode frequency. The EGAMs investigated in this work are destabilized primarily by the passing particles whose transit frequencies are close to the EGAM frequency.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2013

Characteristics of surface photorefractive waves in a nonlinear SBN-75 crystal coated with a metal film

Based on the calculation of the electrostatic field potential of space charges, we have analysed the characteristic features of light-induced scattering of extraordinary polarised light in photorefractive (PR) crystals (for example, an SBN-75 crystal). Using the method of images, the electrostatic field is analysed for surface (aperiodic) waves along the crystal - dielectric (air) interface. It is shown that the field distributions satisfying the boundary conditions can emerge only upon accumulation of a screening electric charge in a narrow transition layer of thickness {approx}1 mm, the sign of the charge being opposite that of the space charge in the illuminated region of the crystal. A model is proposed to explain the observed features of the surface PR waves in a metal-film coated PR crystal. In considering the contact potential difference at the PR crystal - film interface it is shown that in the crystal layer (adjacent to the film) enriched with charge carriers, i.e., electrons, the refractive index can be significantly reduced. In the case of small excitation angles (0 - 1.5deg ), this layer can act as an optical barrier, the reflection from which can result in near-surface waves; a characteristic difference from the previously observed oscillatory surface waves is the presence of a broadened intensity distribution shifted inside the crystal. (nonlinear optical phenomena)

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2024

Nonstationarity in the global terrestrial water cycle and its interlinkages in the Anthropocene

Climate change and human activities alter the global freshwater cycle, causing nonstationary processes as its distribution shifting over time, yet a comprehensive understanding of these changes remains elusive. Here, we develop a remote sensing–informed terrestrial reanalysis and assess the nonstationarity of and interconnections among global water cycle components from 2003 to 2020. We highlight 20 hotspot regions where terrestrial water storage exhibits strong nonstationarity, impacting 35% of the global population and 45% of the area covered by irrigated agriculture. Emerging long-term trends dominate the most often (48.2%), followed by seasonal shifts (32.8%) and changes in extremes (19%). Notably, in mid-latitudes, this encompasses 34% of Asia and 27% of North America. The patterns of nonstationarity and their dominant types differ across other water cycle components, including precipitation, evapotranspiration, runoff, and gross primary production. These differences also manifest uniquely across hotspot regions, illustrating the intricate ways in which each component responds to climate change and human water management. Our findings emphasize the importance of considering nonstationarity when assessing water cycle information toward the development of strategies for sustainable water resource usage, enhancing resilience to extreme events, and effectively addressing other challenges associated with climate change.

Science & Technology - Other Topics↗
DOE OSTI2021

Geographic patterns of genomic diversity and structure in the C4 grass Panicum hallii across its natural distribution

Abstract Geographic patterns of within-species genomic diversity are shaped by evolutionary processes, life history and historical and contemporary factors. New genomic approaches can be used to infer the influence of such factors on the current distribution of infraspecific lineages. In this study, we evaluated the genomic and morphological diversity as well as the genetic structure of the C4 grass Panicum hallii across its complex natural distribution in North America. We sampled extensively across the natural range of P. hallii in Mexico and the USA to generate double-digestion restriction-associated DNA (ddRAD) sequence data for 423 individuals from 118 localities. We used these individuals to study the divergence between the two varieties of P. hallii, P. hallii var. filipes and P. hallii var. hallii as well as the genetic diversity and structure within these groups. We also examined the possibility of admixture in the geographically sympatric zone shared by both varieties, and assessed distribution shifts related with past climatic fluctuations. There is strong genetic and morphological divergence between the varieties and consistent genetic structure defining seven genetic clusters that follow major ecoregions across the range. South Texas constitutes a hotspot of genetic diversity with the co-occurrence of all genetic clusters and admixture between the two varieties. It is likely a recolonization and convergence point of populations that previously diverged in isolation during fragmentation events following glaciation periods.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2020

Using quasar X-ray and UV flux measurements to constrain cosmological model parameters

ABSTRACT Risaliti and Lusso have compiled X-ray and UV flux measurements of 1598 quasars (QSOs) in the redshift range 0.036 ≤ z ≤ 5.1003, part of which, z ∼ 2.4 − 5.1, is largely cosmologically unprobed. In this paper we use these QSO measurements, alone and in conjunction with baryon acoustic oscillation (BAO) and Hubble parameter [H(z)] measurements, to constrain cosmological parameters in six different cosmological models, each with two different Hubble constant priors. In most of these models, given the larger uncertainties, the QSO cosmological parameter constraints are mostly consistent with those from the BAO + H(z) data. A somewhat significant exception is the non-relativistic matter density parameter Ωm0 where QSO data favour Ωm0 ∼ 0.5 − 0.6 in most models. As a result, in joint analyses of QSO data with H(z) + BAO data the 1D Ωm0 distributions shift slightly towards larger values. A joint analysis of the QSO + BAO + H(z) data is consistent with the current standard model, spatially-flat ΛCDM, but mildly favours closed spatial hypersurfaces and dynamical dark energy. Since the higher Ωm0 values favoured by QSO data appear to be associated with the z ∼ 2 − 5 part of these data, and conflict somewhat with strong indications for Ωm0 ∼ 0.3 from most z < 2.5 data as well as from the cosmic microwave background anisotropy data at z ∼ 1100, in most models, the larger QSO data Ωm0 is possibly more indicative of an issue with the z ∼ 2 − 5 QSO data than of an inadequacy of the standard flat ΛCDM model.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2025

Physics-constrained superresolution diffusion for six-dimensional phase space diagnostics

Adaptive physics-constrained superresolution diffusion is developed for noninvasive virtual diagnostics of the six-dimensional (6D) phase space density of charged particle beams. An adaptive variational autoencoder embeds initial beam condition images and scalar measurements to a low-dimensional latent space from which a 32 6 pixel 6D tensor representation of the beam's 6D phase space density is generated. Projecting from a 6D tensor generates physically consistent two-dimensional projections. Physics-guided superresolution diffusion transforms low-resolution images of the 6D density to high resolution 256 × 256 pixel images. Unsupervised adaptive latent space tuning enables tracking of time-varying beams without knowledge of time-varying initial conditions. The method is demonstrated with experimental data and multiparticle simulations at the HiRES UED. The general approach is applicable to a wide range of complex dynamic systems evolving in high-dimensional phase space. The method is shown to be robust to distribution shift without retraining. Published by the American Physical Society 2025

43 PARTICLE ACCELERATORS↗
DOE OSTI2025

Enhancing synchrotron radiation micro-CT images using deep learning: an application of Noise2Inverse on bone imaging

In bone-imaging research, in situ synchrotron radiation micro-computed tomography (SRµCT) mechanical tests are used to investigate the mechanical properties of bone in relation to its microstructure. Low-dose computed tomography (CT) is used to preserve bone's mechanical properties from radiation damage, though it increases noise. To reduce this noise, the self-supervised deep learning method Noise2Inverse was used on low-dose SRµCT images where segmentation using traditional thresholding techniques was not possible. Simulated-dose datasets were created by sampling projection data at full, one-half, one-third, one-fourth and one-sixth frequencies of an in situ SRµCT mechanical test. After convolutional neural networks were trained, Noise2Inverse performance on all dose simulations was assessed visually and by analyzing bone microstructural features. Visually, high image quality was recovered for each simulated dose. Lacunae volume, lacunae aspect ratio and mineralization distributions shifted slightly in full, one-half and one-third dose network results, but were distorted in one-fourth and one-sixth dose network results. Following this, new models were trained using a larger dataset to determine differences between full dose and one-third dose simulations. Significant changes were found for all parameters of bone microstructure, indicating that a separate validation scan may be necessary to apply this technique for microstructure quantification. Noise present during data acquisition from the testing setup was determined to be the primary source of concern for Noise2Inverse viability. While these limitations exist, incorporating dose calculations and optimal imaging parameters enables self-supervised deep learning methods such as Noise2Inverse to be integrated into existing experiments to decrease radiation dose.

Obata, Yoshihiro (ORCID:0000000303659129)↗
DOE OSTI2024

Counter Data Paucity through Adversarial Invariance Encoding: A Case Study on Modeling Battery Thermal Runaway

Lithium-ion batteries, widely used for their durability and high energy storage, face the risk of internal short circuits leading to catastrophic thermal runaway events. These events, triggered by external stimuli like mechanical loads, pose safety concerns in applications such as electric vehicles. Detecting and understanding thermal runaway events is crucial, but physics-driven models struggle to explain the non-linear evolution of battery temperature during these events, considering factors like material composition and state-of-charge. Due to the rarity of these events and the cost of data collection, we propose a deep learning (DL) model to predict battery temperature responses during thermal runaway. The challenge lies in the scarcity of data, making traditional DL models prone to overfitting and learning low-quality representations of the complex process.Our approach introduces a novel few-shot architecture that incorporates an adversarially governed invariant encoding process. This architecture aims to distill "invariant" relationships by addressing distributional shifts in data across various battery properties, facilitating the detection of thermal runaway events. Specifically, our results demonstrate that deep learning models conditioned on these "invariant" representations outperform state-of-the-art baselines, achieving a remarkable 96.8% performance improvement in terms of the popular metric MAPE. This framework presents a promising direction for enhancing battery safety modeling, particularly in the context of rare and complex events like thermal runaway. Our code and code and dataset used for the paper are public1.

Tabassum, Anika [ORNL] (ORCID:0000000254600955)↗
DOE OSTI2019

Large Scale Unsupervised Domain Adaptation of Segmentation Networks with Adversarial Learning

Most current state-of-the-art methods for semantic segmentation on remote sensing imagery require large labeled data, which is scarcely available. Due to the distribution shifting phenomenon inherent in remote sensing imagery, the reuse of pre-trained models on new areas of interest rarely yield satisfactory results. In this paper, we approach this problem from an adversarial learning perspective toward unsupervised domain adaptation. The core concept is to infuse fully convolutional neural networks and adversarial networks for semantic segmentation assuming the structures in the scene and objects of interest are similar in two set of images. Models are trained on a source dataset where ground truth is available and adapted to new target dataset iteratively via a adversarial loss on unlabeled samples. We use two real large scale datasets to validate the framework: 1) cross city road extraction and 2) cross country building extraction. The preliminary results show the usefulness of considering adversarial learning for indirect re-use of the pre-trained models. Experimental validation suggests significant benefits over models without adaptation.

Deng, Xueqing↗
DOE OSTI2023

Neural networks for large eddy simulations of wall-bounded turbulence: numerical experiments and challenges

Here, we examine the application of neural network-based methods to improve the accuracy of large eddy simulations of incompressible turbulent flows. The networks are trained to learn a mapping between flow features and the subgrid scales, and applied locally and instantaneously—in the same way as traditional physics-based subgrid closures. Models that use only the local resolved strain rate are poorly correlated with the actual subgrid forces obtained from filtering direct numerical simulation data. We see that highly accurate models in a priori testing are inaccurate in forward calculations, owing to the preponderance of numerical errors in implicitly filtered large eddy simulations. A network that accounts for the discretization errors is trained and found to be unstable in a posteriori testing. We identify a number of challenges that the approach faces, including a distribution shift that affects networks that fail to account for numerical errors.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2012

Statistical Simulation of the Performance and Degradation of a PEMFC Membrane Electrode Assembly

A 1-D MEA Performance model was developed that considered transport of liquid water, agglomerate catalyst structure, and the statistical variation of the MEA characteristic parameters. The model was validated against a low surface area carbon supported catalyst across various platinum loadings and operational conditions. The statistical variation was found to play a significant role in creating noise in the validation data and that there was a coupling effect between movement in material properties with liquid water transport. Further, in studying the low platinum loaded catalyst layers it was found that liquid water played a significant role in the increasing the overall transport losses. The model was then further applied to study platinum dissolution via potential cycling accelerated stress tests, in which the platinum was found to dissolve nearest the membrane effectively resulting in reaction distribution shifts within the layer.

08 HYDROGEN↗
DOE OSTI2024

Combined speckle- and propagation-based single shot two-dimensional phase retrieval method

Single-shot two-dimensional (2D) phase retrieval (PR) can recover the phase shift distribution within an object from a single 2D x-ray phase contrast image (XPCI). Two competing XPCI imaging modalities often used for single-shot 2D PR to recover material properties critical for predictive performance capabilities are: speckle-based (SP-XPCI) and propagation-based (PB-XPCI) XPCI imaging. However, PR from SP-XPCI and PB-XPCI images are, respectively, limited to reconstructing accurately slowly and rapidly varying features due to noise and differences in their contrast mechanisms. Herein, we consider a combined speckle- and propagation-based XPCI (SPB-XPCI) image by introducing a mask to generate a reference pattern and imaging in the near-to-holographic regime to induce intensity modulations in the image. We develop a single-shot 2D PR method for SPB-XPCI images of pure phase objects without imposing restrictions such as object support constraints. It is compared against PR methods inspired by those developed for SP-XPCI and PB-XPCI on simulated and experimental images of a thin glass shell before and during shockwave compression. Reconstructed phase maps show improvements in quantitative scores of root-mean-square error and structural similarity index measure using our proposed method.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2024

How to understand limitations of generative networks

Well-trained classifiers and their complete weight distributions provide us with a well-motivated and practicable method to test generative networks in particle physics. We illustrate their benefits for distribution-shifted jets, calorimeter showers, and reconstruction-level events. In all cases, the classifier weights make for a powerful test of the generative network, identify potential problems in the density estimation, relate them to the underlying physics, and tie in with a comprehensive precision and uncertainty treatment for generative networks.

Physics↗
DOE OSTI2015

Fuel Cell Powered Lift Truck

This project, entitled “Recovery Act: Fuel Cell-Powered Lift Truck Sysco (Houston) Fleet Deployment”, was in response to DOE funding opportunity announcement DE-PS36-08GO98009, Topic 7B, which promotes the deployment of fuel cell powered material handling equipment in large, multi-shift distribution centers. This project promoted large-volume commercialdeployments and helped to create a market pull for material handling equipment (MHE) powered fuel cell systems. Specific outcomes and benefits involved the proliferation of fuel cell systems in 5-to 20-kW lift trucks at a high-profile, real-world site that demonstrated the benefits of fuel cell technology and served as a focal point for other nascent customers. The project allowed for the creation of expertise in providing service and support for MHE fuel cell powered systems, growth of existing product manufacturing expertise, and promoted existing fuel cell system and component companies. The project also stimulated other MHE fleet conversions helping to speed the adoption of fuel cell systems and hydrogen fueling technology. This document also contains the lessons learned during the project in order to communicate the successes and difficulties experienced, which could potentially assist others planning similar projects.

30 DIRECT ENERGY CONVERSION↗
DOE OSTI2025

Helium Bubble Physics Experiments

New results from MST-16 indicate that as plutonium ages at close to ambient temperature and helium ingrows, the number density of bubbles increases only slowly, if at all, and their size distribution shifts towards larger diameters. Helium-containing aged plutonium foams when annealed at high temperatures. We attribute these phenomena to a diffusivity large enough that, after a few years of aging, new decay helium diffuses to extant bubbles rather than nucleating new bubbles, and that during annealing larger bubbles grow at the expense of smaller ones (Ostwald ripening). Bulk swelling results from the fact that the Laplace pressure and therefore the helium density are less in larger bubbles. We develop a simple theory and propose to study these phenomena in aluminum alloyed with 10 B in which successive exposure to reactor thermal neutrons produces a stepwise increase in helium content.

36 MATERIALS SCIENCE↗