Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Model Inference”

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 1,729 records · Page 96
All sourcesNASA NTRSDOE OSTI
DOE OSTI2023

Efficient Bayesian inference with latent Hamiltonian neural networks in No-U-Turn Sampling

When sampling for Bayesian inference, one popular approach in the computational field is to use Hamiltonian Monte Carlo (HMC) and specifically the No-U-Turn Sampler (NUTS), which automatically decides the end time of the Hamiltonian trajectory. However, HMC and NUTS can require numerous numerical gradients of the target density and can prove slow in practice when relying on computationally expensive forward models. We propose Latent Hamiltonian neural networks (L-HNNs) with HMC and NUTS for solving Bayesian inference problems. Once trained, L-HNNs do not require numerical gradients of the target density during sampling, and hence numerous evaluations of the forward computational model. Moreover, L-HNNs satisfy important properties such as perfect time reversibility and Hamiltonian conservation, making them well-suited for use within HMC and NUTS because stationarity can be shown. We also propose the integration of L-HNNs in an online error monitoring scheme, in which numerical gradients of the target density are used for a few samples whenever the L-HNNs prediction errors are large. This online error monitor scheme prevents sample degeneracy in regions of low probability density and ensures robust uncertainty quantification. We demonstrate L-HNNs in NUTS with online error monitoring on several analytical examples involving complex, heavy-tailed, and high-local-curvature probability densities. We then demonstrate the applicability of L-HNNs in NUTS to two computational case studies, namely the Allen-Cahn stochastic partial differential equation and an elliptic partial differential equation with 25 and 50 inference parameters, respectively. Overall, the L-HNNs in NUTS with online error monitoring satisfactorily inferred these probability densities. In conclusion, compared to traditional NUTS, L-HNNs in NUTS with online error monitoring required 1–2 orders of magnitude fewer numerical gradients of the target density and improved the effective sample size (ESS) per gradient (which is a measure of both the sampling quality and the computational expense) by an order of magnitude.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Characterizing the Sample Selection for Supernova Cosmology

Type Ia supernovae (SNe Ia) are used as distance indicators to infer the cosmological parameters that specify the expansion history of the universe. Parameter inference depends on the criteria by which the analysis SN sample is selected. Only for the simplest selection criteria and population models can the likelihood be calculated analytically, otherwise it needs to be determined numerically, a process that inherently has error. Numerical errors in the likelihood lead to errors in parameter inference. This article presents toy examples where the distance modulus is inferred given a set of SNe at a single redshift. Parameter estimators and their uncertainties are calculated using Monte Carlo techniques. The relationship between the number of Monte Carlo realizations and numerical errors is presented. The procedure can be applied to more realistic models and used to determine the computational and data management requirements of the transient analysis pipeline.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2019

Multi-fidelity classification using Gaussian processes: Accelerating the prediction of large-scale computational models

We report that machine learning techniques typically rely on large datasets to create accurate classifiers. However, there are situations when data is scarce and expensive to acquire. This is the case of studies that rely on state-of-the-art computational models which typically take days to run, thus hindering the potential of machine learning tools. In this work, we present a novel classifier that takes advantage of lower fidelity models and inexpensive approximations to predict the binary output of expensive computer simulations. We postulate an autoregressive model between the different levels of fidelity with Gaussian process priors. We adopt a fully Bayesian treatment for the hyper-parameters and use Markov Chain Monte Carlo samplers. We take advantage of the probabilistic nature of the classifier to implement active learning strategies. We also introduce a sparse approximation to enhance the ability of the multi-fidelity classifier to handle a large amount of low fidelity samples. We test these multi-fidelity classifiers against their single-fidelity counterpart with synthetic data, showing a median computational cost reduction of 23% for a target accuracy of 90%. In an application to cardiac electrophysiology, the multi-fidelity classifier achieves an F1 score, the harmonic mean of precision and recall, of 99.6% compared to 74.1% of a single-fidelity classifier when both are trained with 50 samples. In general, our results show that the multi-fidelity classifiers outperform their single-fidelity counterpart in terms of accuracy in all cases. Finally, we envision that this new tool will enable researchers to study classification problems that would otherwise be prohibitively expensive. Source code is available at https://github.com/fsahli/MFclass.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2017

Implications of Binary Black Hole Detections on the Merger Rates of Double Neutron Stars and Neutron Star–Black Holes

We show that the inferred merger rate and chirp masses of binary black holes (BBHs) detected by advanced LIGO (aLIGO) can be used to constrain the rate of double neutron star (DNS) and neutron star–black hole (NSBH) mergers in the universe. We explicitly demonstrate this by considering a set of publicly available population synthesis models of Dominik et al. and show that if all the BBH mergers, GW150914, LVT151012, GW151226, and GW170104, observed by aLIGO arise from isolated binary evolution, the predicted DNS merger rate may be constrained to be 2.3–471.0 Gpc{sup −3} yr{sup −1} and that of NSBH mergers will be constrained to 0.2–48.5 Gpc{sup −3} yr{sup −1}. The DNS merger rates are not constrained much, but the NSBH rates are tightened by a factor of ∼4 as compared to their previous rates. Note that these constrained DNS and NSBH rates are extremely model-dependent and are compared to the unconstrained values 2.3–472.5 Gpc{sup −3} yr{sup −1} and 0.2–218 Gpc{sup −3} yr{sup −1}, respectively, using the same models of Dominik et al. (2012a). These rate estimates may have implications for short Gamma Ray Burst progenitor models assuming they are powered (solely) by DNS or NSBH mergers. While these results are based on a set of open access population synthesis models, which may not necessarily be the representative ones, the proposed method is very general and can be applied to any number of models, thereby yielding more realistic constraints on the DNS and NSBH merger rates from the inferred BBH merger rate and chirp mass.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2017

Strong lensing analysis of the galaxy cluster MACS J1319.9+7003 and the discovery of a shell galaxy

We present a strong-lensing (SL) analysis of the galaxy cluster MACS J1319.9+7003 (z = 0.33, also known as Abell 1722), as part of our ongoing effort to analyze massive clusters with archival Hubble Space Telescope (HST) imaging. We spectroscopically measured with Keck/Multi-Object Spectrometer For Infra-Red Exploration (MOSFIRE) two galaxies multiply imaged by the cluster. Our analysis reveals a modest lens, with an effective Einstein radius of θ{sub e}(z=2)=12±1{sup ′′}, enclosing 2.1±0.3×10{sup 13} M{sub ⊙}. We briefly discuss the SL properties of the cluster, using two different modeling techniques (see the text for details), and make the mass models publicly available (ftp://wise-ftp.tau.ac.il/pub/adiz/MACS1319/). Independently, we identified a noteworthy, young shell galaxy (SG) system forming around two likely interacting cluster members, 20″ north of the brightest cluster galaxy. SGs are rare in galaxy clusters, and indeed, a simple estimate reveals that they are only expected in roughly one in several dozen, to several hundred, massive galaxy clusters (the estimate can easily change by an order of magnitude within a reasonable range of characteristic values relevant for the calculation). Taking advantage of our lens model best-fit, mass-to-light scaling relation for cluster members, we infer that the total mass of the SG system is ∼1.3×10{sup 11} M{sub ⊙}, with a host-to-companion mass ratio of about 10:1. Despite being rare in high density environments, the SG constitutes an example to how stars of cluster galaxies are efficiently redistributed to the intra-cluster medium. Dedicated numerical simulations for the observed shell configuration, perhaps aided by the mass model, might cast interesting light on the interaction history and properties of the two galaxies. An archival HST search in galaxy cluster images can reveal more such systems.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2023

Tabulated equations of state from models informed by chiral effective field theory

Abstract We construct four equation of state (EoS) tables, tabulated over a range of temperatures, densities, and charge fractions, relevant for neutron star applications such as simulations of neutron star mergers. The EoS are computed from a relativistic mean-field theory constrained by the pure neutron matter EoS from chiral effective field theory, inferred properties of isospin-symmetric nuclear matter, and astrophysical observations of neutron star structure. To model nuclear matter at low densities, we attach an EoS that models inhomogeneous nuclear matter at arbitrary temperatures and charge fractions. The four EoS tables we develop are available from the CompOSE EoS repository https://compose.obspm.fr/eos/297 and gitlab.com/ahaber/qmc-rmf-tables .

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2019

Aquaplanet Models on Eccentric Orbits: Effects of the Rotation Rate on Observables

Rotation and orbital eccentricity both strongly influence planetary climate. Eccentricities can often be measured for exoplanets, but rotation rates are currently difficult or impossible to constrain. Here we examine how the combined effects of rotation and eccentricity on observed emission from ocean-rich terrestrial planets can be used to infer their rotation rates in circumstances where their eccentricities are known. We employ an Earth climate model with no land and a slab ocean, and consider two eccentricities (e = 0.3 and 0.6) and two rotation rates: a fast Earth-like period of 24 hr, and a slower pseudo-synchronous period that generalizes spin synchronization for eccentric orbits. We adopt bandpasses of the Mid-Infrared Instrument on the James Webb Space Telescope as a template for future photometry. At e = 0.3 the rotation rates can be distinguished if the planet transits near periastron, because slow rotation produces a strong day–night contrast and thus an emission minimum during periastron. However, light curves behave similarly if the planet is eclipsed near periastron, as well as for either viewing geometry at e = 0.6. Rotation rates can nevertheless be distinguished using ratios of emission in different bands, one in the water vapor window with another in a region of strong water absorption. These ratios vary over an orbit by ≲0.1 dex for Earth-like rotation, but by 0.3–0.5 dex for pseudo-synchronous rotation because of large day–night contrast in upper-tropospheric water. For planets with condensible atmospheric constituents in eccentric orbits, rotation regimes might thus be distinguished with infrared observations for a range of viewing geometries.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2022

Continuous-time probabilistic models for longitudinal electronic health records

Analysis of longitudinal Electronic Health Record (EHR) data is an important goal for precision medicine. Difficulty in applying Machine Learning (ML) methods, either predictive or unsupervised, stems in part from the heterogeneity and irregular sampling of EHR data. Here, we present an unsupervised probabilistic model that captures nonlinear relationships between variables over continuous-time. This method works with arbitrary sampling patterns and captures the joint probability distribution between variable measurements and the time intervals between them. Inference algorithms are derived that can be used to evaluate the likelihood of future using under a trained model. As an example, we consider data from the United States Veterans Health Administration (VHA) in the areas of diabetes and depression. Likelihood ratio maps are produced showing the likelihood of risk for moderate-severe vs minimal depression as measured by the Patient Health Questionnaire-9 (PHQ-9).

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2013

(Lack of) Cosmological evidence for dark radiation after Planck

We use Bayesian model comparison to determine whether extensions to Standard-Model neutrino physics — primarily additional effective numbers of neutrinos and/or massive neutrinos — are merited by the latest cosmological data. Given the significant advances in cosmic microwave background (CMB) observations represented by the Planck data, we examine whether Planck temperature and CMB lensing data, in combination with lower redshift data, have strengthened (or weakened) the previous findings. We conclude that the state-of-the-art cosmological data do not show evidence for deviations from the standard (ΛCDM) cosmological model (which has three massless neutrino families). This does not mean that the model is necessarily correct — in fact we know it is incomplete as neutrinos are not massless — but it does imply that deviations from the standard model (e.g., non-zero neutrino mass) are too small compared to the current experimental uncertainties to be inferred from cosmological data alone.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2020

Dynamic Equivalence of Large-Scale Power Systems Based on Boundary Measurements

Parallel computing helped speed up many tasks that can be done independently. The advance in gaming industry did not produce a technology that helps the power industry to reduce the computation time for dynamic simulation in individual cases. To perform faster than real-time simulation for the purpose of predicting power system dynamic trajectory, power industry continues to struggle to reduce the system size and simulation time of large-scale power systems while keeping its dynamic behavior under various disturbances. There is also an acute need for real-time dynamic model reduction as more renewables enter the generation mix with dramatic changes in the generation outputs. All existing model reduction solutions are based on having access to a detailed dynamic model of the system. The system changes by the minutes, dynamic models are only updated annually. To solve this problem, wide-area measurements obtained by the phasor measurement unit (PMU) at the boundaries between the reduced system and the study system is used to represent the external area. An artificial neuro-fuzzy inference system (ANFIS) is established to perform the mapping of the measurement to the external equivalent model. Here the external area is regarded as a black-box. Model reduction studies are conducted on the Northeast Power Coordinating Council (NPCC) under various types of contingencies, by using a co-simulation approach between PSS/E and MATLAB. The results look very promising and will be discussed in this paper.

Tong, Ning↗
DOE OSTI2024

Projective Integral Updates for High-Dimensional Variational Inference

Variational inference is an approximation framework for Bayesian inference that seeks to improve quantified uncertainty in predictions by optimizing a simplified distribution over parameters to stand in for the full posterior. Capturing model variations that remain consistent with training data enables more robust predictions by reducing parameter sensitivity. This work introduces a fixed-point optimization for variational inference that is applicable when every feasible log density can be expressed as a linear combination of functions from a given basis. In such cases, the optimizer becomes a fixed-point of projective integral updates. When the basis spans univariate quadratics in each parameter, the feasible distributions are Gaussian mean-fields and the projective integral updates yield quasi-Newton variational Bayes (QNVB). Other bases and updates are also possible. Since these updates require high-dimensional integration, this work begins by proposing an efficient quasirandom sequence of quadratures for mean-field distributions. Each iterate of the sequence contains two evaluation points that combine to correctly integrate all univariate quadratic functions and, if the mean-field factors are symmetric, all univariate cubics. More importantly, averaging results over short subsequences achieves periodic exactness on a much larger space of multivariate polynomials of quadratic total degree. The corresponding variational updates require four loss evaluations with standard (not second-order) backpropagation to eliminate error terms from over half of all multivariate quadratic basis functions. Furthermore, this integration technique is motivated by first proposing stochastic blocked mean-field quadratures, which may be useful in other contexts. A PyTorch implementation of QNVB allows for better control over model uncertainty during training than competing methods. Experiments demonstrate superior generalizability for multiple learning problems and architectures.

Gaussian mean-field↗
DOE OSTI2017

Balmer Filaments in Tycho’s Supernova Remnant: An Interplay between Cosmic-ray and Broad-neutral Precursors

We present Hα spectroscopic observations and detailed modeling of the Balmer filaments in the supernova remnant (SNR) Tycho (SN 1572). We used GH α FaS (Galaxy Hα Fabry–Pérot Spectrometer) on the William Herschel Telescope with a 3.′4 × 3.′4 field of view, 0.″2 pixel scale, and σ{sub instr}=8.1 km s{sup −1} resolution at 1″ seeing for ∼10 hr, resulting in 82 spatial−spectral bins that resolve the narrow Hα line in the entire SN 1572 northeastern rim. For the first time, we can therefore mitigate artificial line broadening from unresolved differential motion and probe Hα emission parameters in varying shock and ambient medium conditions. Broad Hα line remains unresolved within spectral coverage of 392 km s{sup −1}. We employed Bayesian inference to obtain reliable parameter confidence intervals and to quantify the evidence for models with multiple line components. The median Hα narrow-line (NL) FWHM of all bins and models is W{sub NL}=(54.8±1.8) km s{sup −1} at the 95% confidence level, varying within [35, 72] km s{sup −1} between bins and clearly broadened compared to the intrinsic (thermal) ≈20 km s{sup −1}. Possible line splits are accounted for, significant in ≈18% of the filament, and presumably due to remaining projection effects. We also find widespread evidence for intermediate-line emission of a broad-neutral precursor, with a median W{sub IL}=(180±14) km s{sup −1} (95% confidence). Finally, we present a measurement of the remnant’s systemic velocity, V{sub LSR}=−34 km s{sup −1}, and map differential line-of-sight motions. Our results confirm the existence and interplay of shock precursors in Tycho’s remnant. In particular, we show that suprathermal NL emission is near-universal in SN 1572, and that, in the absence of an alternative explanation, collisionless SNR shocks constitute a viable acceleration source for Galactic TeV cosmic-ray protons.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2022

Optical depth to reionization from perturbative 21 cm clustering

The optical depth τ is the least well determined parameter in the standard model of cosmology, and one whose precise value is important for both understanding reionization and for inferring fundamental physics from cosmological measurements. Here, we forecast how well future epoch of reionization experiments could constraint τ using a symmetries-based bias expansion that highlights the special role played by anisotropies in the power spectrum on large scales. Given a parametric model for the ionization evolution inspired by the physical behavior of more detailed reionization simulations, we find that future 21 cm experiments could place tight constraints on the timing and duration of reionization and hence constraints on τ that are competitive with proposed, space-based CMB missions provided they can measure k ≈ 0.1 h Mpc -1 with a clean foreground wedge across redshifts spanning the most active periods of reionization, corresponding to ionization fractions 0.2 ≲ x ≲ 0.8. Significantly improving upon existing CMB-based measurements with next-generation 21 cm surveys would require substantially longer observations (~ 5 years) than standard $\mathscr{O}$(1000 hour) integration times. Precise measurements of smaller scales will not improve constraints on τ until a better understanding of the astrophysics of reionization is achieved. In the presence of noise and foregrounds even future 21 cm experiments will struggle to constrain τ if the ionization evolution deviates significantly from simple parametric forms.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2014

Network Analysis of Breast Cancer Progression and Reversal Using a Tree-Evolving Network Algorithm

The HMT3522 progression series of human breast cells have been used to discover how tissue architecture, microenvironment and signaling molecules affect breast cell growth and behaviors. However, much remains to be elucidated about malignant and phenotypic reversion behaviors of the HMT3522-T4-2 cells of this series. We employed a ‘‘pan-cell-state’’ strategy, and analyzed jointly microarray profiles obtained from different state-specific cell populations from this progression and reversion model of the breast cells using a tree-lineage multi-network inference algorithm, Treegl. We found that different breast cell states contain distinct gene networks. The network specific to non-malignant HMT3522-S1 cells is dominated by genes involved in normal processes, whereas the T4-2-specific network is enriched with cancer-related genes. The networks specific to various conditions of the reverted T4-2 cells are enriched with pathways suggestive of compensatory effects, consistent with clinical data showing patient resistance to anticancer drugs. We validated the findings using an external dataset, and showed that aberrant expression values of certain hubs in the identified networks are associated with poor clinical outcomes. Thus, analysis of various reversion conditions (including non-reverted) of HMT3522 cells using Treegl can be a good model system to study drug effects on breast cancer.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2024

3D multi-system Bayesian calibration with energy conservation to study rapidity-dependent dynamics of nuclear collisions

Considerable information about the early-stage dynamics of heavy ion collisions is encoded in the rapidity dependence of measurements. To leverage the large amount of experimental data, we perform a systematic analysis using three-dimensional hydrodynamic simulations of multiple collision systems — large and small, symmetric and asymmetric. Specifically, we perform fully 3D multi-stage hydrodynamic simulations initialized by a parameterized model for rapidity-dependent energy deposition, which we calibrate on the hadron multiplicity and anisotropic flow coefficients. We utilize Bayesian inference to constrain properties of the early- and late-time dynamics of the system, and highlight the impact of enforcing global energy conservation in our 3D model.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2020

Investigating effects of soil chemicals on density of small mammal bioindicators using spatial capture-recapture models

Monitoring the ecological impacts of environmental pollution and the effectiveness of remediation efforts requires identifying relationships between contaminants and the disruption of biological processes in populations, communities, or ecosystems. Wildlife are useful bioindicators, but traditional comparative experimental approaches rely on a staunch and typically unverifiable assumption that, in the absence of contaminants, reference and contaminated sites would support the same densities of bioindicators, thereby inferring direct causation from indirect data. We demonstrate the utility of spatial capture-recapture (SCR) models for overcoming these issues, testing if community density of common small mammal bioindicators was directly influenced by soil chemical concentrations. By modeling density as an inhomogeneous Poisson point process, we found evidence for an inverse spatial relationship between Peromyscus density and soil mercury concentrations, but not other chemicals, such as polychlorinated biphenyls, at a site formerly occupied by a nuclear reactor. Although the coefficient point estimate supported Peromyscus density being lower where mercury concentrations were higher (β = –0.44), the 95% confidence interval overlapped zero, suggesting no effect was also compatible with our data. Estimated density from the most parsimonious model (2.88 mice/ha; 95% CI = 1.63–5.08), which did not support a density-chemical relationship, was within the range of reported densities for Peromyscus that did not inhabit contaminated sites elsewhere. Environmental pollution remains a global threat to biodiversity and ecosystem and human health, and our study provides an illustrative example of the utility of SCR models for investigating the effects that chemicals may have on wildlife bioindicator populations and communities.

36 MATERIALS SCIENCE↗
DOE OSTI2025

Analyzing line-of-sight selection biases in galaxy-scale strong lensing with external convergence and shear

The upcoming Vera Rubin Observatory Legacy Survey of Space and Time (LSST) will dramatically increase the number of strong gravitational lensing systems, requiring precise modeling of line-of-sight (LOS) effects to mitigate biases in lensing observations and cosmological inferences. We develop a method to construct joint distributions of external convergence (κ ext ) and shear (γ ext ) for strong lensing LOS by aggregating large-scale structure simulations with high-resolution halo renderings and non-linear correction. Our approach captures both smooth background matter and perturbations from halos, enabling accurate modeling of LOS effects. Here, we apply non-linear LOS corrections to κ ext and γ ext that address the non-additive lensing effects caused by objects along the LOS in strong lensing. We find that, with a minimum image separation of 1.0'', non-linear LOS correction due to the presence of a dominant deflector slightly increases the ratio of quadruple to double lenses; this non-linear LOS correction also introduces systematic biases of ∼ 0.1% for galaxy-AGN lens in the inferred Hubble constant (H 0 ) if not accounted for. We also observe a 0.66% bias for galaxy-galaxy lenses on H 0 , and even larger biases 1.02% for galaxy-AGN systems if LOS effects are not accounted for. These results highlight the importance of LOS for precision cosmology. The publicly available code and datasets provide tools for incorporating LOS effects in future analyses.

Hubble constant↗