Search NASA⌕ Search

SEARCH · Search NASA

Results for “Bayesian Framework”

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 1,027 records · Page 57

BAYESIAN ANALYSIS OF WHITE NOISE LEVELS IN THE FIVE-YEAR WMAP DATA

We develop a new Bayesian method for estimating white noise levels in CMB sky maps, and apply this algorithm to the five-year Wilkinson Microwave Anisotropy Probe (WMAP) data. We assume that the amplitude of the noise rms is scaled by a constant value, {alpha}, relative to a pre-specified noise level. We then derive the corresponding conditional density, P({alpha} | s, C {sub l}, d), which is subsequently integrated into a general CMB Gibbs sampler. We first verify our code by analyzing simulated data sets, and then apply the framework to the WMAP data. For the foreground-reduced five-year WMAP sky maps and the nominal noise levels initially provided in the five-year data release, we find that the posterior means typically range between {alpha} = 1.005 {+-} 0.001 and {alpha} = 1.010 {+-} 0.001 depending on differencing assembly, indicating that the noise level of these maps are biased low by 0.5%-1.0%. The same problem is not observed for the uncorrected WMAP sky maps. After the preprint version of this letter appeared on astro-ph., the WMAP team has corrected the values presented on their web page, noting that the initially provided values were in fact estimates from the three-year data release, not from the five-year estimates. However, internally in their five-year analysis the correct noise values were used, and no cosmological results are therefore compromised by this error. Thus, our method has already been demonstrated in practice to be both useful and accurate.

79 ASTRONOMY AND ASTROPHYSICS↗

A Framework for Quantifying Measurement Uncertainties and Uncertainty Propagation in HCCI/LTGC Engine Experiments

In this study, a framework for estimating experimental measurement uncertainties for a Homogenous Charge Compression Ignition (HCCI)/Low-Temperature Gasoline Combustion (LTGC) engine testing facility is presented. Detailed uncertainty quantification is first carried out for the measurement of the in-cylinder pressure, whose variations during the cycle provide most of the information for performance evaluation. Standard uncertainties of other measured quantities, such as the engine geometry and speed, the air and fuel flow rate and the intake/exhaust dry molar fractions are also estimated. Propagating those uncertainties using a Monte Carlo simulation and Bayesian inference methods then allows for estimation of uncertainties of the mass-average temperature and composition at IVC and throughout the cycle; and also of the engine performances such as gross Integrated Mean Effective Pressure, Heat Release and Ringing Intensity. Throughout the analysis, nominal values for uncertainty inputs were taken from a well-characterized engine test facility. However, the analysis did not take into account the calibration practice of experiments run in that facility and the resulting uncertainty values are therefore not indicative of the expected accuracy of those experimental results. A future study will employ the methodology developed here to explore the effects of different calibration methods on the different uncertainty values in order to evaluate best practices for accurate engine measurements.

02 PETROLEUM↗

On a framework for generating PoD curves assisted by numerical simulations

The Probability of Detection (PoD) curve method has emerged as an important tool for the assessment of the performance of NDE techniques, a topic of particular interest to the nuclear industry where inspection qualification is very important. The conventional experimental means of generating PoD curves though, can be expensive, requiring large data sets (covering defects and test conditions), and equipment and operator time. Several methods of achieving faster estimates for PoD curves using physics-based modelling have been developed to address this problem. Numerical modelling techniques are also attractive, especially given the ever-increasing computational power available to scientists today. Here we develop procedures for obtaining PoD curves, assisted by numerical simulation and based on Bayesian statistics. Numerical simulations are performed using Finite Element analysis for factors that are assumed to be independent, random and normally distributed. PoD curves so generated are compared with experiments on austenitic stainless steel (SS) plates with artificially created notches. We examine issues affecting the PoD curve generation process including codes, standards, distribution of defect parameters and the choice of the noise threshold. We also study the assumption of normal distribution for signal response parameters and consider strategies for dealing with data that may be more complex or sparse to justify this. These topics are addressed and illustrated through the example case of generation of PoD curves for pulse-echo ultrasonic inspection of vertical surface-breaking cracks in SS plates.

36 MATERIALS SCIENCE↗

Exploration of lattice Hamiltonians for functional and structural discovery via Gaussian process-based exploration–exploitation

Statistical physics models ranging from simple lattice to complex quantum Hamiltonians are one of the mainstays of modern physics that have allowed both decades of scientific discovery and provided a universal framework to understand a broad range of phenomena from alloying to frustrated and phase separated materials to quantum systems. Traditionally, exploration of the phase diagrams corresponding to multidimensional parameter spaces of Hamiltonians was performed using a combination of basic physical principles, analytical approximations, and extensive numerical modeling. However, exploration of complex multidimensional parameter spaces is subject to the classic dimensionality problem, and the behaviors of interest concentrated on low dimensional manifolds remain undiscovered. Here, we demonstrate that a combination of exploration and exploration–exploitation with Gaussian process modeling and Bayesian optimization allows effective exploration of the parameter space for lattice Hamiltonians and effectively maps the regions at which specific macroscopic functionalities or local structures are maximized. We argue that this approach is general and can be further extended well beyond the lattice Hamiltonians to effectively explore the parameter space of more complex off-lattice and dynamic models.

42 ENGINEERING↗

A New Ensemble of Perturbed-Input-Parameter Simulations by the Community Atmosphere Model

Uncertainty quantification (UQ) is a fundamental challenge in the numerical simulation of Earth's weather and climate, and other complex systems. It entails much more than attaching defensible error bars to predictions: in particular it includes assessing low-probability but high-consequence events. To achieve these goals with models containing a large number of uncertain input parameters, structural uncertainties, etc., raw computational power is needed. An automated, self-adapting search of the possible model configurations is also useful. Our UQ initiative at the Lawrence Livermore National Laboratory has produced the most extensive set to date of simulations from the US Community Atmosphere Model. We are examining output from about 3,000 twelve-year climate simulations generated with a specialized UQ software framework, and assessing the model's accuracy as a function of 21 to 28 uncertain input parameter values. Most of the input parameters we vary are related to the boundary layer, clouds, and other sub-grid scale processes. Our simulations prescribe surface boundary conditions (sea surface temperatures and sea ice amounts) to match recent observations. Fully searching this 21+ dimensional space is impossible, but sensitivity and ranking algorithms can identify input parameters having relatively little effect on a variety of output fields, either individually or in nonlinear combination. Bayesian statistical constraints, employing a variety of climate observations as metrics, also seem promising. Observational constraints will be important in the next step of our project, which will compute sea surface temperatures and sea ice interactively, and will study climate change due to increasing atmospheric carbon dioxide.

54 ENVIRONMENTAL SCIENCES↗

ETHOS: An automated framework to generate multi-fidelity constitutive data tables and propagate uncertainties to hydrodynamic simulations

Accurate constitutive data, such as equations of state and plasma transport coefficients, are necessary for reliable hydrodynamic simulations of plasma systems such as fusion targets, planets, and stars. Here, we develop a framework for automatically generating transport-coefficient tables using a parameterized model that incorporates data from both high-fidelity sources (e.g., density functional theory calculations and reference experiments) and lower-fidelity sources (e.g., average-atom and analytic models). The framework incorporates uncertainties from these multi-fidelity sources, generating ensembles of optimally diverse tables that are suitable for uncertainty quantification of hydrodynamic simulations. We illustrate the utility of the framework with magnetohydrodynamic simulations of magnetically launched flyer plates, which are used to measure material properties in pulsed-power experiments. We explore how changes in the uncertainties assigned to the multi-fidelity data sources propagate to changes in simulation outputs and find that our simulations are most sensitive to uncertainties near the melting transition. The presented framework enables computationally efficient uncertainty quantification that readily incorporates new high-fidelity measurements or calculations and identifies plasma regimes where additional data will have high impact.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

In-Time Safety Assessment & Risk Prediction for Unmanned Aerial Systems

One of the critical challenges in emerging autonomous systems is timely mitigation of hazards encountered during operation which may not be known or accounted for at the time of design. Efficient execution of unmanned systems therefore demands a paradigm shift from scheduled periodic maintenance to predictive risk analysis that includes condition-based-monitoring, real-time reliability assessment and hazard mitigation. Particularly, the state-of-health parameters needs to be computed at the component level, unit level as well as the integrated system level. While in the former two levels, the physics of health propagation may be based on underlying electro-mechanical properties, system level prognostics often relies on data-driven models. Further, uncertainty from model, measurements and input sources should be accurately quantified to generate meaningful prediction results that can be fed into reliable decision making processes. Finally, the expected risk and time to failure has to be computed based on the current state-of-health of the overall system. This talk presents a conceptual design of such an in-time safety assurance approach for unmanned aerial vehicles (UAV) operating at low altitudes near and over populated areas. Typical in-flight hazard incidents include unplanned detour, proximity to obstacles, mid-flight component faults, limited battery life and poor quality of GPS measurements. Safety assessment therefore comprises trajectory generation and re-plan, battery RUL computation, distributed fault diagnostics and uncertainty management of predicted trajectory based on GPS measurement noise. The entire monitoring framework will be demonstrated on simulated as well as real UAV flight experiments conducted at the NASA Langley Research Center. This tutorial will therefore guide the audience through a step-by-step tracking of an autonomous system with focus on in-time risk prediction in the presence of unforeseen hazards and uncertain environment.

diagnostics↗

Uncertainty quantification for Multiphase-CFD simulations of bubbly flows: a machine learning-based Bayesian approach supported by high-resolution experiments

In this paper, we developed a machine learning-based Bayesian approach to inversely quantify and reduce the uncertainties of multiphase computational fluid dynamics (MCFD) simulations for bubbly flows. The proposed approach is supported by high-resolution two-phase flow measurements, including those by double-sensor conductivity probes, high-speed imaging, and particle image velocimetry. Local distributions of key physical quantities of interest (QoIs), including the void fraction and phasic velocities, are obtained to support the Bayesian inference. In the process, the epistemic uncertainties of the closure relations are inversely quantified while the aleatory uncertainties from stochastic fluctuations of the system are evaluated based on experimental uncertainty analysis. The combined uncertainties are then propagated through the MCFD solver to obtain uncertainties of the QoIs, based on which probability-boxes are constructed for validation. The proposed approach relies on three machine learning methods: feedforward neural networks and principal component analysis for surrogate modeling, and Gaussian processes for model form uncertainty modeling. The whole process is implemented within the framework of an open-source deep learning library PyTorch with graphics processing unit (GPU) acceleration, thus ensuring the efficiency of the computation. The results demonstrate that with the support of high-resolution data, the uncertainties of MCFD simulations can be significantly reduced. The proposed approach has the potential for other applications that involve numerical models with empirical parameters.

42 ENGINEERING↗

Breaking Computational Barriers: Real-time Analysis and Optimization with Large-scale Nonlinear Models via Model Reduction

Model reduction for dynamical systems is a promising approach for reducing the computational cost of large-scale physics-based simulations to enable high-fidelity models to be used in many- query (e.g., Bayesian inference) and near-real-time (e.g., fast-turnaround simulation) contexts. While model reduction works well for specialized problems such as linear time-invariant systems, it is much more difficult to obtain accurate, stable, and efficient reduced-order models (ROMs) for systems with general nonlinearities. This report describes several advances that enable nonlinear reduced-order models (ROMs) to be deployed in a variety of time-critical settings. First, we present an error bound for the Gauss-Newton with Approximated Tensors (GNAT) nonlinear model reduction technique. This bound allows the state-space error for the GNAT method to be quantified when applied with the backward Euler time-integration scheme. Second, we present a methodology for preserving classical Lagrangian structure in nonlinear model reduction. This technique guarantees that important properties--such as energy conservation and symplectic time-evolution maps--are preserved when performing model reduction for models described by a Lagrangian formalism (e.g., molecular dynamics, structural dynamics). Third, we present a novel technique for decreasing the temporal complexity --defined as the number of Newton-like iterations performed over the course of the simulation--by exploiting time-domain data. Fourth, we describe a novel method for refining projection-based reduced-order models a posteriori using a goal-oriented framework similar to mesh-adaptive h -refinement in finite elements. The technique allows the ROM to generate arbitrarily accurate solutions, thereby providing the ROM with a 'failsafe' mechanism in the event of insufficient training data. Finally, we present the reduced-order model error surrogate (ROMES) method for statistically quantifying reduced- order-model errors. This enables ROMs to be rigorously incorporated in uncertainty-quantification settings, as the error model can be treated as a source of epistemic uncertainty. This work was completed as part of a Truman Fellowship appointment. We note that much additional work was performed as part of the Fellowship. One salient project is the development of the Trilinos-based model-reduction software module Razor , which is currently bundled with the Albany PDE code and currently allows nonlinear reduced-order models to be constructed for any application supported in Albany. Other important projects include the following: 1. ROMES-equipped ROMs for Bayesian inference: K. Carlberg, M. Drohmann, F. Lu (Lawrence Berkeley National Laboratory), M. Morzfeld (Lawrence Berkeley National Laboratory). 2. ROM-enabled Krylov-subspace recycling: K. Carlberg, V. Forstall (University of Maryland), P. Tsuji, R. Tuminaro. 3. A pseudo balanced POD method using only dual snapshots: K. Carlberg, M. Sarovar. 4. An analysis of discrete v. continuous optimality in nonlinear model reduction: K. Carlberg, M. Barone, H. Antil (George Mason University). Journal articles for these projects are in progress at the time of this writing.

97 MATHEMATICS AND COMPUTING↗

Deep-inelastic and quasielastic electron scattering from A = 3 nuclei

We perform a combined analysis of inclusive electron scattering data from $A=3$ nuclei in the deep-inelastic and quasielastic scattering regions, using Monte Carlo analysis methods and the nuclear weak binding approximation to establish the range over which the data can be described within the same theoretical framework. Comparison with quasielastic $^3$He cross sections from SLAC and Jefferson Lab suggests that most features of the $x \gtrsim 1$ data can be reasonably well described in the impulse approximation with finite-$Q^2$ nuclear smearing functions for momentum transfers $Q^2 \gtrsim 1$~GeV$^2$. For the DIS region, we analyze the recent $^3$He to deuterium cross section ratio from the Jefferson Lab E03-103 experiment to explore the possible isospin dependence of the nuclear effects. We discuss the implications of this for the MARATHON experiment, and outline how a Bayesian analysis of $^3$He, $^3$H and deuterium data can robustly determine the free neutron structure function.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Efficient Stochastic Inversion Using Adjoint Models and Kernel-PCA

Performing stochastic inversion on a computationally expensive forward simulation model with a high-dimensional uncertain parameter space (e.g. a spatial random field) is computationally prohibitive even with gradient information provided. Moreover, the ‘nonlinear’ mapping from parameters to observables generally gives rise to non-Gaussian posteriors even with Gaussian priors, thus hampering the use of efficient inversion algorithms designed for models with Gaussian assumptions. In this work, we propose a novel Bayesian stochastic inversion methodology, characterized by a tight coupling between a gradient-based Langevin Markov Chain Monte Carlo (LMCMC) method and a kernel principal component analysis (KPCA). This approach addresses the ‘curse-of-dimensionality’ via KPCA to identify a low-dimensional feature space within the high-dimensional and nonlinearly correlated spatial random field. Moreover, non-Gaussian full posterior probability distribution functions are estimated via an efficient LMCMC method on both the projected low-dimensional feature space and the recovered high-dimensional parameter space. We demonstrate this computational framework by integrating and adapting recent developments such as data-driven statistics-on-manifolds constructions and reduction-through-projection techniques to solve inverse problems in linear elasticity.

97 MATHEMATICS AND COMPUTING↗

astroABC : An Approximate Bayesian Computation Sequential Monte Carlo sampler for cosmological parameter estimation

Given the complexity of modern cosmological parameter inference where we arefaced with non-Gaussian data and noise, correlated systematics and multi-probecorrelated data sets, the Approximate Bayesian Computation (ABC) method is apromising alternative to traditional Markov Chain Monte Carlo approaches in thecase where the Likelihood is intractable or unknown. The ABC method is called"Likelihood free" as it avoids explicit evaluation of the Likelihood by using aforward model simulation of the data which can include systematics. Weintroduce astroABC, an open source ABC Sequential Monte Carlo (SMC) sampler forparameter estimation. A key challenge in astrophysics is the efficient use oflarge multi-probe datasets to constrain high dimensional, possibly correlatedparameter spaces. With this in mind astroABC allows for massive parallelizationusing MPI, a framework that handles spawning of jobs across multiple nodes. Akey new feature of astroABC is the ability to create MPI groups with differentcommunicators, one for the sampler and several others for the forward modelsimulation, which speeds up sampling time considerably. For smaller jobs thePython multiprocessing option is also available. Other key features include: aSequential Monte Carlo sampler, a method for iteratively adapting tolerancelevels, local covariance estimate using scikit-learn's KDTree, modules forspecifying optimal covariance matrix for a component-wise or multivariatenormal perturbation kernel, output and restart files are backed up everyiteration, user defined metric and simulation methods, a module for specifyingheterogeneous parameter priors including non-standard prior PDFs, a module forspecifying a constant, linear, log or exponential tolerance level,well-documented examples and sample scripts. This code is hosted online athttps://github.com/EliseJ/astroABC

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Robust Multi-fidelity Bayesian Optimization with Deep Kernel and Partition

Multi-fidelity Bayesian optimization (MFBO) is a powerful approach that utilizes lowfidelity, cost-effective sources to expedite the exploration and exploitation of a high-fidelity objective function. Existing MFBO methods with theoretical foundations either lack justification for performance improvements over single-fidelity optimization or rely on strong assumptions about the relationships between fidelity sources to construct surrogate models and direct queries to low-fidelity sources. To mitigate the dependency on cross-fidelity assumptions while maintaining the advantages of low-fidelity queries, we introduce a random sampling and partition-based MFBO framework with deep kernel learning. This framework is robust to cross-fidelity model misspecification and explicitly illustrates the benefits of low-fidelity queries. Our results demonstrate that the proposed algorithm effectively manages complex cross-fidelity relationships and efficiently optimizes the target fidelity function.

Zhang, Fengxue [University of Chicago, Illinois, U↗

On unifying randomized methods for inverse problems

This work unifies the analysis of various randomized methods for solving linear and nonlinear inverse problems with Gaussian priors by framing the problem in a stochastic optimization setting. By doing so, we show that many randomized methods are variants of a sample average approximation (SAA). More importantly, we are able to prove a single theoretical result that guarantees the asymptotic convergence for a variety of randomized methods. Additionally, viewing randomized methods as an SAA enables us to prove, for the first time, a single non-asymptotic error result that holds for randomized methods under consideration. Another important consequence of our unified framework is that it allows us to discover new randomization methods. Here, we present various numerical results for linear, nonlinear, algebraic, and PDE-constrained inverse problems that verify the theoretical convergence results and provide a discussion on the apparently different convergence rates and the behavior for various randomized methods.

42 ENGINEERING↗

AI-optimized detector design for the future Electron-Ion Collider: the dual-radiator RICH case

Advanced detector R&D requires performing computationally intensive and detailed simulations as part of the detector-design optimization process. Here, we propose a general approach to this process based on Bayesian optimization and machine learning that encodes detector requirements. As a case study, we focus on the design of the dual-radiator Ring Imaging Cherenkov (dRICH) detector under development as a potential component of the particle-identification system at the future Electron-Ion Collider (EIC). The EIC is a US-led frontier accelerator project for nuclear physics, which has been proposed to further explore the structure and interactions of nuclear matter at the scale of sea quarks and gluons. We show that the detector design obtained with our automated and highly parallelized framework outperforms the baseline dRICH design within the assumptions of the current model. Our technique can be applied to any detector R&D, provided that realistic simulations are available.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Next-Level Energy Management in Manufacturing: Facility-Level Energy Digital Twin Framework Based on Machine Learning and Automated Data Collection

This research introduces an energy prediction framework at the facility level supported by automated data collection and machine learning models. It investigates whether reducing the prediction time scale allows for applying more complex machine learning techniques and if those techniques improve the prediction accuracy. The primary advantages of this framework lie in its automation of the energy prediction process and its provision of real-time energy data suitable for use in energy dashboards or digital twins. A sitewide dataset was created by combining 15 min energy and daily production data of five shops—assembly, battery, body (electric), body (gas), and paint—from a globally recognized electric vehicle manufacturer. Various machine learning models were evaluated on daily, weekly, and monthly datasets, including, in increasingly complex order: naïve, simple linear regression, net regularized generalized linear regression, principal component regression, k-nearest neighbor, random forest, and Bayesian regularized neural network. Compared to the current state-of-the-art energy consumption prediction for the industrial facility level, this research investigates more complex models and smaller time intervals for higher accuracy. The findings revealed that the more complex monthly models require a minimum of a year and a half of data to operate, while weekly models demand a year of data to achieve improved accuracy. Daily models can operate with only six months of data but exhibit poor performance due to reduced prediction accuracy of production. Key challenges identified include access to reliable, high-quality energy and production data and the initial demand for human labor.

digital twin↗

R2U2: Monitoring and Diagnosis of Security Threats for Unmanned Aerial Systems

We present R2U2, a novel framework for runtime monitoring of security properties and diagnosing of security threats on-board Unmanned Aerial Systems (UAS). R2U2, implemented in FPGA hardware, is a real-time, REALIZABLE, RESPONSIVE, UNOBTRUSIVE Unit for security threat detection. R2U2 is designed to continuously monitor inputs from the GPS and the ground control station, sensor readings, actuator outputs, and flight software status. By simultaneously monitoring and performing statistical reasoning, attack patterns and post-attack discrepancies in the UAS behavior can be detected. R2U2 uses runtime observer pairs for linear and metric temporal logics for property monitoring and Bayesian networks for diagnosis of security threats. We discuss the design and implementation that now enables R2U2 to handle security threats and present simulation results of several attack scenarios on the NASA DragonEye UAS.

Formal Methods↗

ASTROMETRY AND RADIAL VELOCITIES OF THE PLANET HOST M DWARF GJ 317: NEW TRIGONOMETRIC DISTANCE, METALLICITY, AND UPPER LIMIT TO THE MASS OF GJ 317b

We have obtained precision astrometry of the planet host M dwarf GJ 317 in the framework of the Carnegie Astrometric Planet Search project. The new astrometric measurements give a distance determination of 15.3 pc, 65% further than previous estimates. The resulting absolute magnitudes suggest that it is metal-rich and more massive than previously assumed. This result strengthens the correlation between high metallicity and the presence of gas giants around low-mass stars. At 15.3 pc, the minimal astrometric amplitude for planet candidate GJ 317b is 0.3 mas (edge-on orbit), just below our astrometric sensitivity. However, given the relatively large number of observations and good astrometric precision, a Bayesian Monte Carlo Markov Chain analysis indicates that the mass of planet b has to be smaller than twice the minimum mass with a 99% confidence level, with a most likely value of 2.5 M{sub Jup}. Additional radial velocity (RV) measurements obtained with Keck by the Lick-Carnegie Planet search program confirm the presence of an additional very long period planet candidate, with a period of 20 years or more. Even though such an object will imprint a large astrometric wobble on the star, its curvature is yet not evident in the astrometry. Given high metallicity, and the trend indicating that multiple systems are rich in low-mass companions, this system is likely to host additional low-mass planets in its habitable zone that can be readily detected with state-of-the-art optical and near-infrared RV measurements.

79 ASTRONOMY AND ASTROPHYSICS↗