SEARCH · Engineering Papers
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.
Solvent-free quantifications of arsenic species in foods are attained by applying lasso regression to the analysis of X-ray absorption spectra and assigning uncertainties by bootstrapping.
The transport electron scattering due to micro-twin (MT) defects in InSb quantum wells (QWs) has been investigated at room temperature (RT). A linear-regression-based scattering analysis showed that Matthiessen's rule is applicable to the RT electron mobility in 20-nm-thick InSb QWs that contain MTs (whose density is 5.6x10{sup 2}-1.2x10{sup 4} /cm) and threading dislocations (8.7x10{sup 8}-3.2x10{sup 9} /cm{sup 2}) as dominant structural defects. For such an InSb QW whose local electron mobility in its non-MT regions is 2.8x10{sup 4}-4.5x10{sup 4} cm{sup 2}/(Vs), the MT-originated energy barrier against the electron transport is deduced to be 0.081-0.093 eV at RT.
We demonstrate that the space formed by the star formation rate (SFR), gas-phase metallicity (Z), and stellar mass (M {sub *}) can be reduced to a plane, as first proposed by Lara-Lopez et al. We study three different approaches to find the best representation of this 3D space, using a principal component analysis (PCA), a regression fit, and binning of the data. The PCA shows that this 3D space can be adequately represented in only two dimensions, i.e., a plane. We find that the plane that minimizes the {chi}{sup 2} for all variables, and hence provides the best representation of the data, corresponds to a regression fit to the stellar mass as a function of SFR and Z, M {sub *}= f(Z, SFR). We find that the distribution resulting from the median values in bins for our data gives the highest {chi}{sup 2}. We also show that the empirical calibrations to the oxygen abundance used to derive the Fundamental Metallicity Relation have important limitations, which contribute to the apparent inconsistencies. The main problem is that these empirical calibrations do not consider the ionization degree of the gas. Furthermore, the use of the N2 index to estimate oxygen abundances cannot be applied for 12 + log(O/H) {approx}> 8.8 because of the saturation of the [N II] {lambda}6584 line in the high-metallicity regime. Finally, we provide an update of the Fundamental Plane derived by Lara-Lopez et al.
We investigate whether magnetic interaction between close-in giant planets and their host stars produce observable statistical enhancements in stellar coronal or chromospheric activity. New Chandra observations of 12 nearby (d < 60 pc) planet-hosting solar analogs are combined with archival Chandra, XMM-Newton, and ROSAT coverage of 11 similar stars to construct a sample inoculated against inherent stellar class and planet-detection biases. Survival analysis and Bayesian regression methods (incorporating both measurements errors and X-ray upper limits; 13/23 stars have secure detections) are used to test whether ''hot Jupiter'' hosts are systematically more X-ray luminous than comparable stars with more distant or smaller planets. No significant correlations are present between common proxies for interaction strength (M {sub P}/a {sup 2} or 1/a) versus coronal activity (L {sub X} or L {sub X}/L {sub bol}). In contrast, a sample of 198 FGK main-sequence stars does show a significant (∼99% confidence) increase in X-ray luminosity with M {sub P}/a {sup 2}. While selection biases are incontrovertibly present within the main-sequence sample, we demonstrate that the effect is primarily driven by a handful of extreme hot-Jupiter systems with M {sub P}/a {sup 2} > 450 M {sub Jup} AU{sup –2}, which here are all X-ray luminous but to a degree commensurate with their Ca II H and K activity, in contrast to presented magnetic star-planet interaction scenarios that predict enhancements relatively larger in L {sub X}. We discuss these results in the context of cumulative tidal spin-up of stars hosting close-in gas giants (potentially followed by planetary infall and destruction). We also test our main-sequence sample for correlations between planetary properties and UV luminosity or Ca II H and K emission, and find no significant dependence.
Abstract We propose the combination of k-means clustering with Gaussian Process (GP) regression in the analysis and exploration of 4D angle-resolved photoemission spectroscopy (ARPES) data. Using cluster labels as the driving metric on which the GP is trained, this method allows us to reconstruct the experimental phase diagram from as low as 12% of the original dataset size. In addition to the phase diagram, the GP is able to reconstruct spectra in energy-momentum space from this minimal set of data points. These findings suggest that this methodology can be used to improve the efficiency of ARPES data collection strategies for unknown samples. The practical feasibility of implementing this technology at a synchrotron beamline and the overall efficiency implications of this method are discussed with a view on enabling the collection of more samples or rapid identification of regions of interest.
The Singular Value Decomposition (SVD) is one of the most important matrix factorizations, enjoying a wide variety of applications across numerous application domains. In statistics and data analysis, the common applications of SVD inclue Principal Components Analysis (PCA) and regression. Usually these applications arise on data that has far more rows than columns, so-called "tall/skinny" matrices. In the big data analytics context, this may take the form of hundreds of millions to billions of rows with only a few hundred columns. There is a need, therefore, for fast, accurate, and scalable tall/skinny SVD implementations which can fully utilize modern computing resources. To that end, we present a survey of three different algorithms for computing the SVD for these kinds of tall/skinny data layouts using MPI for communication. We contextualize these with common big data analytics techniques. Finally, we present both CPU and GPU timing results from the Summit supercomputer, and discuss possible alternative approaches.
Sepsis is a life-threatening medical condition that, if not treated promptly, can result in tissue damage, organ failure, and death. According to the Centers for Disease Control, about 270,000 individuals die of sepsis in the US each year. Further, sepsis expenditures accounted for 13% of total US hospital costs in 2013, totaling more than $24 billion. Our project objectives were to determine if Machine Learning algorithms could reliably predict hospital stay duration for patients with sepsis. The data set we used has been de-identified and is freely available through the BupaR package. The data includes 1050 cases, 15214 events, and 16 types of actions related to sepsis patient care. First, we used process mining to determine how long each patient was in the hospital. Using BupaR’s functions, we created several process model graphs. These process models depict the movement of patients at a hospital and provide duration data for each patent case. Second, we identified outlier data and created two dataset versions: one with and one without outliers. We then applied the following analysis methods: Linear Regression, Random Forest, K-Nearest Neighbors, Neural Networks, XGBoost, and lightGBM. We compared the model validations for the six machine learning models using the same data-splitting method. We found that the XGBoost model had the best prediction accuracy of 73.9 percent for cases with outliers, and 79 percent for cases without outliers. We also found that the lightGBM model had the lowest mean absolute error between prediction and actual duration in days with 3.66 days for the case with outliers, and 2.4 days for the case without outliers. These two models outperformed the other four models. This work will be enhanced in the future by exploring new prediction algorithms and comparing them with the results of this study.
HVAC units are currently one of the major resources providing demand response (DR) in residential buildings. Models of HVAC with DR function can improve understanding of its impact on power system operations and facilitate the deployment of DR technologies. This paper investigates the importance of various physical parameters and their distributions to the HVAC response to DR signals, which is a key step to the construction of HVAC models for a population of units with insufficient data. These parameters include the size of floors, insulation efficiency, the amount of solid mass in the house, and efficiency of the HVAC units. These parameters are usually assumed to follow Gaussian or Uniform distributions. We study the effect of uncertainty in the chosen parameter distributions on the aggregate HVAC response to DR signals, during transient phase and in steady state. We use a quasi-Monte Carlo sampling method with linear regression and Prony analysis to evaluate sensitivity of DR output to the uncertainty in the distribution parameters. The significance ranking on the uncertainty sources is given for future guidance in the modeling of HVAC demand response.
Detecting suspicious lesions in medical imaging is the important first step in computer-aided detection (CAD) systems. However, detecting abnormalities in breast tissue is difficult due to the lesion's varying size, shape, margin, and contrast with the background tissue. We focused on mass segmentation, a method that provides notable morphological features by outlining contours of masses. Accurate segmentation is crucial for correct diagnosis. Recent advancements in deep learning have improved object detection and segmentation, and these techniques are also being applied to medical imaging studies. We focused on U-net, which is a recently developed mass segmentation algorithm based on a fully convolutional network. The U-net architecture consists of (1) a contracting path to increase the resolution of the output and (2) a symmetric expanding path to better locate the region of interest. The performance of a U-net model was tested with 63 digital mammograms from INbreast, a publicly available database. We trained the model with images resized to 40x40 pixels and conducted 10-fold cross-validation to prevent overfitting. The model's performance with respect to breast density and the lesion's BI-RADS rating was also investigated. Dice coefficients (DC) were used as a performance measure to compare the predicted segmentation of the model with the ground truth. Logistic regression and an analysis of variance were performed to determine the significance of the DCs with regards to breast density and lesion behavior and to calculate the 95% confidence interval. The average DC was 0.80. The difference between DCs for BI-RADS 2 and 4c and for BI-RADS 2 and 5 were significant, suggesting that the model has more difficulty in segmenting benign lesions.
Purpose: Spinal cord tolerance for SBRT has been recommended for the maximum point dose level or at irradiated volumes such as 0.35 mL or 10% of contoured volumes. In this study, we investigated an inherent functional relationship that associates these dose surrogates for irradiated spinal cord volumes of up to 3.0 mL. Methods: A hidden variable termed as Effective Dose Radius (EDR) was formulated based on a dose fall-off model to correlate dose at irradiated spinal cord volumes ranging from 0 mL (point maximum) to 3.0 mL. A cohort of 15 spine SBRT cases was randomly selected to derive an EDR-parameterized formula. The mean prescription dose for the studied cases was 21.0±8.0 Gy (range, 10–40Gy) delivered in 3±1 fractions with target volumes of 39.1 ± 70.6 mL. Linear regression and variance analysis were performed for the fitting parameters of variable EDR values. Results: No direct correlation was found between the dose at maximum point and doses at variable spinal cord volumes. For example, Pearson R{sup 2} = 0.643 and R{sup 2}= 0.491 were obtained when correlating the point maximum dose with the spinal cord dose at 1 mL and 3 mL, respectively. However, near perfect correlation (R{sup 2} ≥0.99) was obtained when corresponding parameterized EDRs. Specifically, Pearson R{sup 2}= 0.996 and R{sup 2} = 0.990 were obtained when correlating EDR (maximum point dose) with EDR (dose at 1 mL) and EDR(dose at 3 mL), respectively. As a result, high confidence level look-up tables were established to correlate spinal cord doses at the maximum point to any finite irradiated volumes. Conclusion: An inherent functional relationship was demonstrated for spine SBRT. Such a relationship unifies dose surrogates at variable cord volumes and proves that a single dose surrogate (e.g. point maximum dose) is mathematically sufficient in constraining the overall spinal cord dose tolerance for SBRT.
Problem Solved: A novel technique for regression and sensitivity analysis of C1 function f problematic input dimension given observations (x_i, f(x_i), f'(x_i)). Solution Provided by the Code: Active Manifolds reduces the problem to analysis/regression of a 1 dimensional analogue by exploiting geometric properties of the function.
SAND2025-01851O GP-BayesOpInf is a software tool that uses algorithms to combine Gaussian process regression, principal component analysis, and linear Bayesian inference to produce a probabilistic reduced-order model for time-dependent systems. Numerical examples include the compressible Euler equations for an ideal gas, a heat diffusion process with a nonlinear reaction term, and a set of ordinary differential equations describing a compartmental model in epidemiology. Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525.
To accurately model flows with shock waves using staggered-grid Lagrangian hydrodynamics, artificial viscosity has to be introduced to convert kinetic energy into internal energy, thereby increasing the entropy across shocks. Determining the appropriate strength of the artificial viscosity is an art and strongly depends on the particular problem and experience of the researcher. The objective of this study is to pose the problem of finding the appropriate strength of artificial viscosity as an optimization problem and solve this problem using machine learning (ML) tools, specifically using surrogate models based on Gaussian Process regression and Bayesian analysis. We describe the optimization method and discuss various practical details of its implementation. The shock-containing problems for which we apply this method all have been implemented in the LANL code FLAG. First, we apply ML to find optimal values to isolated shock problems of different strengths. Second, we apply ML to optimize viscosity for a 1D propagating detonation problem based on Zel’dovich-von Neumann-Doring (ZND) detonation theory using a reactive burn model. We compare results for default (currently used values in FLAG) and optimized values of artificial viscosity for these problems demonstrating the potential for significant improvement in the accuracy of computations.
Typical QRAs provide deterministic estimates and understanding of risks posed but are constructed using significant assumptions and uncertainties due to limited data availability and historical momentum of using nominal estimates. This report presents a hydrogen QRA analysis using HyRAM+ that incorporates uncertainty with Latin hypercube sampling and sensitivity analysis using linear regression.
Cloud drop number concentrations (N d ) over the western North Atlantic Ocean (WNAO) are generally highest during the winter (DJF) and lowest in summer (JJA), in contrast to aerosol proxy variables (aerosoloptical depth, aerosol index, surface aerosol mass concentrations, surface cloud condensation nuclei (CCN) concentrations) that generally peak inspring (MAM) and JJA with minima in DJF. Using aircraft, satellite remote sensing, ground-based in situ measurement data, and reanalysis data, we characterize factors explaining the divergent seasonal cycles and furthermore probe into factors influencing N d on seasonal timescales. The results can be summarized well by features most pronounced in DJF, including features associated with cold-air outbreak (CAO) conditions such as enhanced values of CAO index, planetary boundary layer height (PBLH),low-level liquid cloud fraction, and cloud-top height, in addition to winds aligned with continental outflow. Data sorted into high- and low-N d days in each season, especially in DJF, revealed that all of these conditions were enhanced on the high-N d days, including reduced sea level pressure and stronger wind speeds. Although aerosols may be more abundant in MAM and JJA, the conditions needed to activate those particles into cloud droplets are weaker than in colder months, which is demonstrated by calculations of the strongest (weakest) aerosol indirect effects in DJF (JJA) based on comparing N d to perturbations in four different aerosol proxy variables (total and sulfate aerosol optical depth, aerosol index, surface mass concentration of sulfate). We used three machine learning models and up to 14 input variables to infer about most influential factors related to N d for DJF and JJA, with the best performance obtained with gradient-boosted regression tree (GBRT) analysis. The model results indicated that cloud fraction was the most important input variable, followed by some combination (depending on season) of CAO index and surface mass concentrations of sulfate and organic carbon. Future work is recommended to further understand aspects uncovered here such as impacts of free tropospheric aerosol entrainment on clouds, degree of boundary layer coupling, wet scavenging, and giant CCN effects on aerosol–N d relationships, updraft velocity, and vertical structure of cloud properties such as adiabaticity that impact the satellite estimation of N d .
A simulation model is applied to reproduce some observed brightness temperatures at a wavelength of 21 cm. The simulated results calculated with two different soil textures are compared directly with observations measured over fields in Arizona and South Dakota. It is found that good agreement is possible by properly adjusting the surface roughness parameter. Correlation analysis and linear regression of the brightness temperatures versus soil moistures are also carried out.
The contents of mobile trace elements (Co, Au, Sb, Ga, Se, Rb, Cs, Te, Bi, Ag, In, Tl, Zn, and Cd) in Antarctic and non-Antarctic populations of H4-6 and L4-6 chondrites, were compared using standard multivariate discriminant functions borrowed from linear discriminant analysis and logistic regression. A nonstandard randomization-simulation method was developed, making it possible to carry out probability assignments on a distribution-free basis. Compositional differences were found both between the Antarctic and non-Antarctic H4-6 chondrite populations and between two L4-6 chondrite populations. It is shown that, for various types of meteorites (in particular, for the H4-6 chondrites), the Antarctic/non-Antarctic compositional difference is due to preterrestrial differences in the genesis of their parent materials.
This article summarizes some of the methodology and results for studies of the earth's mantle and crust. Mantle conductivity studies can be made either by studying signals impressed on the earth from outside, e.g., the ionosphere or magnetosphere, or by studing signals originating in the core and transmitted through the mantle. Crustal field studies begin with a careful selection of the data and subsequent removal of core and external fields by some sort of filtering. Average maps from different local times sometimes differ, presumably due to the remaining presence of fields of external origin. Several techniques for further filtering are discussed. Where large-area aeromagnetic maps are available, crustal maps derived from satellite data can be compared with upward continued data. In general, the comparisons show agreement, with some differences, particularly in and near the auroral belts. The satellite data are further reduced by various methods of inverse and forward modeling, sometimes including reduction to the pole. These techniques are generally unstable at the equator. Common methods of stabilizing the inversions include principle components analysis and ridge regression. Because of the presence of the core field, the entire crustal contribution from the field is not known. Also, there is a basic nonuniqueness to the inverse solutions. Nevertheless, magnetizations that are interpretable can be derived.