Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “optimization, inverse problems”

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 289 records · Page 16

Stochastic inverse problems: Models and metrics

In past work, we introduced model-based inverse methods, and applied them to problems in which the anomaly could be reasonably modeled by simple canonical shapes, such as rectangular solids. In these cases the parameters to be inverted would be length, width and height, as well as the occasional probe lift-off or rotation. We are now developing a formulation that allows more flexibility in modeling complex flaws. The idea consists of expanding the flaw in a sequence of basis functions, and then solving for the expansion coefficients of this sequence, which are modeled as independent random variables, uniformly distributed over their range of values. There are a number of applications of such modeling: 1. Connected cracks and multiple half-moons, which we have noted in a POD set. Ideally we would like to distinguish connected cracks from one long shallow crack. 2. Cracks of irregular profile and shape which have appeared in cold work holes during bolt-hole eddy-current inspection. One side of such cracks is much deeper than other. 3. L or C shaped crack profiles at the surface, examples of which have been seen in bolt-hole cracks. By formulating problems in a stochastic sense, we are able to leverage the stochastic global optimization algorithms in NLSE, which is resident in VIC-3D®, to answer questions of global minimization and to compute confidence bounds using the sensitivity coefficient that we get from NLSE. We will also address the issue of surrogate functions which are used during the inversion process, and how they contribute to the quality of the estimation of the bounds.

36 MATERIALS SCIENCE↗

Progressive transfer learning for low-frequency data prediction in full waveform inversion

To effectively overcome the cycle-skipping issue in full waveform inversion (FWI), we developed a deep neural network (DNN) approach to predict the absent low-frequency components by exploiting the hidden physical relation connecting the low- and the high-frequency data. To efficiently solve this challenging nonlinear regression problem, two novel strategies were proposed to design the DNN architecture and to optimize the learning process: (1) dual data feed structure; (2) progressive transfer learning. With the dual data feed structure, not only the high-frequency data, but also the corresponding beat tone data are fed into the DNN to relieve the burden of feature extraction. The second strategy, progressive transfer learning, enables us to train the DNN using a single evolving training dataset. Within the framework of the progressive transfer learning, the training dataset continuously evolves in an iterative manner by gradually retrieving the subsurface information through the physics-based inversion module, progressively enhancing the prediction accuracy of the DNN and propelling the inversion process out of the local minima. Here, the synthetic numerical experiments suggest that, without any a priori geological information, the low-frequency data predicted by the progressive transfer learning are sufficiently accurate for an FWI engine to produce reliable subsurface velocity models free of cycle-skipping artifacts.

02 PETROLEUM↗

On the Spatial Localization of Free Electrons in 4H-SiC MOSFETS with an n Channel

The problem of the spatial localization of free electrons in 4H-SiC metal—oxide—semiconductor field effect transistors (MOSFETS) with an accumulation- and inversion-type n channel is theoretically analyzed. The analysis demonstrates that, in optimally designed accumulation transistors (ACCUFETs), the average distance from the surface, at which free electrons are localized, may be an order of magnitude larger than that in inversion MOSFETs. This can make 4H-SiC ACCUFETs advantageous as regards the effective carrier mobility in a conducting channel.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Community Project for Accelerator Science and Simulation (ComPASS)

After concluding our initial exercise (solving a simplified statistical inverse problem with unknown parameter laser intensity) of coupling Vorpal and our parallel statistical library QUESO, we shifted the application focus to DLA. Our efforts focused on developing a Gaussian process (GP) emulator within QUESO for efficient optimization of power couplers within woodpiles. The smaller simulation size (compared with LPA) allows for sufficient “training runs” to develop a reasonable GP statistical emulator for a parameter space of moderate dimension.

43 PARTICLE ACCELERATORS↗

Efficient shallow Ritz method for 1D diffusion problems

This paper studies the shallow Ritz method for solving the one-dimensional diffusion problem. It is shown that the shallow Ritz method improves the order of approximation dramatically for non-smooth problems. To realize this optimal or nearly optimal order of the shallow Ritz approximation, we develop a damped block Newton (dBN) method that alternates between updates of the linear and non-linear parameters. Per each iteration, the linear and the non-linear parameters are updated by exact inversion and one step of a modified, damped Newton method applied to a reduced non-linear system, respectively. The computational cost of each dBN iteration is $\mathcal{O}$(n). Starting with the non-linear parameters as a uniform partition of the interval, numerical experiments show that the dBN is capable of efficiently moving mesh points to nearly optimal locations. In conclusion, to improve the efficiency of the dBN further, we propose an adaptive damped block Newton (AdBN) method by combining the dBN with the adaptive neuron enhancement (ANE) method [28].

Diffusion problems↗

Iterative subspace algorithms for finite-temperature solution of Dyson equation

One-particle Green’s functions obtained from the self-consistent solution of the Dyson equation can be employed in the evaluation of spectroscopic and thermodynamic properties for both molecules and solids. However, typical acceleration techniques used in the traditional quantum chemistry self-consistent algorithms cannot be easily deployed for the Green’s function methods because of a non-convex grand potential functional and a non-idempotent density matrix. Moreover, the optimization problem can become more challenging due to the inclusion of correlation effects, changing chemical potential, and fluctuations of the number of particles. In this paper, we study acceleration techniques to target the self-consistent solution of the Dyson equation directly. We use the direct inversion in the iterative subspace (DIIS), the least-squared commutator in the iterative subspace (LCIIS), and the Krylov space accelerated inexact Newton method (KAIN). We observe that the definition of the residual has a significant impact on the convergence of the iterative procedure. Based on the Dyson equation, we generalize the concept of the commutator residual used in DIIS and LCIIS and compare it with the difference residual used in DIIS and KAIN. The commutator residuals outperform the difference residuals for all considered molecular and solid systems within both GW and GF2. For a number of bond-breaking problems, we found that an easily obtained high-temperature solution with effectively suppressed correlations is a very effective starting point for reaching convergence of the problematic low-temperature solutions through a sequential reduction of temperature during calculations.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

iTOUGH2 v7.1

iTOUGH2 (inverse TOUGH2) provides inverse modeling capabilities for TOUGH2, a simulator for multi-dimensional, multi-phase, multi-component, non-isothermal flow and transport in fractured porous media. iTOUGH2 performs sensitivity analyses, data-worth analyses, parameter estimation, and uncertainty propagation analyses in geosciences and reservoir engineering and other application areas. iTOUGH2 supports a number of different combinations of fluids and components (equation-of-state (EOS) modules). In addition, the optimization routines implemented in iTOUGH2 can also be used for sensitivity analysis, automatic model calibration, and uncertainty quantification of any external code that uses text-based input and output files using the PEST protocol. iTOUGH2 solves the inverse problem by minimizing a non-linear objective function of the weighted differences between model output and the corresponding observations. Multiple minimization algorithms (derivative-free, gradient-based, and second-order; local and global) are available. iTOUGH2 also performs Latin Hypercube Monte Carlo simulations for uncertainty propagation analyses. A detailed residual and error analysis is provided. This upgrade includes (a) global sensitivity analysis methods, (b) dynamic memory allocation (c) additional input features and output analyses, (d) increased forward simulation capabilities, (e) parallel execution on multicore PCs and Linux clusters, and (f) bug fixes.

Finsterle, Stefan↗

Global optimization of data quality checks on 2-D and 3-D networks of GPR cross-well tomographic data for automatic correction of unknown well deviations

Significant errors related to poor time zero estimation, well deviation or mislocation of the transmitter (TX) and receiver (RX) stations can render even the most sophisticated modeling and inversion routine useless. Previous examples of methods for the analysis and correction of data errors in geophysical tomography include the works of Maurer and Green (1997), Squires et al. (1992) and Peterson (2001). Here we follow the analysis and techniques of Peterson (2001) for data quality control and error correction. Through our data acquisition and quality control procedures we have very accurate control on the surface locations of wells, the travel distance of both the transmitter and receiver within the boreholes, and the change in apparent zero time. However, we often have poor control on well deviations, either because of economic constraints or the nature of the borehole itself prevented the acquisition of well deviation logs. Also, well deviation logs can sometimes have significant errors. Problems with borehole deviations can be diagnosed prior to inversion of travel-time tomography data sets by plotting the apparent velocity of a straight ray connecting a transmitter (TX) to a receiver (RX) against the take-off angle of the ray. Issues with the time-zero pick or distances between wells appear as symmetric smiles or frown in these QC plots. Well deviation or dipping-strong anisotropy will result in an asymmetric correlation between apparent velocity and take-off angle (Figure 1-B). In addition, when a network of interconnected GPR tomography data is available, one has the additional quality constraint of insuring that there is continuity in velocity between immediately adjacent tomograms. A sudden shift in the mean velocity indicates that either position deviations are present or there is a shift in the pick times. Small errors in well geometry may be effectively treated during inversion by including weighting, or relaxation, parameters into the inversion (e.g. Bautu et al., 2006). In the technique of algebraic reconstruction tomography (ART), which is used herein for the travel time inversion (Peterson et al., 1985), a small relaxation parameter will smooth imaging artifacts caused by data errors at the expense of resolution and contrast (Figure 2). However, large data errors such as unaccounted well deviations cannot be adequately suppressed through inversion weighting schemes. Previously, problems with tomograms were treated manually. However, in large data sets and/or networks of data sets, trial and error changes to well geometries become increasingly difficult and ineffective. Mislocation of the transmitter and receiver stations of GPR cross-well tomography data sets can lead to serious imaging artifacts if not accounted for prior to inversion. Previously, problems with tomograms have been treated manually prior to inversion. In large data sets and/or networks of tomographic data sets, trial and error changes to well geometries become increasingly difficult and ineffective. Our approach is to use cross-well data quality checks and a simplified model of borehole deviation with particle swarm optimization (PSO) to automatically correct for source and receiver locations prior to tomographic inversion. We present a simple model of well deviation, which is designed to minimize potential corruption of actual data trends. We also provide quantitative quality control measures based on minimizing correlations between take-off angle and apparent velocity, and a quality check on the continuity of velocity between adjacent wells. This methodology is shown to be accurate and robust for simple 2-D synthetic test cases. Plus, we demonstrate the method on actual field data where it is compared to deviation logs. This study shows the promise for automatic correction of well deviations in GPR tomographic data. Analysis of synthetic data shows that very precise estimates of well deviation can be made for small deviations, even in the presence of static data errors. However, the analysis of the synthetic data and the application of the method to a large network of field data show that the technique is sensitive to data errors varying between neighboring tomograms.

54 ENVIRONMENTAL SCIENCES↗

TU-EF-304-07: Monte Carlo-Based Inverse Treatment Plan Optimization for Intensity Modulated Proton Therapy

Purpose: Intensity-modulated proton therapy (IMPT) is increasingly used in proton therapy. For IMPT optimization, Monte Carlo (MC) is desired for spots dose calculations because of its high accuracy, especially in cases with a high level of heterogeneity. It is also preferred in biological optimization problems due to the capability of computing quantities related to biological effects. However, MC simulation is typically too slow to be used for this purpose. Although GPU-based MC engines have become available, the achieved efficiency is still not ideal. The purpose of this work is to develop a new optimization scheme to include GPU-based MC into IMPT. Methods: A conventional approach using MC in IMPT simply calls the MC dose engine repeatedly for each spot dose calculations. However, this is not the optimal approach, because of the unnecessary computations on some spots that turned out to have very small weights after solving the optimization problem. GPU-memory writing conflict occurring at a small beam size also reduces computational efficiency. To solve these problems, we developed a new framework that iteratively performs MC dose calculations and plan optimizations. At each dose calculation step, the particles were sampled from different spots altogether with Metropolis algorithm, such that the particle number is proportional to the latest optimized spot intensity. Simultaneously transporting particles from multiple spots also mitigated the memory writing conflict problem. Results: We have validated the proposed MC-based optimization schemes in one prostate case. The total computation time of our method was ∼5–6 min on one NVIDIA GPU card, including both spot dose calculation and plan optimization, whereas a conventional method naively using the same GPU-based MC engine were ∼3 times slower. Conclusion: A fast GPU-based MC dose calculation method along with a novel optimization workflow is developed. The high efficiency makes it attractive for clinical usages.

60 APPLIED LIFE SCIENCES↗

Structure-property maps and optimal inversion in configurational thermodynamics

Cluster expansions of first-principles density-functional databases in multicomponent systems are now used as a routine tool for the prediction of zero- and finite-temperature physical properties. The ability of producing large databases of various degrees of accuracy, i.e., high-throughput calculations, makes pertinent the analysis of error propagation during the inversion process. This is a very demanding task as both data and numerical noise have to be treated on equal footing. We have addressed this problem by using an analysis that combines the variational and evolutionary approaches to cluster expansions. Simulated databases were constructed ex professo to sample the configurational space in two different and complementary ways. These databases were in turn treated with different levels of both systematic and random numerical noise. The effects of the cross-validation level, size of the database, type of numerical imprecisions on the forecasting power of the expansions were extensively analyzed. We found that the size of the database is the most important parameter. Upon this analysis, we have determined criteria for selecting the optimal expansions, i.e., transferable expansions with constant forecasting power in the configurational space (a structure-property map). As a by-product, our study provides a detailed comparison between the variational cluster expansion and the genetic-algorithm approaches.

97 MATHEMATICS AND COMPUTING↗

Towards an Abstraction-Friendly Programming Model for High Productivity and High Performance Computing

General purpose languages, such as C++, permit the construction of various high-level abstractions to hide redundant, low-level details and accelerate programming productivity. Example abstractions include functions, data structures, classes, templates and so on. However, the use of abstractions significantly impedes static code analyses and optimizations, including parallelization, applied to the abstraction's complex implementations. As a result, there is a common perception that performance is inversely proportional to the level of abstraction. On the other hand, programming large scale, possibly heterogeneous high-performance computing systems is notoriously difficult, and programmers are less likely to abandon the help from high level abstractions when solving real-world, complex problems. Therefore, the need for programming models balancing both programming productivity and execution performance has reached a new level of criticality.

97 MATHEMATICS AND COMPUTING↗

Inverse Analysis with Variational Autoencoders: A Comparison of Shallow and Deep Networks

Inverse problems are applied to determine unknown properties by matching observational data with a physical model that often takes many parameters as input. To overcome the underconstrained nature of inverse problems and achieve good performance, an approach is presented involving regularization with a technique known as a variational autoencoder (VAE), which is trained to map a high-dimensional parameter space with a complex structure to a low-dimensional latent space with a simple structure. We apply this approach to unconditioned realizations of the parameters (heterogeneous hydraulic fields) for a hydrogeological inverse problem. Two types of hydraulic conductivity fields are used to evaluate the characterization for the different levels of heterogeneity complexity of the physical inputs. This approach keeps the computational cost of generating the training data low. The reason is unconditioned realizations neither rely on the observational data used to perform the inverse analysis nor require any groundwater flow model forward runs. In addition, this approach applies regularization on a low-dimensional latent space from the VAE and increases optimization efficiency through automatic differentiation. Furthermore, two different neural network (NN) structures are tested for their utility in using the VAE for inverse analysis. The performance of a deep, convolutional neural network strongly depends on the dimensionality of the latent space, which requires tuning. In contrast, a shallow, dense neural network provides consistently accurate characterization without tuning. Furthermore, our approach evaluates the advantages of the shallow, dense neural network over the deep, convolutional one and enables future application to a wide range of inverse problems.

97 MATHEMATICS AND COMPUTING↗

J-variational inequalities and zeroes of a family of maximal monotone operators by sunny generalized nonexpansive retraction

In this paper, using generalized resolvents and retractions onto shrinking closed subsets, we present new extragradient algorithms for finding the solution of a J-variational-like inequality, the zero points of a family of maximal monotone operators, and the solution set of mixed equilibrium problems for a J-α-inverse-strongly monotone-like operator in a 2-uniformly convex and 2-uniformly smooth Banach space. By introducing the new definitions, we prove strong convergence of generated iterates in the extragradient methods. Using FMINCON optimization toolbox in MATLAB, we give some numerical examples and compare them with several existence results in the literature to illustrate the usability of our results.

97 MATHEMATICS AND COMPUTING↗

Errors in Radiotherapy: Motivation for Development of New Radiotherapy Quality Assurance Paradigms

Modern radiotherapy practice has rapidly evolved during the past decade, making use of many highly complex and/or automated processes for planning and delivery, including new techniques, like intensity-modulated radiotherapy driven by inverse planning optimization methods, or near real-time image-guided adaptive therapy based on fluoroscopic or tomographic imaging on the treatment machine. In spite of the modern technology, or potentially because of it in some instances, errors and other problems continue to have a significant impact on the field. This report reviews example errors and problems, discusses some of the quality assurance issues that these types of problems raise, and motivates the development of more modern and sophisticated approaches to assure quality for our clinical radiotherapy treatment methods.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Topology optimization of 3D photonic crystals with complete bandgaps

The design of photonic crystals with complete bandgaps has recently received considerable research focus for numerous reasons. This work leverages well-known nonlinear programming techniques to alleviate the non-smoothness caused by degenerate eigenvalues such that topology optimization problems can be solved with the open-source IPOPT software. A fully-vectorial plane wave expansion technique is used with an iterative eigensolver to efficiently predict dispersion properties of candidate structures. Nonlinear programming is employed to solve the inverse problem of designing three-dimensional periodic structures that exhibit complete two-dimensional (2D) and three-dimensional (3D) photonic bandgaps. Mesh refinement is performed to alleviate the large computational burden of designing and analyzing photonic crystals, and a periodic density filter is implemented to impose a minimum feature size for manufacturability considerations.

97 MATHEMATICS AND COMPUTING↗

Rapid 3D nanoscale coherent imaging via physics-aware deep learning

Phase retrieval, the problem of recovering lost phase information from measured intensity alone, is an inverse problem that is widely faced in various imaging modalities ranging from astronomy to nanoscale imaging. The current process of phase recovery is iterative in nature. As a result, the image formation is time consuming and computationally expensive, precluding real-time imaging. Here, we use 3D nanoscale X-ray imaging as a representative example to develop a deep learning model to address this phase retrieval problem. We introduce 3D-CDI-NN, a deep convolutional neural network and differential programing framework trained to predict 3D structure and strain, solely from input 3D X-ray coherent scattering data. Our networks are designed to be “physics-aware” in multiple aspects; in that the physics of the X-ray scattering process is explicitly enforced in the training of the network, and the training data are drawn from atomistic simulations that are representative of the physics of the material. We further refine the neural network prediction through a physics-based optimization procedure to enable maximum accuracy at lowest computational cost. 3D-CDI-NN can invert a 3D coherent diffraction pattern to real-space structure and strain hundreds of times faster than traditional iterative phase retrieval methods. Our integrated machine learning and differential programing solution to the phase retrieval problem is broadly applicable across inverse problems in other application areas.

77 NANOSCIENCE AND NANOTECHNOLOGY↗