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 343 records · Page 19

Quantification of Type I Interferon Inhibition by Viral Proteins: Ebola Virus as a Case Study

Type I interferons (IFNs) are cytokines with both antiviral properties and protective roles in innate immune responses to viral infection. They induce an antiviral cellular state and link innate and adaptive immune responses. Yet, viruses have evolved different strategies to inhibit such host responses. One of them is the existence of viral proteins which subvert type I IFN responses to allow quick and successful viral replication, thus, sustaining the infection within a host. We propose mathematical models to characterise the intra-cellular mechanisms involved in viral protein antagonism of type I IFN responses, and compare three different molecular inhibition strategies. We study the Ebola viral protein, VP35, with this mathematical approach. Approximate Bayesian computation sequential Monte Carlo, together with experimental data and the mathematical models proposed, are used to perform model calibration, as well as model selection of the different hypotheses considered. Finally, we assess if model parameters are identifiable and discuss how such identifiability can be improved with new experimental data.

59 BASIC BIOLOGICAL SCIENCES↗

Calculation of atmospheric loss from microwave radiometric noise temperature measurements

Microwave propagation loss in the atmosphere can be inferred from microwave radiometric noise temperature measurements. The relevant equations are given and a derivation and calculation is made assuming various physical models. Comparison is made with the commonly used lumped element atmospheric model (isothermal and uniform loss) and the model with linear temperature and exponential loss distributions. The results are useful for estimating the integral inversion differences due to the model selection. This indicates that the commonly used lumped element atmospheric model is a very good approximation with judicious choice of the effective physical temperature. For the worst case comparison, the lumped element model agrees with the variable parameter model within 0.2 dB up to a propagation loss of 3 dB.

Stelzried, C.↗

Implementation of and Ada real-time executive: A case study

Current Ada language implementations and runtime environments are immature, unproven and are a key risk area for real-time embedded computer system (ECS). A test-case environment is provided in which the concerns of the real-time, ECS community are addressed. A priority driven executive is selected to be implemented in the Ada programming language. The model selected is representative of real-time executives tailored for embedded systems used missile, spacecraft, and avionics applications. An Ada-based design methodology is utilized, and two designs are considered. The first of these designs requires the use of vendor supplied runtime and tasking support. An alternative high-level design is also considered for an implementation requiring no vendor supplied runtime or tasking support. The former approach is carried through to implementation.

Laird, James D.↗

Selection of a model of Earth's albedo radiation, practical calculation of its effect on the trajectory of a satellite

Theoretical models of Earth's albedo radiation was proposed. By comparing disturbing accelerations computed from a model to those measured in flight with the CACTUS Accelerometer, modified according to the results. Computation of the satellite orbit perturbations from a model is very long because for each position of this satellite the fluxes coming from each elementary surface of the terrestrial portion visible from the satellite must be summed. The speed of computation is increased ten times without significant loss of accuracy thanks to a stocking of some intermediate results. Now it is possible to confront the orbit perturbations computed from the selected model with the measurements of these perturbations found with satellite as LAGEOS.

Walch, J. J.↗

Application-specific optimal model weighting of global climate models: A red tide example

Global climate models (GCMs) and Earth system models (ESMs) provide many climate services with environmental relevance. The High Resolution Model Inter-comparison Project (HighResMIP) of the Coupled Model Intercomparison Project Phase 6 (CMIP6) provides model runs of GCMs and ESMs to address regional phenomena. Developing a parsimonious ensemble of CMIP6 requires multiple ensemble methods such as independent-model subset selection, prescreening-based subset selection, and model weighting. The work presented here focuses on application-specific optimal model weighting, with prescreening-based subset selection. As such, independent ensemble members are categorized, selected, and weighted based on their ability to reproduce physically-interpretable features of interest that are problem-specific. We discuss the strengths and caveats of optimal model weighting using a case study of red tide prediction in the Gulf of Mexico along the West Florida Shelf. Red tide is a common name of specific harmful algal blooms that occur worldwide, causing adverse socioeconomic and environmental impacts. Our results indicate the importance of prescreening-based subset selection as optimal model weighting can underplay robust ensemble members by optimizing error cancellation. Prescreening-based subset selection also provides insights about the validity of the model weights. By illustrating the caveats of using non-representative models when optimal model weighting is used, the findings and discussion of this study are pertinent to many other climate services.

54 ENVIRONMENTAL SCIENCES↗

Leveraging Compiler-Based Translation to Evaluate a Diversity of Exascale Platforms

Accelerator-based heterogeneous computing is the de facto standard in current and upcoming exascale machines. These heterogeneous resources empower computational scientists to select a machine or platform well-suited to their domain or applications. However, this diversity of machines also poses challenges related to programming model selection: inconsistent availability of programming models across different exascale systems, lack of performance portability for those programming models that do span several systems, and inconsistent performance between different models on a single platform. We explore these challenges on exascale-similar hardware, including AMD MI100 and NVIDIA A100 GPUs. By extending the sourceto-source compiler OpenARC, we demonstrate the power of automated translation of applications written in a single frontend programming model (OpenACC) into a variety of backend models (OpenMP, OpenCL, CUDA, HIP) that span the upcoming exascale environments. This translation enables us to compare performance within and across devices and to analyze programming model behavior with profiling tools.

Lambert, Jacob↗

Improving ChemCam LIBS long-distance elemental compositions using empirical abundance trends

The ChemCam instrument on the Curiosity rover provides chemical compositions of Martian rocks and soils using remote laser-induced breakdown spectroscopy (LIBS). The elemental calibration is stable as a function of distance for Ti, Fe, Mg, and Ca. The calibration shows small, systematically increasing abundance trends as a function of distance for Al, Na, K, and to some extent, Si. The distance effect is known to be due to a dependence with distance on the relative strengths of atomic transition lines. Emission lines representing transitions from relatively low energy levels remain intense at longer distances while emission lines representing transitions from higher energy levels decrease in intensity more rapidly as a function of distance. The multivariate algorithms used to determine elemental compositions rely on a large number of emission lines in many cases, so rather than trying to correct all emission lines, a study was made of the predicted compositions as a function of distance, in order to determine an empirical correction. Abundance trends can be well approximated by a linear trend with distance within the ranges of abundances and distances observed up to ~6 m. Data from 11 distinct geological members and data groups of the Murray formation in Gale crater, Mars, were used to form the model, selecting the members and data groups yielding the best statistics. The model was tested using data from several targets observed from two different distances, and using data from the Kimberley formation, the composition of which is significantly different from the Murray formation, showing that the model works on other compositions beyond those used to build the model. For long-distance observations up to ~6 m, corrections can be made back to an equivalent composition at the median distance of ChemCam observations (2.6 m). Finally, the model has been validated up to 6.2 m, although ChemCam is able to observe bedrock targets to >7 m, and iron meteorites to distances of >9 m.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Perspective: Ab initio force field methods derived from quantum mechanics

It is often desirable to accurately and efficiently model the behavior of large molecular systems in the condensed phase (thousands to tens of thousands of atoms) over long time scales (from nanoseconds to milliseconds). In these cases, ab initio methods are difficult due to the increasing computational cost with the number of electrons. A more computationally attractive alternative is to perform the simulations at the atomic level using a parameterized function to model the electronic energy. Many empirical force fields have been developed for this purpose. However, the functions that are used to model interatomic and intermolecular interactions contain many fitted parameters obtained from selected model systems, and such classical force fields cannot properly simulate important electronic effects. Furthermore, while such force fields are computationally affordable, they are not reliable when applied to systems that differ significantly from those used in their parameterization. They also cannot provide the information necessary to analyze the interactions that occur in the system, making the systematic improvement of the functional forms that are used difficult. Ab initio force field methods aim to combine the merits of both types of methods. The ideal ab initio force fields are built on first principles and require no fitted parameters. Ab initio force field methods surveyed in this perspective are based on fragmentation approaches and intermolecular perturbation theory. This perspective summarizes their theoretical foundation, key components in their formulation, and discusses key aspects of these methods such as accuracy and formal computational cost. The ab initio force fields considered here were developed for different targets, and this perspective also aims to provide a balanced presentation of their strengths and shortcomings. Finally, this perspective suggests some future directions for this actively developing area.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

A comparison of cosmological models using strong gravitational lensing galaxies

Strongly gravitationally lensed quasar-galaxy systems allow us to compare competing cosmologies as long as one can be reasonably sure of the mass distribution within the intervening lens. In this paper, we assemble a catalog of 69 such systems from the Sloan Lens ACS and Lens Structure and Dynamics surveys suitable for this analysis, and carry out a one-on-one comparison between the standard model, ΛCDM, and the R{sub h}=ct universe, which has thus far been favored by the application of model selection tools to other kinds of data. We find that both models account for the lens observations quite well, though the precision of these measurements does not appear to be good enough to favor one model over the other. Part of the reason is the so-called bulge-halo conspiracy that, on average, results in a baryonic velocity dispersion within a fraction of the optical effective radius virtually identical to that expected for the whole luminous-dark matter distribution modeled as a singular isothermal ellipsoid, though with some scatter among individual sources. Future work can greatly improve the precision of these measurements by focusing on lensing systems with galaxies as close as possible to the background sources. Given the limitations of doing precision cosmological testing using the current sample, we also carry out Monte Carlo simulations based on the current lens measurements to estimate how large the source catalog would have to be in order to rule out either model at a ∼99.7% confidence level. We find that if the real cosmology is ΛCDM, a sample of ∼200 strong gravitational lenses would be sufficient to rule out R{sub h}=ct at this level of accuracy, while ∼300 strong gravitational lenses would be required to rule out ΛCDM if the real universe were instead R{sub h}=ct. The difference in required sample size reflects the greater number of free parameters available to fit the data with ΛCDM. We point out that, should the R{sub h}=ct universe eventually emerge as the correct cosmology, its lack of any free parameters for this kind of work will provide a remarkably powerful probe of the mass structure in lensing galaxies, and a means of better understanding the origin of the bulge-halo conspiracy.

79 ASTRONOMY AND ASTROPHYSICS↗

Testing averaged cosmology with type Ia supernovae and BAO data

An important problem in precision cosmology is the determination of the effects of averaging and backreaction on observational predictions, particularly in view of the wealth of new observational data and improved statistical techniques. In this paper, we discuss the observational viability of a class of averaged cosmologies which consist of a simple parametrized phenomenological two-scale backreaction model with decoupled spatial curvature parameters. We perform a Bayesian model selection analysis and find that this class of averaged phenomenological cosmological models is favored with respect to the standard ΛCDM cosmological scenario when a joint analysis of current SNe Ia and BAO data is performed. In particular, the analysis provides observational evidence for non-trivial spatial curvature.

79 ASTRONOMY AND ASTROPHYSICS↗

Seasonal Variations in Mercury Deposition over the Yellow Sea, July 2007 through April 2008

Spatial and temporal variations of mercury, including dry and wet deposition fluxes, were assessed over Northeast Asia, targeting the Yellow Sea, using meteorology and chemistry models. Four modeling periods, each representative of one of the four seasons, were selected. Modeling results captured general patterns and behaviors, and fell within similar ranges with respect to observations. However, temporal variations of mercury were not closely matched, possibly owing to the effects of localized emissions. Modeling results indicated that dry deposition is correlated with wind speed, while wet deposition is correlated with precipitation amount. Overall, the wet deposition flux of 66 ng/m2-day was about twice as large as the dry deposition flux of 32 ng/m2-day, when averaged over the four modeling periods. Dry deposition occurred predominantly in the form of reactive gaseous mercury (RGM). In contrast, RGM accounted for only about two-thirds of wet deposition, while particulate mercury accounted for the remainder.

54 ENVIRONMENTAL SCIENCES↗

Regional Replay: A Unique Reanalysis-Based Tool for Addressing Model Error

Understanding and correcting errors in general circulation and climate models has long been part intuition and part trial and error. Efforts to diagnose the errors and provide some guidance to developers have been of some value, though such efforts, with few exceptions, have been more successful in identifying and documenting the errors in the model simulations rather than the model deficiencies that produced them. Modern atmospheric reanalyses such as MERRA-2 provide much-improved estimates of our climate system at hourly to interannual and longer time scales and have become an important tool for assessing model performance. Here we use MERRA-2 to address biases in the NASA/GMAO GEOS model by employing a "regional replay" approach developed in the GMAO. The regional replay approach constrains the model to remain close to the reanalysis over arbitrary regions and selected model variables, thus allowing us to examine how model error generated over one area is spatially translated across the globe. Several examples are given including an assessment of the global impact of errors produced over the Tibet region.

Tibet Region↗

Combined selection of the dynamic model and modeling error in nonlinear aeroelastic systems using Bayesian Inference

Here, we report a Bayesian framework for concurrent selection of physics-based models and (modeling) error models. We investigate the use of colored noise to capture the mismatch between the predictions of calibrated models and observational data that cannot be explained by measurement error alone within the context of Bayesian estimation for stochastic ordinary differential equations. Proposed models are characterized by the average data-fit, a measure of how well a model fits the measurements, and the model complexity measured using the Kullback–Leibler divergence. The use of a more complex error models increases the average data-fit but also increases the complexity of the combined model, possibly over-fitting the data. Bayesian model selection is used to find the optimal physical model as well as the optimal error model. The optimal model is defined using the evidence, where the average data-fit is balanced by the complexity of the model. The effect of colored noise process is illustrated using a nonlinear aeroelastic oscillator representing a rigid NACA0012 airfoil undergoing limit cycle oscillations due to complex fluid–structure interactions. Several quasi-steady and unsteady aerodynamic models are proposed with colored noise or white noise for the model error. The use of colored noise improves the predictive capabilities of simpler models.

42 ENGINEERING↗

Candidate massive galaxies at z ~ 4 in the Dark Energy Survey

Using stellar population models, we predicted that the Dark Energy Survey (DES) - due to its special combination of area (5000 deg. sq.) and depth ($i = 24.3$) - would be in the position to detect massive ($\gtrsim 10^{11}$ M$_{\odot}$) galaxies at $z \sim 4$. We confront those theoretical calculations with the first $\sim 150$ deg. sq. of DES data reaching nominal depth. From a catalogue containing $\sim 5$ million sources, $\sim26000$ were found to have observed-frame $g-r$ vs $r-i$ colours within the locus predicted for $z \sim 4$ massive galaxies. We further removed contamination by stars and artefacts, obtaining 606 galaxies lining up by the model selection box. We obtained their photometric redshifts and physical properties by fitting model templates spanning a wide range of star formation histories, reddening and redshift. Key to constrain the models is the addition, to the optical DES bands $g$, $r$, $i$, $z$, and $Y$, of near-IR $J$, $H$, $K_{s}$ data from the Vista Hemisphere Survey. We further applied several quality cuts to the fitting results, including goodness of fit and a unimodal redshift probability distribution. We finally select 233 candidates whose photometric redshift probability distribution function peaks around $z\sim4$, have high stellar masses ($\log($M$^{*}$/M$_{\odot})\sim 11.7$ for a Salpeter IMF) and ages around 0.1 Gyr, i.e. formation redshift around 5. These properties match those of the progenitors of the most massive galaxies in the local universe. This is an ideal sample for spectroscopic follow-up to select the fraction of galaxies which is truly at high redshift. These initial results and those at the survey completion, which we shall push to higher redshifts, will set unprecedented constraints on galaxy formation, evolution, and the re-ionisation epoch.

79 ASTRONOMY AND ASTROPHYSICS↗

RL10A-3-3A Rocket Engine Modeling Project

Two RL10A-3-3A rocket engines comprise the main propulsion system for the Centaur upper stage vehicle. Centaur is used with bod Titan and Atlas launch vehicles, carrying military and civilian payloads from high altitudes into orbit and beyond. The RL10 has delivered highly reliable service for the past 30 years. Recently, however, there have been two in-flight failures which have refocused attention on the RL10. This heightened interest has sparked a desire for an independent RL10 modeling capability within NASA and th Air Force. Pratt & Whitney, which presently has the most detailed model of the RL10, also sees merit in having an independent model which could be used as a cross-check with their own simulations. The Space Propulsion Technology Division (SPTD) at the NASA Lewis Research Center has developed a computer model of the RL10A-3-3A. A project team was formed, consisting of experts in the areas of turbomachinery, combustion, and heat transfer. The overall goal of the project was to provide a model of the entire RL10 rocket engine for government use. In the course of the project, the major engine components have been modeled using a combination of simple correlations and detailed component analysis tools (computer codes). The results of these component analyses were verified with data provided by Pratt & Whitney. Select modeling results and test data curves were then integrated to form the RL10 engine system model The purpose of this report is to introduce the reader to the RL10 rocket engine and to describe the engine system model. The RL10 engine and its application to U.S. launch vehicles are described first, followed by a summary of the SPTD project organization, goals, and accomplishments. Simulated output from the system model are shown in comparison with test and flight data for start transient, steady state, and shut-down transient operations. Detailed descriptions of all component analyses, including those not selected for integration with the system model, are included as appendices.

Binder, Michael↗

A comparison of probabilistic generative frameworks for molecular simulations

Generative artificial intelligence is now a widely used tool in molecular science. Despite the popularity of probabilistic generative models, numerical experiments benchmarking their performance on molecular data are lacking. Here, in this work, we introduce and explain several classes of generative models, broadly sorted into two categories: flow-based models and diffusion models. We select three representative models: neural spline flows, conditional flow matching, and denoising diffusion probabilistic models, and examine their accuracy, computational cost, and generation speed across datasets with tunable dimensionality, complexity, and modal asymmetry. Our findings are varied, with no one framework being the best for all purposes. In a nutshell, (i) neural spline flows do best at capturing mode asymmetry present in low-dimensional data, (ii) conditional flow matching outperforms other models for high-dimensional data with low complexity, and (iii) denoising diffusion probabilistic models appear the best for low-dimensional data with high complexity. Our datasets include a Gaussian mixture model and the dihedral torsion angle distribution of the Aib9 peptide, generated via a molecular dynamics simulation. We hope our taxonomy of probabilistic generative frameworks and numerical results may guide model selection for a wide range of molecular tasks.

Artificial intelligence↗

Comparing GIS-based habitat models for applications in EIA and SEA

Land use changes, urbanisation and infrastructure developments in particular, cause fragmentation of natural habitats and threaten biodiversity. Tools and measures must be adapted to assess and remedy the potential effects on biodiversity caused by human activities and developments. Within physical planning, environmental impact assessment (EIA) and strategic environmental assessment (SEA) play important roles in the prediction and assessment of biodiversity-related impacts from planned developments. However, adapted prediction tools to forecast and quantify potential impacts on biodiversity components are lacking. This study tested and compared four different GIS-based habitat models and assessed their relevance for applications in environmental assessment. The models were implemented in the Stockholm region in central Sweden and applied to data on the crested tit (Parus cristatus), a sedentary bird species of coniferous forest. All four models performed well and allowed the distribution of suitable habitats for the crested tit in the Stockholm region to be predicted. The models were also used to predict and quantify habitat loss for two regional development scenarios. The study highlighted the importance of model selection in impact prediction. Criteria that are relevant for the choice of model for predicting impacts on biodiversity were identified and discussed. Finally, the importance of environmental assessment for the preservation of biodiversity within the general frame of biodiversity conservation is emphasised.

54 ENVIRONMENTAL SCIENCES↗

Entropy Stable Discretization of Compressible Flows in Thermochemical Nonequilibrium

Entropy stable numerical methods for compressible flow have been demonstrated to exhibit better robustness than purely linearly stable methods and need less overall artificial dissipation for long simulations in subsonic and transonic flows. In this work we seek to extend these benefits to multicomponent, multitemperature flows in thermochemical nonequilibrium such as combustion and hypersonic flight. We first derive entropy functions that symmetrize the governing equations and allow stability proofs for such systems. The impact of diffusion model selection on provable entropy stability is considered in detail, including both rigorous models of irreversible thermodynamics and simplified models of greater practical interest. Based on the proven entropy functions we develop affordable, entropy conservative two-point flux functions for solution in conservation form. We derive entropy conservative fluxes for calorically and thermally perfect mixtures, with heat capacities described by either polynomials of the temperature or formulas from statistical thermodynamics.

42 ENGINEERING↗