Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “nonlinear regression”

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 379 records · Page 21

Model Development for MODIS Thermal Band Electronic Crosstalk

MODerate-resolution Imaging Spectroradiometer (MODIS) has 36 bands. Among them, 16 thermal emissive bands covering a wavelength range from 3.8 to 14.4 m. After 16 years on-orbit operation, the electronic crosstalk of a few Terra MODIS thermal emissive bands developed substantial issues that cause biases in the EV brightness temperature measurements and surface feature contamination. The crosstalk effects on band 27 with center wavelength at 6.7 m and band 29 at 8.5 m increased significantly in recent years, affecting downstream products such as water vapor and cloud mask. The crosstalk effect is evident in the near-monthly scheduled lunar measurements, from which the crosstalk coefficients can be derived. The development of an alternative approach is very helpful for independent verification.In this work, a physical model was developed to assess the crosstalk impact on calibration as well as in Earth view brightness temperature retrieval. This model was applied to Terra MODIS band 29 empirically to correct the Earth brightness temperature measurements. In the model development, the detectors nonlinear response is considered. The impact of the electronic crosstalk is assessed in two steps. The first step consists of determining the impact on calibration using the on-board blackbody (BB). Due to the detectors nonlinear response and large background signal, both linear and nonlinear coefficients are affected by the crosstalk from sending bands. The second step is to calculate the effects on the Earth view brightness temperature retrieval. The effects include those from affected calibration coefficients and the contamination of Earth view measurements. This model links the measurement bias with crosstalk coefficients, detector non-linearity, and the ratio of Earth measurements between the sending and receiving bands. The correction of the electronic cross talk can be implemented empirically from the processed bias at different brightness temperature. The implementation can be done through two approaches. As routine calibration assessment for thermal infrared bands, the trending over select Earth scenes is processed for all the detectors in a band and the band averaged bias is derived at a certain time. In this case, the correction of an affected band can be made using the regression of the model with band averaged bias and then corrections of detector differences are applied. The second approach requires the trending for individual detectors and the bias for each detector is used for regression with the model. A test using the first approach was made for Terra MODIS band 29 with the biases derived from long-term trending of brightness temperature over ocean and Dome-C.

Chang, Tiejun↗

Development of a GPU-based high-performance radiative transfer model for the Infrared Atmospheric Sounding Interferometer (IASI)

Satellite-observed radiance is a nonlinear functional of surface properties and atmospheric temperature and absorbing gas profiles as described by the radiative transfer equation (RTE). In the era of hyperspectral sounders with thousands of high-resolution channels, the computation of the radiative transfer model becomes more time-consuming. The radiative transfer model performance in operational numerical weather prediction systems still limits the number of channels we can use in hyperspectral sounders to only a few hundreds. To take the full advantage of such high-resolution infrared observations, a computationally efficient radiative transfer model is needed to facilitate satellite data assimilation. In recent years the programmable commodity graphics processing unit (GPU) has evolved into a highly parallel, multi-threaded, many-core processor with tremendous computational speed and very high memory bandwidth. The radiative transfer model is very suitable for the GPU implementation to take advantage of the hardware's efficiency and parallelism where radiances of many channels can be calculated in parallel in GPUs. In this paper, we develop a GPU-based high-performance radiative transfer model for the Infrared Atmospheric Sounding Interferometer (IASI) launched in 2006 onboard the first European meteorological polar-orbiting satellites, METOP-A. Each IASI spectrum has 8461 spectral channels. The IASI radiative transfer model consists of three modules. The first module for computing the regression predictors takes less than 0.004% of CPU time, while the second module for transmittance computation and the third module for radiance computation take approximately 92.5% and 7.5%, respectively. Our GPU-based IASI radiative transfer model is developed to run on a low-cost personal supercomputer with four GPUs with total 960 compute cores, delivering near 4 TFlops theoretical peak performance. By massively parallelizing the second and third modules, we reached 364x speedup for 1 GPU and 1455x speedup for all 4 GPUs, both with respect to the original CPU-based single-threaded Fortran code with the -O{sub 2} compiling optimization. The significant 1455x speedup using a computer with four GPUs means that the proposed GPU-based high-performance forward model is able to compute one day's amount of 1,296,000 IASI spectra within nearly 10 min, whereas the original single CPU-based version will impractically take more than 10 days. This model runs over 80% of the theoretical memory bandwidth with asynchronous data transfer. A novel CPU-GPU pipeline implementation of the IASI radiative transfer model is proposed. The GPU-based high-performance IASI radiative transfer model is suitable for the assimilation of the IASI radiance observations into the operational numerical weather forecast model.

97 MATHEMATICS AND COMPUTING↗

The M-dwarf Ultraviolet Spectroscopic Sample. I. Determining Stellar Parameters for Field Stars

Accurate stellar properties are essential for precise stellar astrophysics and exoplanetary science. In the M-dwarf regime, much effort has gone into defining empirical relations that can use readily accessible observables to assess physical stellar properties. Often, these relations for the quantity of interest are cast as a nonlinear function of available data; in Bayesian modeling, however, the reverse is needed. In this article, we introduce a new Bayesian framework to self-consistently and simultaneously apply multiple empirical calibrations to fully characterize the mass, luminosity, radius, and effective temperature of a field age M-dwarf. This framework includes a new M-dwarf mass–radius relation with a scatter of 3.1% at fixed mass. We further introduce the M-dwarf Ultraviolet Spectroscopic Sample (MUSS), and apply our methodology to provide consistent stellar parameters for these nearby low-mass stars, selected as having available spectroscopic data in the ultraviolet. These targets are of interest largely as either exoplanet hosts or benchmarks in multiwavelength stellar activity. We use the field MUSS stars to define a low-mass main sequence in the solar neighborhood through Gaussian Process (GP) regression. These results enable us to empirically measure a feature in the GP derivative at M ⊙ that indicates where the MUSS transitions from fully to partly convective interiors.

J. Sebastian Pineda↗

Analysis of calibration data for the uranium active neutron coincidence counting collar with attention to errors in the measured neutron coincidence rate

The declared linear density of 238 U and 235 U in fresh low enriched uranium light water reactor fuel assemblies can be verified for nuclear safeguards purposes using a neutron coincidence counter collar in passive and active mode, respectively. The active mode calibration of the Uranium Neutron Collar - Light water reactor fuel (UNCL) instrument is normally performed using a non-linear fitting technique. The fitting technique relates the measured neutron coincidence rate (the predictor) to the linear density of 235 U (the response) in order to estimate model parameters of the nonlinear Parte equation, which traditionally is used to model the calibration data. Alternatively, following a simple data transformation, the fitting can also be performed using standard linear fitting methods. This article compares performance of the nonlinear technique to the linear technique, using a range of possible error variance magnitudes in the measured neutron coincidence rate. We develop the required formalism and then apply the traditional (nonlinear) and alternative approaches (linear) to the same experimental and corresponding simulated representative datasets. Here, we find that, in this context, because of the magnitude of the errors in the predictor, it is preferable not to transform to a linear model, and it is preferable not to adjust for the errors in the predictor when inferring the model parameters.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

Colossal photon bunching in quasiparticle-mediated nanodiamond cathodoluminescence

Nanoscale control over the second-order photon correlation function g (2) (τ) is critical to emerging research in nonlinear nanophotonics and integrated quantum information science. Here we report on quasiparticle control of photon bunching with g (2) (0) > 45 in the cathodoluminescence of nanodiamond nitrogen vacancy (NV 0 ) centers excited by a converged electron beam in an aberration-corrected scanning transmission electron microscope. Plasmon-mediated NV 0 cathodoluminescence exhibits a 16-fold increase in luminescence intensity correlated with a threefold reduction in photon bunching compared with that of uncoupled NV 0 centers. This effect is ascribed to the excitation of single temporally uncorrelated NV 0 centers by single surface plasmon polaritons. Spectrally resolved Hanbury Brown–Twiss interferometry is employed to demonstrate that the bunching is mediated by the NV 0 phonon sidebands, while no observable bunching is detected at the zero-phonon line. As a result, the data are consistent with fast phonon-mediated recombination dynamics, a conclusion substantiated by agreement between Bayesian regression and Monte Carlo models of superthermal NV 0 luminescence.

74 ATOMIC AND MOLECULAR PHYSICS↗

Evaluating the Trustworthiness of Explainable Artificial Intelligence (XAI) Methods Applied to Regression Predictions of Arctic Sea Ice Motion

Abstract Recent advances in explainable artificial intelligence (XAI) methods show promise for understanding predictions made by machine learning (ML) models. XAI explains how the input features are relevant or important for the model predictions. We train linear regression (LR) and convolutional neural network (CNN) models to make 1-day predictions of sea ice velocity in the Arctic from inputs of present-day wind velocity and previous-day ice velocity and concentration. We apply XAI methods to the CNN and compare explanations to variance explained by LR. We confirm the feasibility of using a novel XAI method [i.e., global layerwise relevance propagation (LRP)] to understand ML model predictions of sea ice motion by comparing it to established techniques. We investigate a suite of linear, perturbation-based, and propagation-based XAI methods in both local and global forms. Outputs from different explainability methods are generally consistent in showing that wind speed is the input feature with the highest contribution to ML predictions of ice motion, and we discuss inconsistencies in the spatial variability of the explanations. Additionally, we show that the CNN relies on both linear and nonlinear relationships between the inputs and uses nonlocal information to make predictions. LRP shows that wind speed over land is highly relevant for predicting ice motion offshore. This provides a framework to show how knowledge of environmental variables (i.e., wind) on land could be useful for predicting other properties (i.e., sea ice velocity) elsewhere. Significance Statement Explainable artificial intelligence (XAI) is useful for understanding predictions made by machine learning models. Our research establishes trustability in a novel implementation of an explainable AI method known as layerwise relevance propagation for Earth science applications. To do this, we provide a comparative evaluation of a suite of explainable AI methods applied to machine learning models that make 1-day predictions of Arctic sea ice velocity. We use explainable AI outputs to understand how the input features are used by the machine learning to predict ice motion. Additionally, we show that a convolutional neural network uses nonlinear and nonlocal information in making its predictions. We take advantage of the nonlocality to investigate the extent to which knowledge of wind on land is useful for predicting sea ice velocity elsewhere.

Hoffman, Lauren [Scripps Institution of Oceanograp↗

Adding In-Plane Flexibility to the Equations of Motion of a Single Rotor Helicopter

This report describes a way to add the effects of main rotor blade flexibility in the in- plane or lead-lag direction to a large set of non-linear equations of motion for a single rotor helicopter with rigid blades(l). Differences between the frequency of the regressing lag mode predicted by the equations of (1) and that measured in flight (2) for a UH-60 helicopter indicate that some element is missing from the analytical model of (1) which assumes rigid blades. A previous study (3) noted a similar discrepancy for the CH-53 helicopter. Using a relatively simple analytical model in (3), compared to (1), it was shown that a mechanical lag damper increases significantly the coupling between the rigid lag mode and the first flexible mode. This increased coupling due to a powerful lag damper produces an increase in the lowest lag frequency when viewed in a frame rotating with the blade. Flight test measurements normally indicate the frequency of this mode in a non-rotating or fixed frame. This report presents the additions necessary to the full equations of motion, to include main rotor blade lag flexibility. Since these additions are made to a very complex nonlinear dynamic model, in order to provide physical insight, a discussion of the results obtained from a simplified set of equations of motion is included. The reduced model illustrates the physics involved in the coupling and should indicate trends in the full model.

Curtiss, H. C., Jr.↗

Practical Use of Computationally Frugal Model Analysis Methods

Computationally frugal methods of model analysis can provide substantial benefits when developing models of groundwater and other environmental systems. Model analysis includes ways to evaluate model adequacy and to perform sensitivity and uncertainty analysis. Frugal methods typically require 10s of parallelizable model runs; their convenience allows for other uses of the computational effort. We suggest that model analysis be posed as a set of questions used to organize methods that range from frugal to expensive (requiring 10,000 model runs or more). This encourages focus on method utility, even when methods have starkly different theoretical backgrounds. We note that many frugal methods are more useful when unrealistic process-model nonlinearities are reduced. Inexpensive diagnostics are identified for determining when frugal methods are advantageous. Examples from the literature are used to demonstrate local methods and the diagnostics. We suggest that the greater use of computationally frugal model analysis methods would allow questions such as those posed in this work to be addressed more routinely, allowing the environmental sciences community to obtain greater scientific insight from the many ongoing and future modeling efforts

54 ENVIRONMENTAL SCIENCES↗

Improved accuracy of cortical bone mineralization measured by polychromatic microcomputed tomography using a novel high mineral density composite calibration phantom

Purpose: Microcomputed tomography (micro-CT) is increasingly used as a nondestructive alternative to ashing for measuring bone mineral content. Phantoms are utilized to calibrate the measured x-ray attenuation to discrete levels of mineral density, typically including levels up to 1000 mg HA/cm{sup 3}, which encompasses levels of bone mineral density (BMD) observed in trabecular bone. However, levels of BMD observed in cortical bone and levels of tissue mineral density (TMD) in both cortical and trabecular bone typically exceed 1000 mg HA/cm{sup 3}, requiring extrapolation of the calibration regression, which may result in error. Therefore, the objectives of this study were to investigate (1) the relationship between x-ray attenuation and an expanded range of hydroxyapatite (HA) density in a less attenuating polymer matrix and (2) the effects of the calibration on the accuracy of subsequent measurements of mineralization in human cortical bone specimens. Methods: A novel HA-polymer composite phantom was prepared comprising a less attenuating polymer phase (polyethylene) and an expanded range of HA density (0-1860 mg HA/cm{sup 3}) inclusive of characteristic levels of BMD in cortical bone or TMD in cortical and trabecular bone. The BMD and TMD of cortical bone specimens measured using the new HA-polymer calibration phantom were compared to measurements using a conventional HA-polymer phantom comprising 0-800 mg HA/cm{sup 3} and the corresponding ash density measurements on the same specimens. Results: The HA-polymer composite phantom exhibited a nonlinear relationship between x-ray attenuation and HA density, rather than the linear relationship typically employed a priori, and obviated the need for extrapolation, when calibrating the measured x-ray attenuation to high levels of mineral density. The BMD and TMD of cortical bone specimens measured using the conventional phantom was significantly lower than the measured ash density by 19% (p<0.001, ANCOVA) and 33% (p<0.05, Tukey's HSD), on average, respectively. The BMD and TMD of cortical bone specimens measured using the HA-polymer phantom with an expanded range of HA density was significantly lower than the measured ash density by 8% (p<0.001, ANCOVA) and 10% (p<0.05, Tukey's HSD), on average, respectively. Conclusions: The new HA-polymer calibration phantom with a less attenuating polymer and an expanded range of HA density resulted in a more accurate measurement of micro-CT equivalent BMD and TMD in human cortical bone specimens compared to a conventional phantom, as verified by ash density measurements on the same specimens.

07 ISOTOPE AND RADIATION SOURCES↗

Lithium-Ion Battery Diagnostics Using Electrochemical Impedance via Machine-Learning

Diagnosing battery states such as health, state-of-charge, or temperature is crucial for ensuring the safety and reliability of electrochemical energy storage systems. While some states, such as temperature, may be measured using cheap sensors, accurate diagnosis of battery health metrics usually requires time-consuming performance measurements, making them infeasible for use in real-world operation. These health metrics can be measured during lab-testing and then estimated on-line using predictive life models or via state observer algorithms such as Kalman filters, but these predictive methods should be supplemented by actual measurement of battery health whenever possible to ensure reliability. Rapid measurement of battery health may be done by various types of fast diagnostic techniques such as electrochemical impedance spectroscopy (EIS), which can be performed in only a few minutes and require only a fraction of the energy and power needed for a full charge and discharge measurement. But there is a substantial challenge for estimating battery health using EIS data, as EIS is sensitive to cell temperature, state-of-charge, current, and resting time in addition to health. Thus, utilizing EIS data to predict battery capacity requires correcting for all these additional variables, a task that is extremely difficult to handle analytically. This talk utilizes machine-learning methods to estimate the effectiveness of battery capacity prediction from EIS data, leveraging a data set of hundreds of EIS measurements recorded at varying temperature and state-of-charge throughout a 500-day aging study of 32 commercial, large-format NMC-Graphite lithium-ion batteries. Using EIS as input to machine-learning models is complicated by the nonlinear response of impedance to battery health, temperature, and state-of-charge, as well as the collinearity between the impedance response at neighboring frequencies, which can easily lead to overfit models. To train robust models, features from EIS data need to be extracted from the data or some subset of critical frequencies selected. Many approaches for extracting and selecting features from EIS data from electrochemical analysis and machine-learning fields were identified for analysis: using the entire raw spectra; selection of one, two, or many frequencies from the entire spectra; selecting interesting points from the EIS measurement using domain knowledge; fitting EIS with an equivalent-circuit model; calculating statistics on the raw impedance values; and reducing the dimensionality of the data using unsupervised linear (principal component analysis) and non-linear (uniform manifold approximation and projection) methods. These approaches were rigorously compared using a machine-learning pipeline approach, training linear, Gaussian process, and random forest regression models and quantifying performance using cross-validation as well as a held-out test set. An artificial neural network model trained on the raw spectra was also tested. Promising pipelines were fine-tuned via Bayesian hyperparameter optimization using cross-validation loss and training with class-specific weights to counter data set imbalance. The most reliable method for utilizing impedance in this work was the selection of two optimal frequencies through an exhaustive search, resulting in about 2% mean absolute error on test data for both Gaussian process and random forest model architectures. Interrogation of a variety of models reveals critical frequencies of 100 Hz and 103 Hz for this data set, though the optimal set of frequencies is not necessarily intuitive, i.e., the best performing models are not simply those that use impedance at frequencies that have the highest correlation to the relative discharge capacity. The best performing model is an ensemble model, which is able to predict battery capacity with 1.9% mean absolute error for unseen cells using impedance recorded at a variety of temperatures and states-of-charge.

battery↗

Statistical analysis of SSME system data

A statistical methodology to enhance the Space Shuttle Main Engine (SSME) performance prediction accuracy is proposed. This methodology was to be used in conjunction with existing SSME performance prediction computer codes to improve parameter prediction accuracy and to quantify that accuracy. However, after a review of related literature, researchers concluded that the proposed problem required a coverage of areas such as linear and nonlinear system theory, measurement theory, statistics, and stochastic estimation. Since state space theory is the foundation for a more complete study of each of the before mentioned areas, these researchers chose to refocus emphasis to cover the more specialized topic of state vector estimation procedures. State vector estimation was also selected because of current and future concerns by NASA for SSME performance evaluation; i.e., there is a current interest in an improved evaluation procedure for actual SSME post flight performance as well as for post static test performance of a single SSME. A current investigation of analytical methods may be used to improve test stand failure detection. This paper considers the issue of post flight/test state variable reconstruction through the application of observations made on the output of the Space Shuttle propulsion system. Rogers used the Kalman filtering procedure to reconstruct the state variables of the Space Shuttle propulsion system. An objective of this paper is to give the general setup of the Kalman filter and its connection to linear regression. A second objective is to examine the reconstruction methodology for application to the reconstruction of the state vector of a single Space Shuttle Main Engine (SSME) by using static test firing data.

Temple, Enoch C.↗

Horizontal vestibuloocular reflex evoked by high-acceleration rotations in the squirrel monkey. I. Normal responses

The horizontal angular vestibuloocular reflex (VOR) evoked by high-frequency, high-acceleration rotations was studied in five squirrel monkeys with intact vestibular function. The VOR evoked by steps of acceleration in darkness (3,000 degrees /s(2) reaching a velocity of 150 degrees /s) began after a latency of 7.3 +/- 1.5 ms (mean +/- SD). Gain of the reflex during the acceleration was 14.2 +/- 5.2% greater than that measured once the plateau head velocity had been reached. A polynomial regression was used to analyze the trajectory of the responses to steps of acceleration. A better representation of the data was obtained from a polynomial that included a cubic term in contrast to an exclusively linear fit. For sinusoidal rotations of 0.5-15 Hz with a peak velocity of 20 degrees /s, the VOR gain measured 0.83 +/- 0.06 and did not vary across frequencies or animals. The phase of these responses was close to compensatory except at 15 Hz where a lag of 5.0 +/- 0.9 degrees was noted. The VOR gain did not vary with head velocity at 0.5 Hz but increased with velocity for rotations at frequencies of >/=4 Hz (0. 85 +/- 0.04 at 4 Hz, 20 degrees /s; 1.01 +/- 0.05 at 100 degrees /s, P < 0.0001). No responses to these rotations were noted in two animals that had undergone bilateral labyrinthectomy indicating that inertia of the eye had a negligible effect for these stimuli. We developed a mathematical model of VOR dynamics to account for these findings. The inputs to the reflex come from linear and nonlinear pathways. The linear pathway is responsible for the constant gain across frequencies at peak head velocity of 20 degrees /s and also for the phase lag at higher frequencies being less than that expected based on the reflex delay. The frequency- and velocity-dependent nonlinearity in VOR gain is accounted for by the dynamics of the nonlinear pathway. A transfer function that increases the gain of this pathway with frequency and a term related to the third power of head velocity are used to represent the dynamics of this pathway. This model accounts for the experimental findings and provides a method for interpreting responses to these stimuli after vestibular lesions.

NASA Program Biomedical Research and Countermeasur↗

Hybrid NN/SVM Computational System for Optimizing Designs

A computational method and system based on a hybrid of an artificial neural network (NN) and a support vector machine (SVM) (see figure) has been conceived as a means of maximizing or minimizing an objective function, optionally subject to one or more constraints. Such maximization or minimization could be performed, for example, to optimize solve a data-regression or data-classification problem or to optimize a design associated with a response function. A response function can be considered as a subset of a response surface, which is a surface in a vector space of design and performance parameters. A typical example of a design problem that the method and system can be used to solve is that of an airfoil, for which a response function could be the spatial distribution of pressure over the airfoil. In this example, the response surface would describe the pressure distribution as a function of the operating conditions and the geometric parameters of the airfoil. The use of NNs to analyze physical objects in order to optimize their responses under specified physical conditions is well known. NN analysis is suitable for multidimensional interpolation of data that lack structure and enables the representation and optimization of a succession of numerical solutions of increasing complexity or increasing fidelity to the real world. NN analysis is especially useful in helping to satisfy multiple design objectives. Feedforward NNs can be used to make estimates based on nonlinear mathematical models. One difficulty associated with use of a feedforward NN arises from the need for nonlinear optimization to determine connection weights among input, intermediate, and output variables. It can be very expensive to train an NN in cases in which it is necessary to model large amounts of information. Less widely known (in comparison with NNs) are support vector machines (SVMs), which were originally applied in statistical learning theory. In terms that are necessarily oversimplified to fit the scope of this article, an SVM can be characterized as an algorithm that (1) effects a nonlinear mapping of input vectors into a higher-dimensional feature space and (2) involves a dual formulation of governing equations and constraints. One advantageous feature of the SVM approach is that an objective function (which one seeks to minimize to obtain coefficients that define an SVM mathematical model) is convex, so that unlike in the cases of many NN models, any local minimum of an SVM model is also a global minimum.

Rai, Man Mohan↗

Deep Neural Networks are Adaptive to Function Regularity and Data Distribution in Approximation and Estimation

Deep learning has exhibited remarkable results across diverse areas. To understand its success, substantial research has been directed towards its theoretical foundations. Nev- ertheless, the majority of these studies examine how well deep neural networks can model functions with uniform regularities. In this paper, we explore a different angle: how deep neural networks can adapt to varying degrees of smoothness in functions and nonuni- form data distributions across different locations and scales. More precisely, we focus on a broad class of functions defined by nonlinear tree-based approximation methods. This class encompasses a range of function types, such as functions with uniform regularities and discontinuous functions. We develop nonparametric approximation and estimation theories for this class using deep ReLU networks. Our results show that deep neural networks are adaptive to the nonuniform smoothness of functions and nonuniform data distributions at different locations and scales. We apply our results to several function classes, and derive the corresponding approximation and generalization errors. The validity of our results is demonstrated through numerical experiments.

97 MATHEMATICS AND COMPUTING↗

Bayesian Analysis of Savings from Retrofit Projects

Estimates of savings from retrofit projects depend on statistical models, but because of the complicated analysis required to determine the uncertainty of the estimates, savings uncertainty is not often considered. Numerous simplified methods have been proposed to determine savings uncertainty, but in all but the simplest cases, these methods provide approximate results only. The objective of this paper is to show that Bayesian inference provides a consistent framework for estimating savings and savings uncertainty in retrofit projects. We review the mathematical background of Bayesian inference and Bayesian regression, and present two examples of estimating savings and savings uncertainty in retrofit projects. The first is a simple case where both baseline and post-retrofit monthly natural gas use can be modeled as a linear function of monthly heating degree days. The Efficiency Valuation Organization (EVO 2007) defines two methods of determining savings in such cases: reporting period savings, which is an estimate of the savings during the post-retrofit period; and normalized savings, which is an estimate of the savings that would be obtained during a typical year at the project site. For reporting period savings, classical statistical analysis provides exact analytic results for both savings and savings uncertainty in this case. We use Bayesian analysis to calculate reporting period savings and savings uncertainty and show that the results are identical to the analytical results. For normalized savings, the literature contains no exact expression for the uncertainty of normalized savings; we use Bayesian inference to calculate this quantity for the first time, and compare it with the result of an approximate formula that has been proposed. The second example concerns a problem where the baseline data exhibit nonlinearity and serial autocorrelation, both of which are common in real-world retrofit projects. No analytical solutions exist to determine savings or savings uncertainty in this situation, but several simplified formulas have been proposed. We model the data using a 5-parameter model with first-order autoregressive errors, and use Bayesian inference to develop distributions for the model parameters and for the reporting period savings, which allows us to determine the savings uncertainty. We find the energy savings to be about 5% lower than the result obtained by ignoring the autocorrelation. In addition, the Bayesian analysis finds the savings uncertainty to be narrower than the approximate uncertainty calculated using the simplified formula. These results show that Bayesian inference can be used to determine savings and savings uncertainty for a wide variety of real-world problems.

Im, Piljae↗

SU-F-T-118: Characterization of Change in Fractional Anisotropy After Radiation Therapy: Does Nearby Disruption Predict for White Matter Damage?

Purpose: We investigated predictors of fractional anisotropy (FA) change in the corticospinal white matter tract (CST) following radiation therapy (RT). Methods: Diffusion tensor imaging (DTI) is a non-invasive modality which models water diffusion properties. FA quantifies the extent of directional bias—a decrease indicates disrupted white matter integrity. Fifteen patients with high-grade glioma underwent DTI scans before, and ten months after RT to 59.4–60 Gy. The CST was segmented using an automated atlas-based algorithm on all DTI images. Treatment planning CT and DTI images were aligned using non-linear registration allowing for baseline FA, follow-up FA, and absorbed dose to be determined in each voxel. Relative FA change was dichotomized into a binary outcome using 25% decrease as cutoff. Three metrics were assessed as predictors: voxel dose, distance from the voxel to the center of the CST (Rc), and the number of neighboring voxels (Nadj from 0 to 26) with ≥25% decrease in FA. Logistic regression and the area under the receiver-operating characteristics curve (AUC) analysis were performed for each patient. Results: Median age of the cohort was 59 years (range: 40–85). The average number of voxels in the CST amongst all patients was 1181 (±172, SD). In logistic regression, the probability of FA change was highly associated with Nadj in all 15 patients with corresponding AUCs between 0.81 and 0.97. With all three metrics included in the logistic regression models, Nadj was highly significant (p<0.001) in all patients, voxel dose significant (p<0.05) in 3/15 patients, and Rc significant in 12/15 patients (p<0.05). Conclusion: The number of neighboring voxels with change in FA was the dominant predictor of FA change at any given voxel. This suggests that the microenvironment of surrounding white matter disruption after radiation therapy may drive local effects along a white matter tract. Pettersson and Cervino are funded by a Varian Medical Systems grant.

60 APPLIED LIFE SCIENCES↗

Nested polynomial trends for the improvement of Gaussian process-based predictors

The role of simulation keeps increasing for the sensitivity analysis and the uncertainty quantification of complex systems. Such numerical procedures are generally based on the processing of a huge amount of code evaluations. When the computational cost associated with one particular evaluation of the code is high, such direct approaches based on the computer code only, are not affordable. Surrogate models have therefore to be introduced to interpolate the information given by a fixed set of code evaluations to the whole input space. When confronted to deterministic mappings, the Gaussian process regression (GPR), or kriging, presents a good compromise between complexity, efficiency and error control. Such a method considers the quantity of interest of the system as a particular realization of a Gaussian stochastic process, whose mean and covariance functions have to be identified from the available code evaluations. In this context, this work proposes an innovative parametrization of this mean function, which is based on the composition of two polynomials. This approach is particularly relevant for the approximation of strongly non linear quantities of interest from very little information. After presenting the theoretical basis of this method, this work compares its efficiency to alternative approaches on a series of examples.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

MFNets: data efficient all-at-once learning of multifidelity surrogates as directed networks of information sources

We present an approach for constructing a surrogate from ensembles of information sources of varying cost and accuracy. The multifidelity surrogate encodes connections between information sources as a directed acyclic graph, and is trained via gradient-based minimization of a nonlinear least squares objective. While the vast majority of state-of-the-art assumes hierarchical connections between information sources, our approach works with flexibly structured information sources that may not admit a strict hierarchy. The formulation has two advantages: (1) increased data efficiency due to parsimonious multifidelity networks that can be tailored to the application; and (2) no constraints on the training data—we can combine noisy, non-nested evaluations of the information sources. Finally, numerical examples ranging from synthetic to physics-based computational mechanics simulations indicate the error in our approach can be orders-of-magnitude smaller, particularly in the low-data regime, than single-fidelity and hierarchical multifidelity approaches.

97 MATHEMATICS AND COMPUTING↗