Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Sparsity”

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 415 records · Page 23

The potential, limitations, and challenges of divide and conquer quantum electronic structure calculations on energetic materials

High explosives are an important class of energetic materials used in many weapons applications. Even with modern computers, the simulation of the dynamic chemical reactions and energy release is exceedingly challenging. While the scale of the detonation process may be macroscopic, the dynamic bond breaking responsible for the explosive release of energy is fundamentally quantum mechanical. Thus, any method that does not adequately describe bonding is destined to lack predictive capability on some level. Performing quantum mechanics calculations on systems with more than dozens of atoms is a gargantuan task, and severe approximation schemes must be employed in practical calculations. We have developed and tested a divide and conquer (DnC) scheme to obtain total energies, forces, and harmonic frequencies within semi-empirical quantum mechanics. The method is intended as an approximate but faster solution to the full problem and is possible due to the sparsity of the density matrix in many applications. The resulting total energy calculation scales linearly as the number of subsystems, and the method provides a path-forward to quantum mechanical simulations of millions of atoms.

36 MATERIALS SCIENCE↗

Gravity Gradiometry: A Novel Application for Compressed Sensing

During the past decade, compressed sensing (CS) has proven to be extremely useful for sparse signal reconstruction. Here the method of CS is applied to fast mass distribution determination based on cold atom gravity gradiometer measurements. Specifically we consider an array of M gradiometers placed around a 2D target area in order to determine the interior mass distribution in cases where the set of (M) sensor measurements under samples the distribution. This was done by assuming that the system’s sparsity comes from the mass distributions only having K non-zero masses, which led to a non-orthogonal basis dictionary. This lack of orthogonality caused interesting behaviors, including weakened noise performance, and breaking of the typical CS motivated logarithmic scaling of required M values for larger K values. However, for low K values M scaled as expected. Modifications to the gravity sensor model to promote orthogonality and test its impact on signal recovery postponed the onset of anomalous scaling; suggesting that lack of orthogonality is the primary cause. While CS works for this sparse, but intrinsically ill-posed problem, this sensor system displayed increased noise sensitivity and a smaller upper bound on the size of recoverable K sets. However, while these limitations decrease CS performance, significant improvements over traditional sensing approaches are still possible.

74 ATOMIC AND MOLECULAR PHYSICS↗

Efficient Algorithm for Locating and Sizing Series Compensation Devices in Large Transmission Grids: Solutions and Applications (PART II)

In a companion manuscript, we developed a novel optimization method for placement, sizing, and operation of Flexible Alternating Current Transmission System (FACTS) devices to relieve transmission network congestion. Specifically, we addressed FACTS that provide Series Compensation (SC) via modification of line inductance. In this manuscript, this heuristic algorithm and its solutions are explored on a number of test cases: a 30-bus test network and a realistically-sized model of the Polish grid (~2700 nodes and ~3300 lines). The results on the 30-bus network are used to study the general properties of the solutions including non-locality and sparsity. The Polish grid is used as a demonstration of the computational efficiency of the heuristics that leverages sequential linearization of power flow constraints and cutting plane methods that take advantage of the sparse nature of the SC placement solutions. Using these approaches, the algorithm is able to solve an instance of Polish grid in tens of seconds. We explore the utility of the algorithm by analyzing transmission networks congested by (a) uniform load growth, (b) multiple overloaded configurations, and (c) sequential generator retirements

25 ENERGY STORAGE↗

Code Coupling via Jacobian-Free Newton-Krylov Algorithms with Application to Magnetized Fluid Plasma and Kinetic Neutral Models

Jacobian-free Newton-Krylov (JFNK) algorithms are a potentially powerful class of methods for solving the problem of coupling codes that address dfferent physics models. As communication capability between individual submodules varies, different choices of coupling algorithms are required. The more communication that is available, the more possible it becomes to exploit the simple sparsity pattern of the Jacobian, albeit of a large system. The less communication that is available, the more dense the Jacobian matrices become and new types of preconditioners must be sought to efficiently take large time steps. In general, methods that use constrained or reduced subsystems can offer a compromise in complexity. The specific problem of coupling a fluid plasma code to a kinetic neutrals code is discussed as an example.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Three-dimensional Imaging for Large LArTPCs

High-performance event reconstruction is critical for current and future massive liquid argon time projection chambers (LArTPCs) to realize their full scientic potential. LArTPCs with readout using wire planes provides a limited number of 2D projections. In general, without a pixel-type readout it is challenging to achieve unambiguous 3D event reconstruction. As a remedy, we present a novel 3D imaging method, Wire-Cell, which incorporates the charge and sparsity information in addition to the time and geometry through simple and robust mathematics.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

DiaMonD: An Integrated Multifaceted Approach to Mathematics at the Interfaces of Data, Models, and Decisions

The major goal of the project was to invent, analyze, implement, and apply new algorithms to speed up the computational simulation of complex systems in settings in which multiple simulations are needed. The main efforts we undertook are as follows: 1. Develop novel reduced order methods to reduce the costs of simulations, especially in settings (e.g., UQ, inference, optimization and control) for which numerous simulation are needed and also for nonlocal problems (e.g., fractional derivative, anomalous diffusion, nonlocal mechanics) for which lack of sparsity results in more expensive full-order simulation than encountered in partial differential equations settings; rigorous analyses as well as computational tests were carried out, 2. Develop multifidelity methods based on the usage of a variety of models having different fidelities and different costs to accelerate simulations again in settings where numerous simulations are needed or are especially expensive; rigorous analyses as well as computational tests were carried out, 3. Develop ensemble methods to reduced the costs of simulations again in settings where numerous simulations are needed or are especially expensive; rigorous analyses as well as computational tests were carried out, and 4. Analyze a model for Richardson pair dispersion in turbulent flows and develop discretization methods for the efficient and accurate approximations of the solution of the model; rigorous analyses as well as computational tests were carried out.

97 MATHEMATICS AND COMPUTING↗

Optimizing the Performance of Sparse-Matrix Vector Products on Next-Generation Processors

Matrix-vector products are ubiquitous in high-performance scientific applications and have a growing set of occurrences in advanced data analysis activities. Achieving high performance for these kernels is therefore paramount, in part, because these operations can consume vast amounts of application execution time. In this report we document the development of several sparse-matrix vector product kernel implementations using a variety of programming models and approaches. Each kernel is run on a broad set of matrices selected to demonstrate the wide variety of matrix structure and sparsity that is possible with a single, generic kernel. For benchmarking and performance analysis, we utilize leading computing architectures for the NNSA/ASC program including Intel's Knights Landing processor and IBM's POWER8.

97 MATHEMATICS AND COMPUTING↗

Understanding the Effects of Tapering on Gaussian Process Regression

Gaussian process regression is a computationally expensive machine learning algorithm that requires the solution of a system of linear equations with a dense covariance matrix. We propose a modification that uses tapering to introduce sparsity into the covariance matrix followed by matrix reordering schemes to reduce the bandwidth of the sparse matrix. We also apply iterative refinement to our approximate solution in order to recover some of the accuracy lost to tapering. Our numerical tests showed that this modification works best with datasets that have many small values in the covariance matrix. This restricts the usefulness of the modification, so other techniques to reduce computational time may be worth considering in the future.

97 MATHEMATICS AND COMPUTING↗

Gradient-based Optimization for Regression in the Functional Tensor-Train Format

We consider the task of low-multilinear-rank functional regression, i.e., learning a low-rank parametric representation of functions from scattered real-valued data. Our first contribution is the development and analysis of gradient-based optimization procedures, including stochastic gradient descent and quasi-Newton methods, for learning the parameters of a functional tensor-train (FT) that yields improved accuracy over standard alternating least squares methods. The functional tensor-train uses the tensor-train (TT) representation of low-rank arrays as an ansatz for a class of low-multilinear-rank functions. The FT is represented by a set of matrix-valued functions that contain a set of univariate functions, and the regression task is to learn the parameters of these univariate functions. Our second contribution demonstrates that using nonlinearly parameterized univariate functions, e.g., symmetric kernels with moving centers, within each core can outperform the standard approach of using a linear expansion of basis functions. Our final contributions are new rank adaptation and group-sparsity regularization procedures to minimize overfitting. We use several benchmark problems to demonstrate at least an order of magnitude lower accuracy with gradient-based optimization methods than ALS in the low-sample number regime. We also demonstrate an order of magnitude reduction in accuracy on a test problem resulting from nonlinear parameterizations over linear parameterizations. Finally we compare regression performance with 22 other nonparametric and parametric regression methods on 10 real-world data sets. We achieve top-five accuracy for seven of the data sets and best accuracy for two of the data sets. These rankings are the best amongst parametric models and competetive with the best non-parametric methods.

97 MATHEMATICS AND COMPUTING↗

Locating Seismic Events with Local-Distance Data

As the seismic monitoring community advances toward detecting, identifying, and locating ever-smaller natural and anthropogenic events, the need is constantly increasing for higher resolution, higher fidelity data, models, and methods for accurately characterizing events. Local-distance seismic data provide robust constraints on event locations, but also introduce complexity due to the significant geologic heterogeneity of the Earth’s crust and upper mantle, and the relative sparsity of data that often occurs with small events recorded on regional seismic networks. Identifying the critical characteristics for improving local-scale event locations and the factors that impact location accuracy and reliability is an ongoing challenge for the seismic community. Using Utah as a test case, we examine three data sets of varying duration, finesse, and magnitude to investigate the effects of local earth structure and modeling parameters on local-distance event location precision and accuracy. We observe that the most critical elements controlling relocation precision are azimuthal coverage and local-scale velocity structure, with tradeoffs based on event depth, type, location, and range.

42 ENGINEERING↗

Compressed Natural Gas Component Leak Frequency Estimation

The frequency of unintended releases in a compressed natural gas system is an important aspect of the system quantitative risk assessment. The frequencies for possible release scenarios, along with engineering models, are utilized to quantify the risks for compressed natural gas facilities. This report documents component leakage frequencies representative of compressed natural gas components that were estimated as a function of the normalized leak size. A Bayesian statistical method was used which results in leak frequency distributions for each component which represent variation and uncertainty in the leak frequency. The analysis shows that there is high uncertainty in the estimated leak frequencies due to sparsity in compressed natural gas data. These leak frequencies may still be useful in compressed natural gas system risk assessments, as long as this high uncertainty is acknowledged and considered appropriately.

03 NATURAL GAS↗

Final Technical Report for "Cloud-based Low-Scaling Quantum Chemistry Simulations for Materials"

The objective of phase I was to develop the basic infrastructure for performing efficient hybrid DFT calculations on solid materials with a turnaround time amenable to the use in large-scale data-driven approaches. To fulfill the goal we have developed a novel algorithm that significantly reduces the cost of exchange matrix formation. The algorithm does so by making use of three ingredients (a) interpolative decomposition of the electron integrals (b) robust pseudospectral method and (c) occ-RI approach. Our published work demonstrates that the algorithm is orders of magnitude faster than any other hybrid-DFT method and is on the order of only 3-4 times slower than pure DFT. All steps of the algorithm developed during phase I can be accelerated to the point that, for large enough systems, the diagonalization of the Fock matrix will become the most expensive step. To treat such large systems we have created an interface to the ASCR-funded PEXSI library. The PEXSI method is used to compute the density matrix directly from the Fock matrix in a manner that preserves the sparsity of the local representation.

Shiozaki, Toru↗

Comparison of Supervised and Un-Supervised Machine Learning Algorithms for Threat Detection and Scintillator Performance for Radiation Portal Monitoring

Following the events of September 11, 2001, international border crossing have been equipped with radiation portal monitors (RPMs) to identify illicit radioactive material. Polyvinyl toluene (PVT) scintillators are commonly used due to their low cost and reasonable maintainability, however they offer low spectral resolution. Despite the fact that over twenty years has transpired since this event, radioisotopes are still typically identified by hand-crafted classification algorithms, e.g., total counts or energy windowing, and exhibit relatively poor performance in detecting threats at the low false alarm rates required to support the stream of commerce. While some improvement to performance has been realized via the use of supervised machine learning, these classification algorithms typically utilize simulations in lieu of real data due to the sparsity of data for one or more classes. Accordingly, the performance of these algorithms is somewhat less than optimal when examining experiments or simulations with model mismatch. Consequently, in this work, we examine the application of a number of unsupervised machine learning, anomaly detection based algorithms, to circumvent the inverse crime when analyzing spectroscopy data for RPMs. We also compare anomaly detection results with those obtained via the use of supervised classification detection ML algorithms when model mismatch is introduced between the simulated threat items utilized for training/testing. Finally, we compared the performance of the PVT scintillators to those obtained with higher resolution detectors using both anomaly detection and supervised classification algorithms.

98 NUCLEAR DISARMAMENT, SAFEGUARDS, AND PHYSICAL P↗

Enabling Cybersecurity, Situational Awareness and Resilience in Distribution Grids with High Penetration of Photovoltaics (CARE-PV) (Final Report)

Since legacy distribution systems have very limited visibility beyond the substation, high penetration of PV at the grid edge presents some unique operational challenges. One approach to address these challenges is to use information from advanced metering infrastructure (AMI) and µPMUs. However, exploiting this information is impacted by a number of factors, including multi-timescale measurements, volume of data generated, communication network impairments (e.g., information loss and latency) and susceptibility to cyber-attacks. Therefore, one of the critical tasks involved in the management of a distribution grid is to develop complete situational awareness by integrating cyber-security mechanisms with state estimation strategies and leveraging this situational awareness to assure energy services at strategic locations while exploiting AMI/PV inverter/ µPMU data. This CARE-PV project addresses the fundamental challenges in situational awareness and resilience to cyber and physical vectors by exploiting the synergy between innovative modeling, estimation, data analytics, testing and validation using smart PV inverters designed at K-State and facilities at NREL. Specifically, the project involved the development, testing and validation of the following novel enabling technologies: (Thrust 1) Resilience to cyber vectors that impact data integrity was addressed via a two-level defense strategy that combines cyber intrusion detection using self-learning, cooperative smart PV inverters, and a novel moving target defense framework to combat data integrity attacks. (Thrust 2) Resilience to cyber-physical vectors that impact situational awareness by limiting data availability was addressed via novel centralized and decentralized, sparsity-based static and dynamic state estimation approaches that enhance observability even when the underlying system is unobservable. (Thrust 3) Leveraging a unique probabilistic sensitivity analysis approach accompanied by one-of-a-kind dominant influencer set computation, the vulnerability of critical infrastructure at strategic locations was evaluated so that proactive PV-based control strategies can be used to support operations under normal/outage scenarios. These CARE-PV project innovations were demonstrated on both small-scale IEEE and larger utility-scale testbeds (Thrust 4). Feedback from Industry Advisory Board members was used to formulate a commercialization pathway for a subset of CARE-PV technologies. These CARE-PV technologies will ultimately lead to reliable and secure, large-scale integration of renewable energy and mitigate the risk of energy disruption resulting from cyber incidents and other emerging threats within the energy environment.

14 SOLAR ENERGY↗

Exploration with Scalable Gaussian Process Reinforcement Learning

Exploration is a challenging problem in reinforcement learning (RL), especially in environments with sparse rewards. Quantifying and utilizing the parametric uncertainty has been shown to be paramount for successful exploration [Osband et al., 2018]. Bayesian, or approximately Bayesian, methods present a principled means of estimating the parametric uncertainty in RL problems. Gaussian processes, nonparametric Bayesian models, are often impractical due to poor scalability and computational bottlenecks. We introduce a scalable Gaussian process RL (GPRL) method which directly induces sparsity in the covariance matrix to facilitate faster computation. This is a departure from previous GPRL methods which instead rely on data reduction and subsampling. We compare various covariance-based exploration techniques (Thompson sampling, upper confidence bound, and probabilistic maximum variance) which leverage our scalable GP framework in sparse reward environments. Finally, we show favorable comparison against the bootstrapped deep Q-Network.

97 MATHEMATICS AND COMPUTING↗

Efficient Dimension Reduction of Complex Three-dimensional CO2 Saturation using Deep Learning Models

In the domain of deep learning (DL), dimension reduction is crucial for enhancing training efficiency and mitigating overfitting, particularly when managing complex data such as three-dimensional (3D) saturation data. The 3D saturation data in the context of geological carbon storage (GCS) presents unique challenges due to its inherent sparsity and the abrupt transitions at plume boundaries, known as shock fronts. To address the challenges, we proposed a novel DL framework that integrates dimension reduction with advanced 3D reconstruction techniques. Our model leveraged latent variables derived from 2D average saturation data, offering a robust and efficient solution tailored to the intricate dynamics of 3D saturation fields. The proposed framework can extract the critical features of the high-dimensional data while reducing the variable numbers, which is more tractable for DL models and enhances the model robustness and accuracy. Therefore, it provides a novel approach for modeling and analyses in complex geological scenarios, which finds great potential applications in environmental monitoring and energy storage.

Wang, Hongsheng↗

AEOLUS: Advances in Experimental Design, Optimal Control, and Learning for Uncertain Complex Systems

Sustained advances in the mathematics of modeling and simulation have resulted in the capability today for routine simulation of a number of large scale complex DOE-relevant systems. As remarkable as this capability for solving the so-called forward problem is, it is typically only the first step-an inner loop within an outer loop that explores the simulation model's parameter space and decision space to characterize uncertainty in the model's predictions, learn unknown model parameters from data, design the most informative experiments, determine optimal control strategies, and create optimal designs. Broadly, what unifies all of these outer loop problems is that they are, in one form or another, optimization problems over parameter/control/design space that are constrained by complex uncertain models. To fully realize the power of scientific simulation as a basis for scientific discovery, technological innovation, and rational decision-making, it is imperative to move beyond simulation to tackle the outer loop of optimization for learning from data, experimental design, and control with complex uncertain models. When the models under consideration are large-scale and complex, and when the optimization variable and uncertain parameter spaces are high (or infinite) dimensional, this constitutes a grand challenge of the highest order, and is intractable with conventional methods. To overcome these challenges, the AEOLUS Center was established to develop a unified mathematical, computational, and statistical framework for (1) Learning predictive models from complex data via Bayesian inference and optimization, and (2) Optimizing experiments, processes, and designs using the resulting uncertain models. These problems are intractable with conventional methods, for several reasons: (1) The simulation problems that govern the inner loops of the optimization problems are expensive to execute (due to severe nonlinearity, heterogeneity, multiphysics/multiscale coupling); (2) The optimization variable and uncertain parameter spaces are high dimensional, often stemming from discretizations of infinite dimensional fields such as initial conditions, sources, or material properties. We argue that the key to overcoming these challenges is to develop new mathematical, computational, and statistical methods that exploit the structure of the Bayesian inference and optimization problems mediated by their underlying complex uncertain models. This structure includes the regularity, sparsity, geometry, low intrinsic dimensionality, and multifidelity nature of the maps from uncertain parameter/optimization variable spaces to the specific objectives targeted: Bayesian inference, optimal experimental design, and optimal control design. Black box methods developed as generic tools are incapable of exploiting this structure. To be successful, we must create, integrate, and cross-fertilize ideas across multiple areas of applied math--including approximation theory, Bayesian inference, data science, experimental design, information theory, machine learning, model reduction, optimal control theory, parallel algorithms, PDE-constrained optimization, randomized algorithms, stochastic optimization, and uncertainty quantification--all while exploiting the structure of the problems at hand. With this goal in mind, we have marshaled a team of leading authorities in these areas. While the methods we develop will be broadly applicable across a wide spectrum of DOE problems in which experiments inform models and the systems those models describe must be optimized under uncertainty, we have chosen a specific area, advanced manufacturing and materials, to drive our work. AMM is characterized by complex models across multiple scales, and is a rich source of challenging problems in inference, experimental design, and optimal control, requiring multifaceted and integrated advances in applied mathematics. As such, AMM serves as an excellent vehicle to motivate and demonstrate the advances in applied mathematics developed by our center.

97 MATHEMATICS AND COMPUTING↗

Compressed sensing methods with applications to advanced air sampling

Environmental sampling methods developed by the Savannah River National Laboratory (SRNL) employ collectors with sorbent media tubes set at various locations to collect airborne emissions. Laboratory analyses of these tubes results in one-dimensional signals regarding what chemicals are being released and transported within the atmosphere. The analysis process is time consuming especially when analyzing a full year’s worth of tubes (hourly sample collection results in nearly 9,000 tubes per year). Using a signal processing method such as compressed sensing allows for recreation of the full signal while greatly reducing the number of analyzed samples required. Due to the sparsity of data retrieved from the air tubes, it is possible to use measurements a fraction of the size of the original data to gain much of the same information. This would improve the overall time and cost of analysis when modeling one-dimensional sampling signals.

54 ENVIRONMENTAL SCIENCES↗