Search NASA⌕ Search

SEARCH · Search NASA

Results for “Inverse problem”

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 649 records · Page 36

Improvements in surface singularity analysis and design methods

The coupling of the combined source vortex distribution of Green's potential flow function with contemporary numerical techniques is shown to provide accurate, efficient, and stable solutions to subsonic inviscid analysis and design problems for multi-element airfoils. The analysis problem is solved by direct calculation of the surface singularity distribution required to satisfy the flow tangency boundary condition. The design or inverse problem is solved by an iteration process. In this process, the geometry and the associated pressure distribution are iterated until the pressure distribution most nearly corresponding to the prescribed design distribution is obtained. Typically, five iteration cycles are required for convergence. A description of the analysis and design method is presented, along with supporting examples.

Bristow, D. R.↗

On the identification of a harmonic force on a viscoelastic plate from response data

The problem of determining the force acting on a structure from measurements of the response of the structure to the force is an inverse problem. Presented is a method for determining the location, magnitude, and phase of a harmonic point force acting on a simply-supported classical viscoelastic rectangular plate from a number of displacement readings at discrete points on the plate. Presented also is a demonstration of the robustness of the solution technique to the effects of measurement noise as well as a means by which problems involving more general structural and loading configurations may be solved.

D'Cruz, J.↗

Model Calibration for Cancer Risk Projections According to Uncertain Data

This paper presents forward and inverse formulations for the calibration of computational models according to uncertain data. Uncertainty in the data might be caused by a poor metrology system, measurement noise, missing or uncontrollable input variables, or by the inability to directly measure the inputs and/or outputs of interest. The forward approach performs the calibration in the space of the model’s output thereby requiring repeated model simulations. Conversely, the inverse approach leverages an ensemble of solutions to an inverse problem in order to perform the calibration in the space of the model’s parameters. As such, the computational demands of the inverse approach are considerably lower. These strategies are applied to the calibration of a radiation model that in-forms cancer risk projections for future deep space missions.

uncertainty quantification↗

Stochastic gradient descent algorithm for stochastic optimization in solving analytic continuation problems

We propose a stochastic gradient descent based optimization algorithm to solve the analytic continuation problem in which we extract real frequency spectra from imaginary time Quantum Monte Carlo data. The procedure of analytic continuation is an ill-posed inverse problem which is usually solved by regularized optimization methods, such like the Maximum Entropy method, or stochastic optimization methods. The main contribution of this work is to improve the performance of stochastic optimization approaches by introducing a supervised stochastic gradient descent algorithm to solve a flipped inverse system which processes the random solutions obtained by a type of Fast and Efficient Stochastic Optimization Method.

97 MATHEMATICS AND COMPUTING↗

Maximum likelihood techniques applied to quasi-elastic light scattering

There is a necessity of having an automatic procedure for reliable estimation of the quality of the measurement of particle size from QELS (Quasi-Elastic Light Scattering). Getting the measurement itself, before any error estimates can be made, is a problem because it is obtained by a very indirect measurement of a signal derived from the motion of particles in the system and requires the solution of an inverse problem. The eigenvalue structure of the transform that generates the signal is such that an arbitrarily small amount of noise can obliterate parts of any practical inversion spectrum. This project uses the Maximum Likelihood Estimation (MLE) as a framework to generate a theory and a functioning set of software to oversee the measurement process and extract the particle size information, while at the same time providing error estimates for those measurements. The theory involved verifying a correct form of the covariance matrix for the noise on the measurement and then estimating particle size parameters using a modified histogram approach.

Edwards, Robert V.↗

Estimating parameters and uncertainty for three-dimensional flow and transport in a highly heterogeneous sand box experiment.

Heterogeneity plays an important role in groundwater flow and contaminant transport in natural systems. Since it is impossible to directly measure spatial variability of hydraulic conductivity, predictions of solute transport based on mathematical models are always uncertain. While in most cases groundwater flow and tracer transport problems are investigated in two-dimensional (2D) systems, it is important to study more realistic and well-controlled 3D systems to fully evaluate inverse parameter estimation techniques and evaluate uncertainty in the resulting estimates. We used tracer concentration breakthrough curves (BTCs) obtained from a magnetic resonance imaging (MRI) technique in a small flow cell (14 x 8 x 8 cm) that was packed with a known pattern of five different sands (i.e., zones) having cm-scale variability. In contrast to typical inversion systems with head, conductivity and concentration measurements at limited points, the MRI data included BTCs measured at a voxel scale ({approx}0.2 cm in each dimension) over 13 x 8 x 8 cm with a well controlled boundary condition, but did not have direct measurements of head and conductivity. Hydraulic conductivity and porosity were conceptualized as spatial random fields and estimated using pilot points along layers of the 3D medium. The steady state water flow and solute transport were solved using MODFLOW and MODPATH. The inversion problem was solved with a nonlinear parameter estimation package - PEST. Two approaches to parameterization of the spatial fields are evaluated: (1) The detailed zone information was used as prior information to constrain the spatial impact of the pilot points and reduce the number of parameters; and (2) highly parameterized inversion at cm scale (e.g., 1664 parameters) using singular value decomposition (SVD) methodology to significantly reduce the run-time demands. Both results will be compared to measured BTCs. With MRI, it is easy to change the averaging scale of the observed concentration from point to cross-section. This comparison allows us to evaluate which method best matches experimental results at different scales. To evaluate the uncertainty in parameter estimation, the null space Monte Carlo method will be used to reduce computational burden of the development of calibration-constrained Monte Carlo based parameter fields. This study will illustrate how accurately a well-calibrated model can predict contaminant transport.

58 GEOSCIENCES↗

Dynamic Shape Reconstruction of Three-Dimensional Frame Structures Using the Inverse Finite Element Method

A robust and efficient computational method for reconstructing the three-dimensional displacement field of truss, beam, and frame structures, using measured surface-strain data, is presented. Known as shape sensing , this inverse problem has important implications for real-time actuation and control of smart structures, and for monitoring of structural integrity. The present formulation, based on the inverse Finite Element Method (iFEM), uses a least-squares variational principle involving strain measures of Timoshenko theory for stretching, torsion, bending, and transverse shear. Two inverse-frame finite elements are derived using interdependent interpolations whose interior degrees-of-freedom are condensed out at the element level. In addition, relationships between the order of kinematic-element interpolations and the number of required strain gauges are established. As an example problem, a thin-walled, circular cross-section cantilevered beam subjected to harmonic excitations in the presence of structural damping is modeled using iFEM; where, to simulate strain-gauge values and to provide reference displacements, a high-fidelity MSC/NASTRAN shell finite element model is used. Examples of low and high-frequency dynamic motion are analyzed and the solution accuracy examined with respect to various levels of discretization and the number of strain gauges.

Gherlone, Marco↗

Self-absorption correction on 2D X-ray fluorescence maps

Abstract X-ray fluorescence mapping (XRF) is a highly efficient and non-invasive technique for quantifying material composition with micro and nanoscale spatial resolutions. Quantitative XRF analysis, however, confronts challenges from the long-lasting problem called self-absorption. Moreover, correcting two-dimensional XRF mapping datasets is particularly difficult because it is an ill-posed inverse problem. Here we report a semi-empirical method that can effectively correct 2D XRF mapping data. The correction error is generally less than 10% from a comprehensive evaluation of the accuracy in various configurations. The proposed method was applied to quantify the composition distribution around the grain boundaries in an electrochemically corroded stainless steel sample. Highly localized Cr enrichment was found around the crack sites, which was invisible before the absorption correction.

36 MATERIALS SCIENCE↗

SU-E-J-174: Adaptive PET-Based Dose Painting with Tomotherapy

Purpose: PET imaging can be converted into dose prescription directly. Due to the variability of the intensity of PET the image, PET prescription maybe superior over uniform dose prescription. Furthermore, unlike the case in image reconstruction of not knowing the image solution in advance, the prescribed dose is known from a PET image a priori. Therefore, optimum beam orientations are derivable. Methods: We can assume the PET image to be the prescribed dose and invert it to determine the energy fluence. The same method used to reconstruct tissue images from projections could be used to solve the inverse problem of determining beam orientations and modulation patterns from a dose prescription [10]. Unlike standard tomographic reconstruction of images from measured projection profiles, the inversion of the prescribed dose results in photon fluence which may be negative and therefore unphysical. Two-dimensional modulated beams can be modelled in terms of the attenuated or exponential radon transform of the prescribed dose function (assumed to be the PET image in this case), an application of a Ram-Lak filter, and inversion by backprojection. Unlike the case in PET processing, however, the filtered beam obtained from the inversion represents a physical photon fluence. Therefore, a positivity constraint for the fluence (setting negative fluence to zero) must be applied (Brahme et al 1982, Bortfeld et al 1990) Results: Truncating the negative profiles from the PET data results in an approximation of the derivable energy fluence. Backprojection of the deliverable fluence is an approximation of the dose delivered. The deliverable dose is comparable to the original PET image and is similar to the PET image. Conclusion: It is possible to use the PET data or image as a direct indicator of deliverable fluence for cylindrical radiotherapy systems such as TomoTherapy.

60 APPLIED LIFE SCIENCES↗

Mapping the fluid flow and shear near the core surface using the radial and horizontal components of the magnetic field

The problem of calculating the temporal evolution of both the radial and horizontal poloidal components of a field, given an initial field and the flow and shear, is first considered. Attention is then given to the inverse problem of determining the flow and shear, given an initial field and its temporal evolution. The nonuniqueness inherent in such inversions is discussed, and it is shown that part of the nonuniqueness in the shear is closely related to that in the flow derived from just the radial induction equation.

Jackson, Andrew↗

Reconstructing parton distribution functions from Ioffe time data: from Bayesian methods to neural networks

The computation of the parton distribution functions (PDF) or distribution amplitudes (DA) of hadrons from first principles lattice QCD constitutes a central open problem in high energy nuclear physics. In this study, we present and evaluate the efficiency of several numerical methods, well established in the study of inverse problems, to reconstruct the full x-dependence of PDFs. Our starting point are the so called Ioffe time PDFs, which are accessible from Euclidean time simulations in conjunction with a matching procedure. Using realistic mock data tests, we find that the ill-posed incomplete Fourier transform underlying the reconstruction requires careful regularization, for which both the Bayesian approach as well as neural networks are efficient and flexible choices.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

BAYESIAN MAGNETOHYDRODYNAMIC SEISMOLOGY OF CORONAL LOOPS

We perform a Bayesian parameter inference in the context of resonantly damped transverse coronal loop oscillations. The forward problem is solved in terms of parametric results for kink waves in one-dimensional flux tubes in the thin tube and thin boundary approximations. For the inverse problem, we adopt a Bayesian approach to infer the most probable values of the relevant parameters, for given observed periods and damping times, and to extract their confidence levels. The posterior probability distribution functions are obtained by means of Markov Chain Monte Carlo simulations, incorporating observed uncertainties in a consistent manner. We find well-localized solutions in the posterior probability distribution functions for two of the three parameters of interest, namely the Alfven travel time and the transverse inhomogeneity length scale. The obtained estimates for the Alfven travel time are consistent with previous inversion results, but the method enables us to additionally constrain the transverse inhomogeneity length scale and to estimate real error bars for each parameter. When observational estimates for the density contrast are used, the method enables us to fully constrain the three parameters of interest. These results can serve to improve our current estimates of unknown physical parameters in coronal loops and to test the assumed theoretical model.

79 ASTRONOMY AND ASTROPHYSICS↗

Machine learning the spectral function of a hole in a quantum antiferromagnet

Understanding charge motion in a background of interacting quantum spins is a fundamental problem in quantum many-body physics. The most extensively studied model for this problem is the so-called t-t'-t''-J model, where the determination of the parameter t' in the context of cuprate superconductors is challenging. Here we present a theoretical study of the spectral functions of a mobile hole in the t-t'-t''-J model using two machine-learning techniques: K-nearest neighbor regression (KNN) and a feed-forward neural network (FFNN). We employ the self-consistent Born approximation to generate a dataset of about 1.3 x 10 5 spectral functions. Here we show that, for the forward problem, both methods allow for the accurate and efficient prediction of spectral functions, allowing, e.g., rapid searches through parameter space. Furthermore, we find that for the inverse problem (inferring Hamiltonian parameters from spectra), the FFNN can, but the KNN cannot, accurately predict the model parameters using merely the density of states. Our results suggest that it may be possible to use deep-learning methods to predict materials parameters from experimentally measured spectral functions.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Extreme-Scale Bayesian Inference for Uncertainty Quantification of Complex Simulations

Uncertainty quantification (UQ)—that is, quantifying uncertainties in complex mathematical models and their large-scale computational implementations—is widely viewed as one of the outstanding challenges facing the field of CS&E over the coming decade. The EUREKA project set to address the most difficult class of UQ problems: those for which both the underlying PDE model as well as the uncertain parameters are of extreme scale. In the project we worked on these extreme-scale challenges in the following four areas: 1. Scalable parallel algorithms for sampling and characterizing the posterior distribution that exploit the structure of the underlying PDEs and parameter-to-observable map. These include structure-exploiting versions of the randomized maximum likelihood method, which aims to overcome the intractability of employing conventional MCMC methods for solving extreme-scale Bayesian inversion problems by appealing to and adapting ideas from large-scale PDE-constrained optimization, which have been very successful at exploring high-dimensional spaces. 2. Scalable parallel algorithms for construction of prior and likelihood functions based on learning methods and non-parametric density estimation. Constructing problem-specific priors remains a critical challenge in Bayesian inference, and more so in high dimensions. Another challenge is construction of likelihood functions that capture unmodeled couplings between observations and parameters. We will create parallel algorithms for non-parametric density estimation using high dimensional N-body methods and combine them with supervised learning techniques for the construction of priors and likelihood functions. 3. Bayesian inadequacy models, which augment physics models with stochastic models that represent their imperfections. The success of the Bayesian inference framework depends on the ability to represent the uncertainty due to imperfections of the mathematical model of the phenomena of interest. This is a central challenge in UQ, especially for large-scale models. We propose to develop the mathematical tools to address these challenges in the context of extreme-scale problems. 4. Parallel scalable algorithms for Bayesian optimal experimental design (OED). Bayesian inversion yields quantified uncertainties in the model parameters, which can be propagated forward through the model to yield uncertainty in outputs of interest. This opens the way for designing new experiments to reduce the uncertainties in the model parameters and model predictions. Such experimental design problems have been intractable for large-scale problems using conventional methods; we will create OED algorithms that exploit the structure of the PDE model and the parameter-to-output map to overcome these challenges. Parallel algorithms for these four problems were created, analyzed, prototyped, implemented, tuned, and scaled up for leading-edge supercomputers, including UT-Austin’s own 10 petaflops Stampede system, ANL’s Mira system, and ORNL’s Titan system. While our focus is on fundamental mathematical/computational methods and algorithms, we will assess our methods on model problems derived from several DOE mission applications, including multiscale mechanics and ice sheet dynamics.

97 MATHEMATICS AND COMPUTING↗

ON THE ACCURACY OF THE DIFFERENTIAL EMISSION MEASURE DIAGNOSTICS OF SOLAR PLASMAS. APPLICATION TO SDO/AIA. II. MULTITHERMAL PLASMAS

Differential emission measure (DEM) analysis is one of the most used diagnostic tools for solar and stellar coronae. Being an inverse problem, it has limitations due to the presence of random and systematic errors. We present in this series of papers an analysis of the robustness of the inversion in the case of SDO/AIA observations. We completely characterize the DEM inversion and its statistical properties, providing all the solutions consistent with the data along with their associated probabilities, and a test of the suitability of the assumed DEM model. While Paper I focused on isothermal conditions, we now consider multithermal plasmas and investigate both isothermal and multithermal solutions. We demonstrate how the ambiguity between noises and multithermality fundamentally limits the temperature resolution of the inversion. We show that if the observed plasma is multithermal, isothermal solutions tend to cluster on a constant temperature whatever the number of passbands or spectral lines. The multithermal solutions are also found to be biased toward near-isothermal solutions around 1 MK. This is true even if the residuals support the chosen DEM model, possibly leading to erroneous conclusions on the observed plasma. We propose tools for identifying and quantifying the possible degeneracy of solutions, thus helping the interpretation of DEM inversion.

79 ASTRONOMY AND ASTROPHYSICS↗

Flow through cascades in tandem

An exact treatment of the problem of finding the incompressible, inviscid two-dimensional flow around two cascades in tandem is presented. The analysis includes solutions of both the direct and the inverse problems. These problems are solved by conformally mapping the tandem cascade onto the region between two concentric circles in which region there are suitably placed flow singularities. Formulas for the velocity and the potential in the annular region are presented in a closed form by means of elliptic functions. The equations are presented in a form suitable for computation.

Spraglin, William E↗

Stochastic joint inversion of hydrogeophysical data for salt tracer test monitoring and hydraulic conductivity imaging

The assessment of hydraulic conductivity of heterogeneous aquifers is a difficult task using traditional hydrogeological methods (e.g., steady state or transient pumping tests) due to their low spatial resolution. Geophysical measurements performed at the ground surface and in boreholes provide additional information for increasing the resolution and accuracy of the inverted hydraulic conductivity field. We used a stochastic joint inversion of Direct Current (DC) resistivity and Self-Potential (SP) data plus in situ measurement of the salinity in a downstream well during a synthetic salt tracer experiment to reconstruct the hydraulic conductivity field between two wells. The pilot point parameterization was used to avoid over-parameterization of the inverse problem. Bounds on the model parameters were used to promote a consistent Markov chain Monte Carlo sampling of the model parameters. To evaluate the effectiveness of the joint inversion process, we compared 8 cases in which the geophysical data are coupled or not to the in situ sampling of the salinity to map the hydraulic conductivity. We first tested the effectiveness of the inversion of each type of data alone (concentration sampling, self-potential, and DC resistivity), and then we combined the data two by two. We finally combined all the data together to show the value of each type of geophysical data in the joint inversion process because of their different sensitivity map. We also investigated a case in which the data were contaminated with noise and the variogram unknown and inverted stochastically. The results of the inversion revealed that incorporating the self-potential data improves the estimate of hydraulic conductivity field especially when the self-potential data were combined to the salt concentration measurement in the second well or to the time-lapse cross-well electrical resistivity data. Various tests were also performed to quantify the uncertainty in the inverted hydraulic conductivity field.

54 ENVIRONMENTAL SCIENCES↗

Uncertainty Quantification for Large-Scale Ice Sheet Modeling

This report summarizes our work to develop advanced forward and inverse solvers and uncertainty quantification capabilities for a nonlinear 3D full Stokes continental-scale ice sheet flow model. The components include: (1) forward solver: a new state-of-the-art parallel adaptive scalable high-order-accurate mass-conservative Newton-based 3D nonlinear full Stokes ice sheet flow simulator; (2) inverse solver: a new adjoint-based inexact Newton method for solution of deterministic inverse problems governed by the above 3D nonlinear full Stokes ice flow model; and (3) uncertainty quantification: a novel Hessian-based Bayesian method for quantifying uncertainties in the inverse ice sheet flow solution and propagating them forward into predictions of quantities of interest such as ice mass flux to the ocean.

58 GEOSCIENCES↗