Search NASA⌕ Search

SEARCH · Search NASA

Results for “Model selection”

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 1,531 records · Page 85

CASP11 – An Evaluation of a Modular BCL::Fold-Based Protein Structure Prediction Pipeline

In silico prediction of a protein’s tertiary structure remains an unsolved problem. The community-wide Critical Assessment of Protein Structure Prediction (CASP) experiment provides a double-blind study to evaluate improvements in protein structure prediction algorithms. We developed a protein structure prediction pipeline employing a three-stage approach, consisting of low-resolution topology search, high-resolution refinement, and molecular dynamics simulation to predict the tertiary structure of proteins from the primary structure alone or including distance restraints either from predicted residue-residue contacts, nuclear magnetic resonance (NMR) nuclear overhauser effect (NOE) experiments, or mass spectroscopy (MS) cross-linking (XL) data. The protein structure prediction pipeline was evaluated in the CASP11 experiment on twenty regular protein targets as well as thirty-three ‘assisted’ protein targets, which also had distance restraints available. Although the low-resolution topology search module was able to sample models with a global distance test total score (GDT_TS) value greater than 30% for twelve out of twenty proteins, frequently it was not possible to select the most accurate models for refinement, resulting in a general decay of model quality over the course of the prediction pipeline. In this study, we provide a detailed overall analysis, study one target protein in more detail as it travels through the protein structure prediction pipeline, and evaluate the impact of limited experimental data.

59 BASIC BIOLOGICAL SCIENCES↗

The MicroBooNE Search for Single Photon Events

This note describes and presents preliminary results for the MicroBooNE analysis being developed to search for neutrino-induced neutral current single photon events in the MicroBooNE detector. The analysis currently assumes neutrino neutral current Δ production followed by Δ radiative decay on argon (NC Δ → N γ) as the “signal model”; event reconstruction and selection have been developed and are being optimized in order to reduce cosmogenic and other beam-related backgrounds explicit to the NC Δ → N γ process. This note presents the current status of event reconstruction, selection, optimization and validation checks for the early stages of analysis on external (cosmogenic background) events. Also presented are the projected sensitivities for testing the standard model predicted rate for the NC Δ → N γ process and for testing the interpretation of the previously observed MiniBooNE low energy excess as NC Δ → N γ events.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Phase-field investigation on the non-equilibrium interface dynamics of rapid alloy solidification

The research program reported here is focused on critical issues that represent conspicuous gaps in current understanding of rapid solidification, limiting our ability to predict and control microstructural evolution (i.e. morphological dynamics and microsegregation) at high undercooling, where conditions depart significantly from local equilibrium. More specifically, through careful application of phase-field modeling, using appropriate thin-interface and anti-trapping corrections and addressing important details such as transient effects and a velocity-dependent (i.e. adaptive) numerics, the current analysis provides a reasonable simulation-based picture of non-equilibrium solute partitioning and the corresponding oscillatory dynamics associated with single-phase rapid solidification and show that this method is a suitable means for a self-consistent simulation of transient behavior and operating point selection under rapid growth conditions. Moving beyond the limitations of conventional theoretical/analytical treatments of non-equilibrium solute partitioning, these results serve to substantiate recent experimental findings and analytical treatments for single-phase rapid solidification. The departure from the equilibrium solid concentration at the solid-liquid interface was often observed during rapid solidification, and the energetic associated non-equilibrium solute partitioning has been treated in detail, providing possible ranges of interface concentrations for a given growth condition. Use of these treatments for analytical description of specific single-phase dendritic and cellular operating point selection, however, requires a model for solute partitioning under a given set of growth conditions. Therefore, analytical solute trapping models which describe the chemical partitioning as a function of steady state interface velocities have been developed and widely utilized in most of the theoretical investigations of rapid solidification. However, these solute trapping models are not rigorously verified due to the difficulty in experimentally measuring under rapid growth conditions. Moreover, since these solute trapping models include kinetic parameters which are difficult to directly measure from experiments, application of the solute trapping models or the associated analytic rapid solidification model is limited. These theoretical models for steady state rapid solidification which incorporate the solute trapping models do not describe the interdependency of solute diffusion, interface kinetics, and alloy thermodynamics. The phase-field approach allows calculating, spontaneously, the non-equilibrium growth effects of alloys and the associated time-dependent growth dynamics, without making the assumptions that solute partitioning is an explicit function of velocity, as is the current convention. In the research described here, by utilizing the phase-field model in the thin-interface limit, incorporating the anti-trapping current term, more quantitatively valid interface kinetics and solute diffusion across the interface are calculated. In order to sufficiently resolve the physical length scales (i.e. interface thickness and diffusion boundary length), grid spacings are continually adjusted in calculations. The full trajectories of transient planar growth dynamics under rapid directional solidification conditions with different pulling velocities are described. As a validation of a model, the predicted steady state conditions are consistent with the analytic approach for rapid growth. It was confirmed that rapid interface dynamics exhibits the abrupt acceleration of the planar front when the effect of the non-equilibrium solute partitioning at the interface becomes signi ficant. This is consistent with the previous linear stability analysis for the non-equilibrium interface dynamics. With an appropriate growth condition, the continuous oscillation dynamics was able to be simulated using continually adjusting grid spacings. This oscillatory dynamics including instantaneous jump of interface velocities are consistent with a previous phenomenological model by and a numerical investigation, which may cause the formation of banded structures. Additionally, the selection of the steady state growth dynamics in the highly undercooled melt is demonstrated. The transition of the growth morphology, interface velocity selection, and solute trapping phenomenon with increasing melt supersaturations was described by the phase-field simulation. The tip selection for the dendritic growth was consistent with Ivantsov's function, and the non-equilibrium chemical partitioning behavior shows good qualitative agreement with the Aziz's solute trapping model even though the model parameter(V D ) remains as an arbitrary constant. This work is able to show the possibility of comprehensive description of rapid alloy growth over the entire time-dependent non-equilibrium phenomenon.

36 MATERIALS SCIENCE↗

Exploring the dependence of gas cooling and heating functions on the incident radiation field with machine learning

ABSTRACT Gas cooling and heating functions play a crucial role in galaxy formation. But, it is computationally expensive to exactly compute these functions in the presence of an incident radiation field. These computations can be greatly sped up by using interpolation tables of pre-computed values, at the expense of making significant and sometimes even unjustified approximations. Here, we explore the capacity of machine learning to approximate cooling and heating functions with a generalized radiation field. Specifically, we use the machine learning algorithm XGBoost to predict cooling and heating functions calculated with the photoionization code cloudy at fixed metallicity, using different combinations of photoionization rates as features. We perform a constrained quadratic fit in metallicity to enable a fair comparison with traditional interpolation methods at arbitrary metallicity. We consider the relative importance of various photoionization rates through both a principal component analysis (PCA) and calculation of SHapley Additive exPlanation (shap) values for our XGBoost models. We use feature importance information to select different subsets of rates to use in model training. Our XGBoost models outperform a traditional interpolation approach at each fixed metallicity, regardless of feature selection. At arbitrary metallicity, we are able to reduce the frequency of the largest cooling and heating function errors compared to an interpolation table. We find that the primary bottleneck to increasing accuracy lies in accurately capturing the metallicity dependence. This study demonstrates the potential of machine learning methods such as XGBoost to capture the non-linear behaviour of cooling and heating functions.

79 ASTRONOMY AND ASTROPHYSICS↗

MINE: maximally informative next experiment—toward a new GWAS experimental design and methodology

Abstract The computational methodology of Genome Wide Association Studies (GWAS) currently has several limitations: (i) the number of observations (rows) on a quantitative trait tends to be smaller than the number of single nucleotide polymorphisms (SNPs) (columns) in the design matrix; (ii) each SNP is usually modeled separately, failing to acknowledge interaction between each other (ie epistasis); (iii) there is implicit linkage disequilibrium (LD) between neighboring SNPs due to their linkage. To overcome these issues, we developed a tool that uses ensemble methods to fit mixed linear models to GWAS data, and these ensemble methods include the development of a new experimental design approach in GWAS, which uses the resultant models and data to select the next informative experiment over time. This new adaptive and staged approach for GWAS experimental design was developed and tested in a 3 yr adaptive model-guided discovery experiment against a fixed classical design. In Sorghum bicolor a total of 79, 86, and 78 accessions were tested in years 1, 2, and 3, respectively out of 343 accessions available in the Bioenergy Association Panel (BAP) each identified for 232,303 SNPs, 1 every 2–3 kb in the genomes. We demonstrated the feasibility of MINE enacted with 8 people in the field per year over 3 yr vs in 1 large classical design enacted with 20 people in 1 yr. The MINE results for chromosomal regions identified controlling dry weight were confirmed against results from previous sorghum GWAS experiments and 1 large classical design for the BAP panel.

Genetics & Heredity↗

Identification Uncertainty in Inverse Material Model Parameter Determination: A Sensitivity‐Based Decision Process for Load Path Selection

This research proposes a sensitivity-based framework for selecting the optimal prescribed loading path for a biaxial cruciform specimen. Optimality here is determined by the direction and magnitude of the prescribed displacement that minimizes the influence of random noise on the material model parameter identification. Using simulated experimental data based on finite element simulation, in this work, we identify the material model parameters of a Ludwik hardening model and plane stress implementation of the Hill-48 yield criterion using finite element model updating (FEMU). Our analysis reveals that the identification (or estimator) uncertainty of model parameters depends on the displacement boundary conditions (i.e., loading sequence) and the ground-truth value of the individual parameters. Optimal experimental design (OED) criteria based on the Fisher information matrix were investigated to mitigate indecision in the choice of optimal load path when the identification uncertainty of different material model parameters optimized at different load paths. The determinant of the Fisher information matrix was chosen here as the more useful metric due to its ability to capture uncertainty of the most influential material model parameters. The proposed framework demonstrates potential for real-time automated load step selection using scalar criteria derived prior to mechanical loading. The framework can be generalized to other geometries, boundary conditions and material models, allowing this procedure to be utilized for different experimental configurations and materials.

Fayad, Samuel S. [University of Illinois at Urbana↗

ML Pipeline (Machine Learning Pipeline) [SWR-22-35]

The Machine Learning Pipeline (ML Pipeline) allows the user to fit a model to predict a dependent variable, y, based on a feature set, X. ML Pipeline gives the ability to automatically generate features with its Feature Engineering function, automatically select the most important features with its Feature Selection function, fit the model with its Fit function, and evaluate the model with its Evaluate function. ML Pipeline's functionality can not only help the user predict values for the desired variable it also gives the user a better understanding of which features are important and how effective the model is. ML Pipeline accomplishes all this, while remaining simple to use and interpret.

Schiek, Andrew↗

Chandra and NuSTAR Follow-up Observations of Swift-BAT-selected AGNs

Based on current models of the cosmic X-ray background (CXB), heavily obscured active galactic nuclei (AGNs) are expected to make up ∼10% of the peak emission of the CXB and ∼20% of the total population of AGNs, yet few of these sources have been recorded and characterized in current surveys. Here we present the Chandra follow-up observation of 14 AGNs detected by Swift-BAT. For five sources in the sample, NuSTAR observations in the 3–80 keV band are also available. The X-ray spectral fitting over the 0.3–150 keV energy range allows us to determine the main X-ray spectral parameters, such as the photon index and the intrinsic absorption, of these objects and to make hypotheses on the physical structures responsible for the observed spectra. We find that 13 of the 14 objects are absorbed AGNs, and one is a candidate Compton-thick AGN, having intrinsic absorption N{sub H}>10{sup 24} cm{sup −2}. Finally, we verified that the use of NuSTAR observations is strategic to strongly constrain the properties of obscured AGNs, since the best-fit values we obtained for parameters such as the power-law photon index Γ and the intrinsic absorption N{sub H} changed sometimes significantly fitting the spectra with and without the use of NuSTAR data.

79 ASTRONOMY AND ASTROPHYSICS↗

Dark Energy Survey: Modeling strategy for multiprobe cluster cosmology and validation for the Full Six-year Dataset

We introduce an updated To&Krause2021 model for joint analyses of cluster abundances and large-scale two-point correlations of weak lensing and galaxy and cluster clustering (termed CL+3x2pt analysis) and validate that this model meets the systematic accuracy requirements of analyses with the statistical precision of the final Dark Energy Survey (DES) Year 6 (Y6) dataset. The validation program consists of two distinct approaches, (1) identification of modeling and parameterization choices and impact studies using simulated analyses with each possible model misspecification (2) end-to-end validation using mock catalogs from customized Cardinal simulations that incorporate realistic galaxy populations and DES-Y6-specific galaxy and cluster selection and photometric redshift modeling, which are the key observational systematics. In combination, these validation tests indicate that the model presented here meets the accuracy requirements of DES-Y6 for CL+3x2pt based on a large list of tests for known systematics. In addition, we also validate that the model is sufficient for several other data combinations: the CL+GC subset of this data vector (excluding galaxy--galaxy lensing and cosmic shear two-point statistics) and the CL+3x2pt+BAO+SN (combination of CL+3x2pt with the previously published Y6 DES baryonic acoustic oscillation and Y5 supernovae data).

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Adaptive cache management based on programming model information

A processing system selectively allocates space to store a group of one or more cache lines at a cache level of a cache hierarchy having a plurality of cache levels based on memory access patterns of a software application executing at the processing system. The processing system generates bit vectors indicating which cache levels are to allocate space to store groups of one or more cache lines based on the memory access patterns, which are derived from data granularity and movement information. Based on the bit vectors, the processing system provides hints to the cache hierarchy indicating the lowest cache level that can exploit the reuse potential for a particular data.

Na, Weon Taek↗

Field validation of isotropic analytical models for simulating fabric shades

Fabric roller shades are common shading materials used in commercial and residential buildings. Accurately characterizing and modeling shades helps practitioners select the appropriate product and its control strategy based on climate and occupants' priorities, such as visual comfort and view to outdoors. Previous studies established a generalized method for modeling complex fenestration systems using data-driven tabulated bidirectional scattering distribution functions. However, deploying such a method at scale to all fabric shading products on the market is too costly and time-consuming. Analytical models that are based on a limited set of measurements (e.g., normal-normal and normal-hemispherical visible transmittance and reflectance, and directional cut-off angles) can be used to model the wide variety of shading products on the market. This study evaluates the performance of two isotropic analytical models, Roos-Wienold and Modified-Kotey, for modeling fabric roller shades, with a focus on the model's ability to predict occupant visual comfort. The performance evaluation was conducted through laboratory and field measurements and simulations. The results showed that both models are sufficient for predicting vertical illuminance at seated eye-level. Roos-Wienold model was able to predict binary visual comfort classification (glare/no-glare) under a wide range of luminance conditions, while Modified-Kotey model did not perform as well under high-contrast low-adaptation conditions. Both models are insufficient in predicting visual comfort at a four-point scale (e.g., imperceptible, perceptible, disturbing, intolerable). The two isotropic models become less accurate when the fabric exhibits high anisotropy.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Dark energy survey: Modeling strategy for multiprobe cluster cosmology and validation for the full six-year dataset

Here, we introduce an updated To&Krause2021 model for joint analyses of cluster abundances and large-scale two-point correlations of weak lensing and galaxy and cluster clustering (termed CL+3×2 pt analysis) and validate that this model meets the systematic accuracy requirements of analyses with the statistical precision of the final Dark Energy Survey (DES) Year 6 (Y6) dataset. The validation program consists of two distinct approaches, (i) identification of modeling and parametrization choices and impact studies using simulated analyses with each possible model misspecification and (ii) end-to-end validation using mock catalogs from customized Cardinal simulations that incorporate realistic galaxy populations and DES-Y6-specific galaxy and cluster selection and photometric redshift modeling, which are the key observational systematics. In combination, these validation tests indicate that the model presented here meets the accuracy requirements of DES-Y6 for CL+3×2 pt based on a large list of tests for known systematics. In addition, we also validate that the model is sufficient for several other data combinations: the CL+GC subset of this data vector (excluding galaxy–galaxy lensing and cosmic shear two-point statistics) and the CL+3×2 pt+BAO+SN (combination of CL+3×2 pt with the previously published Y6 DES baryonic acoustic oscillation and Y5 supernovae data).

79 ASTRONOMY AND ASTROPHYSICS↗

Collaborative Project: The problem of bias in defining uncertainty in computationally enabled strategies for data-driven climate model development

The objective of the project is to develop strategies for better representing scientific sensibilities within statistical measures of model skill that then can be used within a Bayesian statistical framework for data-driven climate model development and improved measures of model scientific uncertainty. One of the thorny issues in model evaluation is quantifying the effect of biases on climate projections. While any bias is not desirable, only those biases that affect feedbacks affect scatter in climate projections. The effort at the University of Texas is to analyze previously calculated ensembles of CAM3.1 with perturbed parameters to discover how biases affect projections of global warming. The hypothesis is that compensating errors in the control model can be identified by their effect on a combination of processes and that developing metrics that are sensitive to dependencies among state variables would provide a way to select version of climate models that may reduce scatter in climate projections. Gabriel Huerta was responsible for leading the effort to develop statistical methods for evaluating these field dependencies. The UT effort will incorporate these developments into MECS, which is a set of python scripts being developed at the University of Texas for managing the workflow associated with data-driven climate model development over HPC resources. We regularly consult with Peter Gleckler (LLNL), Mark Taylor (SNL), and Rich Neale (NCAR) to provide additional insight into issues concerning CAM5 model development and multi-model evaluation of the AR5 archive. This grant was originally assigned to Indiana University where both the PI (Huerta) and the Postdoc (Nosedal) worked on the project from March 1, 2012-March 31, 2013.

97 MATHEMATICS AND COMPUTING↗

Collaborative Project: The problem of bias in defining uncertainty in computationally enabled strategies for data-driven climate model development. Final Technical Report.

The objective of the project is to develop strategies for better representing scientific sensibilities within statistical measures of model skill that then can be used within a Bayesian statistical framework for data-driven climate model development and improved measures of model scientific uncertainty. One of the thorny issues in model evaluation is quantifying the effect of biases on climate projections. While any bias is not desirable, only those biases that affect feedbacks affect scatter in climate projections. The effort at the University of Texas is to analyze previously calculated ensembles of CAM3.1 with perturbed parameters to discover how biases affect projections of global warming. The hypothesis is that compensating errors in the control model can be identified by their effect on a combination of processes and that developing metrics that are sensitive to dependencies among state variables would provide a way to select version of climate models that may reduce scatter in climate projections. Gabriel Huerta at the University of New Mexico is responsible for developing statistical methods for evaluating these field dependencies. The UT effort will incorporate these developments into MECS, which is a set of python scripts being developed at the University of Texas for managing the workflow associated with data-driven climate model development over HPC resources. This report reflects the main activities at the University of New Mexico where the PI (Huerta) and the Postdocs (Nosedal, Hattab and Karki) worked on the project.

58 GEOSCIENCES↗

Holistic Fuel Cell Electric Vehicle/Hydrogen Station Optimization Model (CRADA CRD-18-00745 Final Report)

The National Renewable Energy Lab (NREL), Argonne National Lab (ANL), Sandia National Lab (SNL), and Frontier Energy, Inc comprise a team that will perform a review of currently available models that can simulate both a hydrogen station and a fuel cell electric vehicle (FCEV), and whose owners are willing to make them available freely. We will then down select to a single model or model pair, secure agreements for their free use, and validate the resulting model or coupled models. For example, we may evaluate an existing 1-D hydrogen fueling model developed by Kyushu University for the Japanese New Energy and Industrial Technology Development Organization (NEDO). The validation will take place using data from various sources, potentially including the European Joint Research Center (JRC) HyTransfer project, SAE testing performed by Powertech to validate SAE J2601, and whole-station validation using the NREL’s Hydrogen Infrastructure Testing and Research Facility (HITRF). The team will make the model and all the validation data open to the public at the end of the project.

1-D hydrogen fueling model↗

Big Adaptive Rotor Phase I (Final Report)

The Big Adaptive Rotor (BAR) project was initiate by DOE in 2018 with the goal of identifying novel technologies that can enable large (>100m) blades for low specific power (SP) turbines. Five distinct tasks were completed to achieve this goal: 1. Assess trends, impacts, and value of low-SP turbines, 2. Wind turbine blade cost reduction roadmap study, 3. Research and Development (R&D) opportunity screening, 4. Detailed design and analysis, and 5. Low-cost carbon fiber. These tasks were completed by the national lab team consisting of Sandia National Laboratories (SNL), the National Renewable Energy Laboratory (NREL), and Lawrence Berkeley National Laboratory (LBNL). The objective of Task 1 was to assess the historical trends of low-SP of onshore deployments, and to assess the impact and value of low-SP turbines. Analysis under this task showed that there is significant value for low-SP turbines, especially in markets where there is a higher saturation of wind energy on the grid. This has to do with the fact that low-SP turbines have increased capacity factors, and can contribute more reliable energy to the grid, even in lower wind conditions. This bodes well as higher renewable scenarios are likely in the coming years. Task 2 was a detailed analysis on the logistical challenges of deploying very large (>100m) onshore blade. This work was completed by experts at DNV-GL. The report looked at various ways to get around the current transportation constraints which are estimated to be around 75m in length. The report concluded that there are viable solutions to this issue including: segmented blades (which the industry is already pursuing), lighter than air (LTA) transportation, on-site manufacturing, and controlled bending of blades on rail. The final option was recommended for further analysis and study by DOE and the national labs because keeping a single piece blade reduces manufacturing and operations and maintenance (O&M) costs. Task 3 focused on identifying novel concepts that are capable of enabling a cost effective 5MW 206m rotor for onshore deployment. The findings from the first two tasks were considered in the analysis, in that the concepts identified must be able to overcome the transportation logistics challenges. Around 20 concepts were identified and evaluated by experts within the industry. Additionally, science and engineering challenges were identified for each concept. Based on these evaluations six concepts were deemed the most impactful and were selected for further analysis in Task 4. The concepts were: highly flexible rail transportable blades, downwind rotors, distributed aerodynamic control (DAC) devices, inflatable blade, bi-wing blade, and 4/5 bladed rotors. Task 4 conducted detailed design, optimization, and analysis on the selected BAR concepts. A modeling gaps analysis was conducted, and modeling improvements were implemented that allowed for the study and design of the novel concepts. The BAR team established a set of baseline designs by which to compare the selected designs through a technoeconomic analysis. It was found that the highly flexible rail transportable blade, the downwind rotor, and the DAC devices have the most promise to deliver the BAR targets. A Phase II for BAR has been proposed to further mature these concepts and address the open science and engineering challenges identified in Task 3. Task 5 conducted research on optimized carbon fiber materials that were used throughout the BAR Phase I project. Overall, the BAR project has identified low-SP turbines as important to continued LCOE reductions for onshore turbines. Furthermore, the project has identified viable solutions to the technical and logistical challenges to realizing these goals. The most promising technologies that were identified in Phase I of the project will be further matured and de-risked in Phase II of the BAR project.

17 WIND ENERGY↗

Viscosity measurements and empirical predictions for coal slags

Slag viscosity in slagging coal gasifier is an important factor affecting the gasification regime and operating cost. Most of the empirical viscosity models of coal slags that are available in the literature are applicable to only limited ranges of temperature and composition. To develop a reliable slag viscosity model, additional data are needed. Slag viscosity was measured under air or reducing atmosphere (calculated pO2~1.2 10-12 atm at 1400°C) at temperatures in the range of 1150-1550°C on 63 statistically designed slags, including 5 actual coal slag compositions and 4 validation slag compositions. The Arrhenius equation, with Arrhenius coefficients A = constant and B expressed as linear function of mass fractions of nine major components was used to fit the viscosity/temperature data. This Arrhenius relationship represents the viscosity–temperature relationship of tested slags reasonably well, = 0.981 (reducing atmosphere) and = 0.974 (air atmosphere). The validation of the model with four randomly selected slags (two from the SciGlass database and two from experimental design) indicated an accurately measured viscosity-temperature data and a fairly good predictive performance of slag viscosity models over designed compositions. The capability of the developed model to predict the viscosity of coal slags under reducing atmosphere was found to be a superior to a number of the most commonly used empirical models in the literature that are based on simplified oxide melts and British or Australian coal ash slags.

01 COAL, LIGNITE, AND PEAT↗