SEARCH · Engineering Papers
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
The following paper is presented in viewgraph format and covers topics including: (1) linear state feedback control system; (2) Kalman filter state estimation; (3) relation between residual and stochastic part of output; (4) obtaining Kalman filter gain; (5) state estimation under unknown system model and unknown noises; and (6) relationship between filter Markov parameters and system Markov parameters.
The project is concerned with the development of decision-theoretic techniques to optimize the scientific return of planetary rovers. Planetary rovers are small unmanned vehicles equipped with cameras and a variety of sensors used for scientific experiments. They must operate under tight constraints over such resources as operation time, power, storage capacity, and communication bandwidth. Moreover, the limited computational resources of the rover limit the complexity of on-line planning and scheduling. We have developed a comprehensive solution to this problem that involves high-level tools to describe a mission; a compiler that maps a mission description and additional probabilistic models of the components of the rover into a Markov decision problem; and algorithms for solving the rover control problem that are sensitive to the limited computational resources and high-level of uncertainty in this domain.
One of the richest potential sources of insight into fundamental physics that LISA will be capable of observing is the inspiral of supermassive black hole binaries (BHBs). However, the data analysis challenge presented by the LISA data stream is quite unlike the situation for present day gravitational wave detectors. In order to make the precision measurements necessary to achieve LISA's science goals, the BHB signal must be distinguished from a data stream that not only contains instrumental noise, but potentially thousands of other signals as well, so that the "background" we wish to separate out to focus on the BHB signal is likely to be highly nonstationary and nongaussian, as well as being of scientific interest in its own right. In addition, whereas the theoretical templates that we calculate in order to ultimately estimate the parameters can afford to be somewhat inaccurate and still be effective for present day and near future detectors, this is not the case for LISA, and extremely high fidelity of the theoretical templates for high signal-to-noise signals will be required to prevent theoretical errors from dominating the parameter estimates. NVe, will describe efforts in the community of LISA data analysts to address the challenges regarding the specific issue of BHB signals. These efforts include using a Markov Chain Monte Carlo approach with the freedom to model the BHB and the other signals present in the data stream simultaneously, rather than trying to remove other signals and risk biasing the remaining data. The Mock LISA Data Challenge is a community of LISA scientists who generate rounds of simulated LISA noise with increasingly difficult signal content, and invite the LISA data analysis community to exercise their methods, or develop new methods, in an attempt to extract the parameters for the signals embedded in the mock data. In addition to practical approaches such ,is this to assess the level of parameter accuracy, one can apply the Fisher matrix formalism to assess both the statistical errors from noise and the theoretical errors
Adaptive psychoacoustic test methods, in which the next signal level depends on the response to the previous signal, are the most efficient for determining psychoacoustic thresholds of individual subjects. In many tests conducted in the NASA psychoacoustic labs, the goal is to determine thresholds representative of the general population. To do this economically, non-adaptive testing methods are used in which three or four subjects are tested at the same time with predetermined signal levels. This approach requires us to identify techniques for assessing the uncertainty in resulting group-average psychoacoustic thresholds. In this presentation we examine the Delta Method of frequentist statistics, the Generalized Linear Model (GLM), the Nonparametric Bootstrap, a frequentist method, and Markov Chain Monte Carlo Posterior Estimation and a Bayesian approach. Each technique is exercised on a manufactured, theoretical dataset and then on datasets from two psychoacoustics facilities at NASA. The Delta Method is the simplest to implement and accurate for the cases studied. The GLM is found to be the least robust, and the Bootstrap takes the longest to calculate. The Bayesian Posterior Estimate is the most versatile technique examined because it allows the inclusion of prior information.
The management of large-scale distributed information systems relies on the effective use and modeling of monitoring data collected at various points in the distributed information systems. A promising approach is to discover invariant relationships among the monitoring data and generate invariant networks, where a node is a monitoring data source (metric) and a link indicates an invariant relationship between two monitoring data. Such an invariant network representation can help system experts to localize and diagnose the system faults by examining those broken invariant relationships and their related metrics, because system faults usually propagate among the monitoring data and eventually lead to some broken invariant relationships. However, at one time, there are usually a lot of broken links (invariant relationships) within an invariant network. Without proper guidance, it is difficult for system experts to manually inspect this large number of broken links. Thus, a critical challenge is how to effectively and efficiently rank metrics (nodes) of invariant networks according to the anomaly levels of metrics. The ranked list of metrics will provide system experts with useful guidance for them to localize and diagnose the system faults. To this end, we propose to model the nodes and the broken links as a Markov Random Field (MRF), and develop an iteration algorithm to infer the anomaly of each node based on belief propagation (BP). Finally, we validate the proposed algorithm on both realworld and synthetic data sets to illustrate its effectiveness.
Here, the Davis equation of state (EOS) is commonly used to model thermodynamic relationships for high explosive (HE) reactants. Typically, the parameters in the EOS are calibrated, with uncertainty, using a Bayesian framework and Markov Chain Monte Carlo (MCMC) methods. However, MCMC methods are computationally expensive, especially for complex models with many parameters. This paper provides a comparison between MCMC and less computationally expensive Variational methods (Variational Bayesian and Hessian Variational Bayesian) for computing the posterior distribution and approximating the posterior covariance matrix based on heterogeneous experimental data. All three methods recover similar posterior distributions and posterior covariance matrices. This study demonstrates that for this EOS parameter calibration application, the assumptions made in the two Variational methods significantly reduce the computational cost but do not substantially change the results compared to MCMC.
At a distance of ∼2 pc, our nearest brown dwarf neighbor, Luhman 16 AB, has been extensively studied since its discovery 3 years ago, yet its most fundamental parameter—the masses of the individual dwarfs—has not been constrained with precision. In this work, we present the full astrometric orbit and barycentric motion of Luhman 16 AB and the first precision measurements of the individual component masses. We draw upon archival observations spanning 31 years from the European Southern Observatory (ESO) Schmidt Telescope, the Deep Near-Infrared Survey of the Southern Sky (DENIS), public FORS2 data on the Very Large Telescope (VLT), and new astrometry from the Gemini South Multiconjugate Adaptive Optics System (GeMS). Finally, we include three radial velocity measurements of the two components from VLT/CRIRES, spanning one year. With this new data sampling a full period of the orbit, we use a Markov chain Monte Carlo algorithm to fit a 16-parameter model incorporating mutual orbit and barycentric motion parameters and constrain the individual masses to be 27.9{sub −1.0}{sup +1.1} M{sub J} for the T dwarf and 34.2{sub −1.1}{sup +1.3} M{sub J} for the L dwarf. Our measurements of Luhman 16 AB’s mass ratio and barycentric motion parameters are consistent with previous estimates in the literature utilizing recent astrometry only. The GeMS-derived measurements of the Luhman 16 AB separation in 2014–2015 agree closely with Hubble Space Telescope (HST) measurements made during the same epoch, and the derived mutual orbit agrees with those measurements to within the HST uncertainties of 0.3–0.4 mas.
Measuring the masses of protoplanetary disks is crucial for understanding their planet-forming potential. Typically, dust masses are derived from (sub-)millimeter flux density measurements plus assumptions for the opacity, temperature, and optical depth of the dust. Here we use radiative transfer models to quantify the validity of these assumptions with the aim of improving the accuracy of disk dust mass measurements. We first carry out a controlled exploration of disk parameter space. We find that the disk temperature is a strong function of disk size, while the optical depth depends on both disk size and dust mass. The millimeter-wavelength spectral index can be significantly shallower than the naive expectation due to a combination of optical depth and deviations from the Rayleigh–Jeans regime. We fit radiative transfer models to the spectral energy distributions (SEDs) of 132 disks in the Taurus-Auriga region using a Markov chain Monte Carlo approach. We used all available data to produce the most complete SEDs used in any extant modeling study. We perform the fitting twice: first with unconstrained disk sizes and again imposing the disk size–brightness relation inferred for sources in Taurus. This constraint generally forces the disks to be smaller, warmer, and more optically thick. From both sets of fits, we find disks to be ∼1–5 times more massive than when derived using (sub-)millimeter measurements and common assumptions. With the uncertainties derived from our model fitting, the previously measured dust mass–stellar mass correlation is present in our study but only significant at the 2σ level.
Pade approximations for linear and dynamic programming using Markov processes and new algorithms
This study introduces a non-intrusive approach in the context of low-rank separated representation to construct a surrogate of high-dimensional stochastic functions, e.g., PDEs/ODEs, in order to decrease the computational cost of Markov Chain Monte Carlo simulations in Bayesian inference. The surrogate model is constructed via a regularized alternative least-square regression with Tikhonov regularization using a roughening matrix computing the gradient of the solution, in conjunction with a perturbation-based error indicator to detect optimal model complexities. The model approximates a vector of a continuous solution at discrete values of a physical variable. The required number of random realizations to achieve a successful approximation linearly depends on the function dimensionality. The computational cost of the model construction is quadratic in the number of random inputs, which potentially tackles the curse of dimensionality in high-dimensional stochastic functions. Furthermore, this vector-valued separated representation-based model, in comparison to the available scalar-valued case, leads to a significant reduction in the cost of approximation by an order of magnitude equal to the vector size. The performance of the method is studied through its application to three numerical examples including a 41-dimensional elliptic PDE and a 21-dimensional cavity flow.
Neutrinos can be used as a unique and informative probe of the structure within nucleons. This thesis presents the tools and methodology for studying the strange quark spin in the nucleon through neutral current elastic neutrino-proton scattering in the MicroBooNE experiment located at the Fermi National Accelerator Lab. An automated boosted decision tree based proton identification algorithm is used to select proton in liquid argon TPC data with a 70\% efficiency. After a set of protons are selected, a logistic regression model is used to select neutral current elastic proton interactions in MicroBooNE. Neutral current elastic proton interactions are selected with an 11\% efficiency and 30\% purity. The number of selected events measured in MicroBooNE data as a function of $Q^2$ is compared directly to the number of selected events in MicroBooNE simulation. An event reweighting scheme is used to vary the expected number of eventsin MicroBooNE simulation based on the values o f the physics parameters of interest. The likelihood for any given value of the physical parameters can then be easily calculated and the probability distributions of the physics parameters can be sampled using Markov Chain Monte Carlo. Future improvements to the detector physics models in MicroBooNE are required for a precise determination of the strange quark spin structure in the proton. With the current level of uncertainty on the detector physics models, we are only able to constrain the net strange quark spin, $\Delta s$ to the range $-1.8 < \Delta s < 3.8$ with 95\% confidence.
The fibrous shape and biopersistence of multi-walled carbon nanotubes (MWCNT) have raised concern over their potential toxicity after pulmonary exposure. As in vivo exposure to MWCNT produced a transient inflammatory and progressive fibrotic response, this study sought to identify significant biological processes associated with lung inflammation and fibrosis pathology data, based upon whole genome mRNA expression, bronchoaveolar lavage scores, and morphometric analysis from C57BL/6J mice exposed by pharyngeal aspiration to 0, 10, 20, 40, or 80 μg MWCNT at 1, 7, 28, or 56 days post-exposure. Using a novel computational model employing non-negative matrix factorization and Monte Carlo Markov Chain simulation, significant biological processes with expression similar to MWCNT-induced lung inflammation and fibrosis pathology data in mice were identified. A subset of genes in these processes was determined to be functionally related to either fibrosis or inflammation by Ingenuity Pathway Analysis and was used to determine potential significant signaling cascades. Two genes determined to be functionally related to inflammation and fibrosis, vascular endothelial growth factor A (vegfa) and C-C motif chemokine 2 (ccl2), were confirmed by in vitro studies of mRNA and protein expression in small airway epithelial cells exposed to MWCNT as concordant with in vivo expression. This study identified that the novel computational model was sufficient to determine biological processes strongly associated with the pathology of lung inflammation and fibrosis and could identify potential toxicity signaling pathways and mechanisms of MWCNT exposure which could be used for future animal studies to support human risk assessment and intervention efforts. - Highlights: • A novel computational model identified toxicity pathways matching in vivo pathology. • Systematic identification of MWCNT-induced biological processes in mouse lungs • MWCNT-induced functional networks of lung inflammation and fibrosis were revealed. • Two functional, representative genes, ccl2 and vegfa, were validated in vitro.
This paper presents an algorithm to identify a state-space model of a linear system using a backward-time approach. The procedure consists of three basic steps. First, the Markov parameters of a backward-time observer are computed from experimental input-output data. Second, the backward-time observer Markov parameters are decomposed to obtain the backward-time system Markov parameters (backward-time pulse response samples) from which a backward-time state-space model is realized using the Eigensystem Realization Algorithm. Third, the obtained backward-time state space model is converted to the usual forward-time representation. Stochastic properties of this approach will be discussed. Experimental results are given to illustrate when and to what extent this concept works.
We present a novel algorithm, FAST-PT, for performing convolution or mode-coupling integrals that appear in nonlinear cosmological perturbation theory. The algorithm uses several properties of gravitational structure formation—the locality of the dark matter equations and the scale invariance of the problem—as well as Fast Fourier Transforms to describe the input power spectrum as a superposition of power laws. This yields extremely fast performance, enabling mode-coupling integral computations fast enough to embed in Monte Carlo Markov Chain parameter estimation. We describe the algorithm and demonstrate its application to calculating nonlinear corrections to the matter power spectrum, including one-loop standard perturbation theory and the renormalization group approach. We also describe our public code (in Python) to implement this algorithm. The code, along with a user manual and example implementations, is available at https://github.com/JoeMcEwen/FAST-PT.
This modeling archive supports the manuscript submitted for publication in the Environmental Modeling and Software. This study is supported by ORNL-SFA and IDEAS-Watershed. This study aims to improve calibration of complex biogeochemical models using datasets from multiple experiments targeting specific subprocesses. The proposed Bayesian joint-fitting scheme calibrates the entire biogeochemical model in one go using all the available datasets and estimate parameter uncertainties using Markov Chain Monte Carlo (MCMC). This allows for complete propagation of uncertainties and utilization of the information shared between different datasets. Mapping joint distribution of parameters guides model improvement by identifying null spaces in the parameter space. This archive contains files used to perform MCMC, post-process outputs and visualize results.
We derive a stochastic model for the surface diffusion of molecules, starting from the classical equations of motion for an N-atom molecule on a surface. The equation of motion becomes a generalized Langevin equation for the center of mass of the molecule, with a non-Markovian friction kernel. In the Markov approximation, a standard Langevin equation is recovered, and the effect of the molecular vibrations on the diffusion is seen to lead to an increase in the friction for center of mass motion. This effective friction has a simple form that depends on the curvature of the lowest energy diffusion path in the 3N-dimensional coordinate space. We also find that so long as the intramolecular forces are sufficiently strong, memory effects are usually not significant and the Markov approximation can be employed, resulting in a simple one-dimensional model that can account for the effect of the dynamics of the molecular vibrations on the diffusive motion.
We demonstrate a Bayesian method that can be used to calibrate computationally expensive 3D RANS (Reynolds Av- eraged Navier Stokes) models with complex response surfaces. Such calibrations, conditioned on experimental data, can yield turbulence model parameters as probability density functions (PDF), concisely capturing the uncertainty in the parameter estimates. Methods such as Markov chain Monte Carlo (MCMC) estimate the PDF by sampling, with each sample requiring a run of the RANS model. Consequently a quick-running surrogate is used instead to the RANS simulator. The surrogate can be very difficult to design if the model's response i.e., the dependence of the calibration variable (the observable) on the parameter being estimated is complex. We show how the training data used to construct the surrogate can be employed to isolate a promising and physically realistic part of the parameter space, within which the response is well-behaved and easily modeled. We design a classifier, based on treed linear models, to model the "well-behaved region". This classifier serves as a prior in a Bayesian calibration study aimed at estimating 3 k - ε parameters ( C μ , C ε2 , C ε1 ) from experimental data of a transonic jet-in-crossflow interaction. The robustness of the calibration is investigated by checking its predictions of variables not included in the cal- ibration data. We also check the limit of applicability of the calibration by testing at off-calibration flow regimes. We find that calibration yield turbulence model parameters which predict the flowfield far better than when the nomi- nal values of the parameters are used. Substantial improvements are still obtained when we use the calibrated RANS model to predict jet-in-crossflow at Mach numbers and jet strengths quite different from those used to generate the ex- perimental (calibration) data. Thus the primary reason for poor predictive skill of RANS, when using nominal values of the turbulence model parameters, was parametric uncertainty, which was rectified by calibration. Post-calibration, the dominant contribution to model inaccuraries are due to the structural errors in RANS.
Motivated by Planck confirmation of an anomalously low value of the CMB temperature fluctuations up to multipole ℓ < 40, we in this paper try to explain such feature by investigating case of punctuated inflation scenario. This form of inflation potential is inspired by Minimal Super-symmetric Standard Model (MSSM) wherein suppression of curvature perturbation power at large scales is produced by introducing period of fast-roll phase of the inflation sandwiched between two stages of slow-roll phase. We apply Markov Chain Monte Carlo analysis to determine posterior distribution and the best fit values of the model parameters using recent WMAP9 and Planck data. We show that WMAP9 and Planck results are consistent with each other and that with Planck data we obtain tighter constraints for punctuated inflation parameters. We find that punctuated inflation leads to better fit in CMB data compared to simple power law model. The improvement in the fit to the WMAP9 data is Δ χ{sup 2} ∼ 3.6 and for Planck the improvement is Δ χ{sup 2} ∼ 5.4. We find that AIC does not discriminate between punctuated inflation and simple power law model for WMAP9 data. However, for Planck data we find that punctuated inflation is moderately preferred over a simple power law model.