Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “approximation algorithms”

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 577 records · Page 32

Fast Query-Optimized Kernel-Machine Classification

A recently developed algorithm performs kernel-machine classification via incremental approximate nearest support vectors. The algorithm implements support-vector machines (SVMs) at speeds 10 to 100 times those attainable by use of conventional SVM algorithms. The algorithm offers potential benefits for classification of images, recognition of speech, recognition of handwriting, and diverse other applications in which there are requirements to discern patterns in large sets of data. SVMs constitute a subset of kernel machines (KMs), which have become popular as models for machine learning and, more specifically, for automated classification of input data on the basis of labeled training data. While similar in many ways to k-nearest-neighbors (k-NN) models and artificial neural networks (ANNs), SVMs tend to be more accurate. Using representations that scale only linearly in the numbers of training examples, while exploring nonlinear (kernelized) feature spaces that are exponentially larger than the original input dimensionality, KMs elegantly and practically overcome the classic curse of dimensionality. However, the price that one must pay for the power of KMs is that query-time complexity scales linearly with the number of training examples, making KMs often orders of magnitude more computationally expensive than are ANNs, decision trees, and other popular machine learning alternatives. The present algorithm treats an SVM classifier as a special form of a k-NN. The algorithm is based partly on an empirical observation that one can often achieve the same classification as that of an exact KM by using only small fraction of the nearest support vectors (SVs) of a query. The exact KM output is a weighted sum over the kernel values between the query and the SVs. In this algorithm, the KM output is approximated with a k-NN classifier, the output of which is a weighted sum only over the kernel values involving k selected SVs. Before query time, there are gathered statistics about how misleading the output of the k-NN model can be, relative to the outputs of the exact KM for a representative set of examples, for each possible k from 1 to the total number of SVs. From these statistics, there are derived upper and lower thresholds for each step k. These thresholds identify output levels for which the particular variant of the k-NN model already leans so strongly positively or negatively that a reversal in sign is unlikely, given the weaker SV neighbors still remaining. At query time, the partial output of each query is incrementally updated, stopping as soon as it exceeds the predetermined statistical thresholds of the current step. For an easy query, stopping can occur as early as step k = 1. For more difficult queries, stopping might not occur until nearly all SVs are touched. A key empirical observation is that this approach can tolerate very approximate nearest-neighbor orderings. In experiments, SVs and queries were projected to a subspace comprising the top few principal- component dimensions and neighbor orderings were computed in that subspace. This approach ensured that the overhead of the nearest-neighbor computations was insignificant, relative to that of the exact KM computation.

Mazzoni, Dominic↗

An Explicit Upwind Algorithm for Solving the Parabolized Navier-Stokes Equations

An explicit, upwind algorithm was developed for the direct (noniterative) integration of the 3-D Parabolized Navier-Stokes (PNS) equations in a generalized coordinate system. The new algorithm uses upwind approximations of the numerical fluxes for the pressure and convection terms obtained by combining flux difference splittings (FDS) formed from the solution of an approximate Riemann (RP). The approximate RP is solved using an extension of the method developed by Roe for steady supersonic flow of an ideal gas. Roe's method is extended for use with the 3-D PNS equations expressed in generalized coordinates and to include Vigneron's technique of splitting the streamwise pressure gradient. The difficulty associated with applying Roe's scheme in the subsonic region is overcome. The second-order upwind differencing of the flux derivatives are obtained by adding FDS to either an original forward or backward differencing of the flux derivative. This approach is used to modify an explicit MacCormack differencing scheme into an upwind differencing scheme. The second order upwind flux approximations, applied with flux limiters, provide a method for numerically capturing shocks without the need for additional artificial damping terms which require adjustment by the user. In addition, a cubic equation is derived for determining Vegneron's pressure splitting coefficient using the updated streamwise flux vector. Decoding the streamwise flux vector with the updated value of Vigneron's pressure splitting improves the stability of the scheme. The new algorithm is applied to 2-D and 3-D supersonic and hypersonic laminar flow test cases. Results are presented for the experimental studies of Holden and of Tracy. In addition, a flow field solution is presented for a generic hypersonic aircraft at a Mach number of 24.5 and angle of attack of 1 degree. The computed results compare well to both experimental data and numerical results from other algorithms. Computational times required for the upwind PNS code are approximately equal to an explicit PNS MacCormack's code and existing implicit PNS solvers.

Korte, John J.↗

Simulated Stochastic Approximation Annealing for Global Optimization with a Square-Root Cooling Schedule

Simulated annealing has been widely used in the solution of optimization problems. As known by many researchers, the global optima cannot be guaranteed to be located by simulated annealing unless a logarithmic cooling schedule is used. However, the logarithmic cooling schedule is so slow that no one can afford to have such a long CPU time. This paper proposes a new stochastic optimization algorithm, the so-called simulated stochastic approximation annealing algorithm, which is a combination of simulated annealing and the stochastic approximation Monte Carlo algorithm. Under the framework of stochastic approximation Markov chain Monte Carlo, it is shown that the new algorithm can work with a cooling schedule in which the temperature can decrease much faster than in the logarithmic cooling schedule, e.g., a square-root cooling schedule, while guaranteeing the global optima to be reached when the temperature tends to zero. The new algorithm has been tested on a few benchmark optimization problems, including feed-forward neural network training and protein-folding. The numerical results indicate that the new algorithm can significantly outperform simulated annealing and other competitors.

Liang, Faming↗

Low-Rank Solution Methods for Stochastic Eigenvalue Problems

In this report, we study efficient solution methods for stochastic eigenvalue problems arising from discretization of self-adjoint partial differential equations with random data, where the underlying operators depend linearly on the random parameters. With the stochastic Galerkin approach, the solutions are represented as generalized polynomial chaos expansions. When these solutions can be approximated well by low-rank objects, we introduce a low-rank variant of the inverse subspace iteration algorithm for computing one or several minimal eigenvalues and corresponding eigenvectors of parameter-dependent matrices. In the algorithm, the iterates are approximated by low-rank matrices, which leads to significant cost savings. Furthermore, the algorithm is tested on two benchmark problems, a stochastic diffusion problem with some poorly separated eigenvalues, and an operator derived from a discrete stochastic Stokes problem whose minimal eigenvalue is related to the inf-sup stability constant. Numerical experiments show that the low-rank algorithm produces accurate solutions compared to the Monte Carlo method, and it uses much less computational time than the original algorithm without low-rank approximation.

97 MATHEMATICS AND COMPUTING↗

Improved Heat-Transfer Calculations for Hypersonic Flow

Lewis number corrected for extremely high airspeeds. Algorithm calculates improved, variable value of Lewis number, factor in equation for heat-transfer coefficient at stagnation point of body in hypersonic flow mach numbers of approximately 25. New algorithm improves accuracy of calculations of heat generated by hot air acting on moving body.

Greenschlag, S. N.↗

Geometry-aware training of factorized layers in tensor Tucker format

Reducing parameter redundancies in neural network architectures is crucial for achieving feasible computational and memory requirements during train and inference of large networks. Given its easy implementation and flexibility, one promising approach is layer factorization, which reshapes weight tensors into a matrix format and parameterizes it as the product of two rank-r matrices. However, this family of approaches often requires an initial full-model warm-up phase, prior knowledge of a feasible rank, and it is sensitive to parameter initialization.In this work, we introduce a novel approach to train the factors of a Tucker decomposition of the weight tensors. Our training proposal proves to be optimal in locally approximating the original unfactorized dynamics and stable for the initialization. Furthermore, the rank of each mode is dynamically updated during training.We provide a theoretical analysis of the algorithm, showing convergence, approximation and local descent guarantees. The method's performance is further illustrated through a variety of experiments, showing remarkable training compression rates and comparable or even better performance than the full baseline and alternative layer factorization strategies.

Zangrando, Emanuele [Gran Sasso Science Institute ↗

Change detection and classification of land cover in multispectral satellite imagery using clustering of sparse approximations (CoSA) over learned feature dictionaries

Neuromimetic machine vision and pattern recognition algorithms are of great interest for landscape characterization and change detection in satellite imagery in support of global climate change science and modeling. We present results from an ongoing effort to extend machine vision methods to the environmental sciences, using adaptive sparse signal processing combined with machine learning. A Hebbian learning rule is used to build multispectral, multiresolution dictionaries from regional satellite normalized band difference index data. Land cover labels are automatically generated via our CoSA algorithm: Clustering of Sparse Approximations, using a clustering distance metric that combines spectral and spatial textural characteristics to help separate geologic, vegetative, and hydrologie features. We demonstrate our method on example Worldview-2 satellite images of an Arctic region, and use CoSA labels to detect seasonal surface changes. In conclusion, our results suggest that neuroscience-based models are a promising approach to practical pattern recognition and change detection problems in remote sensing.

97 MATHEMATICS AND COMPUTING↗

Continued performance improvement and integration of MOOSE's thermal-hydraulics capabilities (M3 Milestone Report)

This work introduces performance, robustness and workflow improvements to Multiphysics Object-Oriented Simulation Environment (MOOSE)-based thermal-hydraulics solvers. It presents work related to the acceleration of segregated fluid dynamics algorithms, which show approximately a factor of 10 speedup compared to the preceding implementation. Additionally, we discuss approaches to use advanced, Schurr complement-based, field split preconditioners for monolithic solution algorithms relying on the finite volume method. The presence of the Rhie-Chow interpolation makes the utilization of this preconditioner challenging, but the results indicate that for a moderately large problem a factor of 3.4 speedup can be achieved in conjunction with a factor of 3.5 reduction in memory usage. Furthermore, we introduce several pseudo-time stepping approaches to MOOSE for the robust convergence to steady-state solutions when steady-state solves don't converge due to the initial guesses being too far from the solution in Newton's method. Every MOOSE-based application has access this algorithm and can benefit from its use. Moreover, several new avenues have been presented for importing meshes from commercial software which make meshing easier. Lastly, the Component system within the Thermal-Hydraulics Module (THM) of MOOSE is abstracted by separating geometry- and physics-related properties.

97 MATHEMATICS AND COMPUTING↗

Adaptive DIT-Based Fringe Tracking and Prediction at IOTA

An automatic fringe tracking system has been developed and implemented at the Infrared Optical Telescope Array (IOTA). In testing during May 2002, the system successfully minimized the optical path differences (OPDs) for all three baselines at IOTA. Based on sliding window discrete Fourier transform (DFT) calculations that were optimized for computational efficiency and robustness to atmospheric disturbances, the algorithm has also been tested extensively on off-line data. Implemented in ANSI C on the 266 MHZ PowerPC processor running the VxWorks real-time operating system, the algorithm runs in approximately 2.0 milliseconds per scan (including all three interferograms), using the science camera and piezo scanners to measure and correct the OPDs. Preliminary analysis on an extension of this algorithm indicates a potential for predictive tracking, although at present, real-time implementation of this extension would require significantly more computational capacity.

Wilson, Edward↗

Margins associated with loss of assured safety for systems with multiple weak links and strong links

Representations for margins associated with loss of assured safety (LOAS) for weak link (WL)/strong link (SL) systems involving multiple time-dependent failure modes are developed. Additionally, the following topics are described: (i) defining properties for WLs and SLs, (ii) background on cumulative distribution functions (CDFs) for link failure time, link property value at link failure, and time at which LOAS occurs, (iii) CDFs for failure time margins defined by (time at which SL system fails) - (time at which WL system fails), (iv) CDFs for SL system property values at LOAS, (v) CDFs for WL/SL property value margins defined by (property value at which SL system fails) - (property value at which WL system fails), and (vi) CDFs for SL property value margins defined by (property value of failing SL at time of SL system failure) - (property value of this SL at time of WL system failure). Included in this presentation is a demonstration of a verification strategy based on defining and approximating the indicated margin results with (i) procedures based on formal integral representations and associated quadrature approximations and (ii) procedures based on algorithms for sampling-based approximations.

42 ENGINEERING↗

Linear image reconstruction for a diffuse optical mammography system in a noncompressed geometry using scattering fluid

Diffuse optical tomography (DOT) is a potential new imaging modality to detect or monitor breast lesions. Recently, Philips developed a new DOT system capable of transmission and fluorescence imaging, where the investigated breast is hanging freely into the measurement cup containing scattering fluid. We present a fast and robust image reconstruction algorithm that is used for the transmission measurements. The algorithm is based on the Rytov approximation. We show that this algorithm can be used over a wide range of tissue optical properties if the reconstruction is adapted to each patient. We use estimates of the breast shape and average tissue optical properties to initialize the reconstruction, which improves the image quality significantly. We demonstrate the capability of the measurement system and reconstruction to image breast lesions by clinical examples.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Margins Associated with Loss of Assured Safety for Systems with Multiple Time-Dependent Failure Modes.

Representations for margins associated with loss of assured safety (LOAS) for weak link (WL)/strong link (SL) systems involving multiple time-dependent failure modes are developed. The following topics are described: (i) defining properties for WLs and SLs, (ii) background on cumulative distribution functions (CDFs) for link failure time, link property value at link failure, and time at which LOAS occurs, (iii) CDFs for failure time margins defined by (time at which SL system fails) – (time at which WL system fails), (iv) CDFs for SL system property values at LOAS, (v) CDFs for WL/SL property value margins defined by (property value at which SL system fails) – (property value at which WL system fails), and (vi) CDFs for SL property value margins defined by (property value of failing SL at time of SL system failure) – (property value of this SL at time of WL system failure). Included in this presentation is a demonstration of a verification strategy based on defining and approximating the indicated margin results with (i) procedures based on formal integral representations and associated quadrature approximations and (ii) procedures based on algorithms for sampling-based approximations.

97 MATHEMATICS AND COMPUTING↗

A Prototype Thick-Target Bremsstrahlung Model with Angularly-Dependent Emission in the MCNP6 ® Code

This document summarizes the current thick-target bremsstrahlung (TTB) model in MCNP and provides test results for an alternative implementation to improve the accuracy with reduced cost compared to full electron transport. It has been observed that the current TTB model produces inaccurate results in problems where the medium is thick with respect to electrons, but the photon distribution in the problem has a strong directionality. An example of such a simulation is detectors surrounding a metal target irradiated with a radiographic beam of high energy photons. The primary cause of this discrepancy is the current TTB method emits all bremsstrahlung photons in the same direction as the primary electron produced from each (γ, e ± ) interaction, leading to artificially forward peaked photon distributions for intermediate to high-energy incident photons. To improve the TTB model, we have implemented an angularly-dependent TTB model in a developer version of the MCNP6 ® code; for developers, this was done on the branch prototype/angular_ttb in the mcnp6 repo on bitbucket. The angularly-dependent TTB model accounts for the energy and scattering of electrons as they slow down in the current material, but does not sample the computationally expensive energy straggling, secondary electron events, and tracking electrons; this approach is significantly less computationally expensive than full electron transport and can be comparable to the original TTB method for problems with sufficiently complex materials and geometry. To evaluate the method, we have modeled a simple problem of a beam of 5 MeV photons incident on a sphere of plutonium surrounded by detectors at different deflection angles. For this problem, the angularly-dependent TTB produces a photon flux within 8.0% for a 90 degree deflection angle and 0.8% along the beam axis, as compared to the electron transport solution. This is an improvement compared to a 43% and 81% discrepancy with the original TTB method, respectively. The rest of this work includes the following: the first section details the current TTB treatment in MCNP, which has not been well documented elsewhere. Then, the modified TTB algorithm is detailed and the approximations compared to the condensed history algorithm are compared. Results are given comparing the two TTB methods to the condensed history transport algorithm. The appendix includes details for code developers on relevant electron transport implementation details and potential code improvements for future work.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

A Regularized Approach for Solving Magnetic Differential Equations and a Revised Iterative Equilibrium Algorithm

A method for approximately solving magnetic differential equations is described. The approach is to include a small diffusion term to the equation, which regularizes the linear operator to be inverted. The extra term allows a "source-correction" term to be defned, which is generally required in order to satisfy the solvability conditions. The approach is described in the context of computing the pressure and parallel currents in the iterative approach for computing magnetohydrodynamic equilibria. __________________________________________________

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

A regularized approach for solving magnetic differential equations and a revised iterative equilibrium algorithm

A method for approximately solving magnetic differential equations is described. The approach is to include a small diffusion term to the equation, which regularizes the linear operator to be inverted. The extra term allows a ''source-correction'' term to be defined, which is generally required in order to satisfy the solvability conditions. The approach is described in the context of computing the pressure and parallel currents in the iterative approach for computing magnetohydrodynamic equilibria.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Identifying Vehicle Signals in Continuous Seismic Data Using Unsupervised Machine-Learning Techniques

Seismic sensors deployed near roadways effectively capture ground vibrations generated by passing vehicles. Although both traditional and machine‐learning algorithms have been utilized for analyzing such signals, independent validation of detected vehicle events remains limited. We applied two unsupervised machine‐learning algorithms, uniform manifold approximation and projection for dimension reduction, and hierarchical density‐based spatial clustering of applications with noise, to continuous seismic data collected along a road on the main campus of Oak Ridge National Laboratory. The algorithms identified seven distinct cluster labels across the entire dataset. By comparing these cluster labels with precipitation records from a nearby weather station and image‐derived labels from a local camera system, we identified one cluster associated with rainfall and another with vehicle activity. Our algorithms identified a greater number of vehicle‐related labels compared to the camera‐derived labels because seismic data are unaffected by poor lighting conditions. The arrival times of the newly detected vehicle signals corresponded well with the road’s speed limit, supporting our findings. Our algorithm outperformed the short‐term average/long‐term average method and k‐means clustering. Our results suggest that seismic data, when analyzed with machine‐learning algorithms, can complement existing vehicle monitoring systems, particularly under challenging environmental conditions.

Chai, Chengping [Oak Ridge National Laboratory (OR↗

An algorithm for linearizing convex extremal problems

This paper suggests a method of approximating the solution of minimization problems for convex functions of several variables under convex constraints is suggested. The main idea of this approach is the approximation of a convex function by a piecewise linear function, which results in replacing the problem of convex programming by a linear programming problem. To carry out such an approximation, the epigraph of a convex function is approximated by the projection of a polytope of greater dimension. In the first part of the paper, the problem is considered for functions of one variable. In this case, an algorithm for approximating the epigraph of a convex function by a polygon is presented, it is shown that this algorithm is optimal with respect to the number of vertices of the polygon, and exact bounds for this number are obtained. After this, using an induction procedure, the algorithm is generalized to certain classes of functions of several variables. Applying the suggested method, polynomial algorithms for an approximate calculation of the L{sub p}-norm of a matrix and of the minimum of the entropy function on a polytope are obtained. Bibliography: 19 titles.

97 MATHEMATICS AND COMPUTING↗

Noise elimination algorithm for modal analysis

Modal analysis is an ongoing interdisciplinary physical issue. Modal parameters estimation is applied to determine the dynamic characteristics of structures under vibration excitation. Modal analysis is more challenging for the measured vibration response signals are contaminated with noise. This study develops a mathematical algorithm of structured low rank approximation combined with the complex exponential method to estimate the modal parameters. Physical experiments using a steel cantilever beam with ten accelerometers mounted, excited by an impulse load, demonstrate that this method can significantly eliminate noise from measured signals and accurately identify the modal frequencies and damping ratios. This study provides a fundamental mechanism of noise elimination using structured low rank approximation in physical fields.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗