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 973 records · Page 54

Projection pursuit adaptation on polynomial chaos expansions

Here, the present work addresses the issue of accurate stochastic approximations in high-dimensional parametric space using tools from uncertainty quantification (UQ). The basis adaptation method and its accelerated algorithm in polynomial chaos expansions (PCE) were recently proposed to construct low-dimensional approximations adapted to specific quantities of interest (QoI). The present paper addresses one difficulty with these adaptations, namely their reliance on quadrature point sampling, which limits the reusability of potentially expensive samples. Projection pursuit (PP) is a statistical tool to find the “interesting” projections in high-dimensional data and thus bypass the curse-of-dimensionality. In the present work, we combine the fundamental ideas of basis adaptation and projection pursuit regression (PPR) to propose a novel method to simultaneously learn the optimal low-dimensional spaces and PCE representation from given data. While this projection pursuit adaptation (PPA) can be entirely data-driven, the constructed approximation exhibits mean-square convergence to the solution of an underlying governing equation and thus captures the supports and probability distributions associated with the physics constraints. The proposed approach is demonstrated on a borehole problem and a structural dynamics problem, demonstrating the versatility of the method and its ability to discover low-dimensional manifolds with high accuracy with limited data. In addition, the method can learn surrogate models for different quantities of interest while reusing the same data set.

97 MATHEMATICS AND COMPUTING↗

Enhancing Monte Carlo proxy applications on GPUs

In Monte Carlo neutron transport simulations, a computational routine commonly known as the ”cross-section lookup” has been identified as being the most computationally expensive part of these applications. A tool which is commonly used as a proxy application for these routines, named ”XSBench”, was created to simulate popular algorithms used in these routines on CPUs. Currently, however, as GPU-based HPC resources have become more widely available, there has been significant interest and efforts invested in moving these traditionally CPU-based simulations to GPUs. Unfortunately, the algorithms commonly used in the cross-section lookup routine were originally devised and developed for CPU-based platforms, and have seen limited study on GPUs to date. Additionally, platforms such as XSBench implement approximations which may have a negligible effect on CPUs, but may be quite impactful to performance on GPUs given the more resource-limited nature of the latter. As a result, we have created VEXS, a new tool for modeling the cross-section lookup routine which removes or at least reduces the approximations made by XSBench in order to provide a more realistic prediction of algorithm performance on GPUs. In this paper, we detail our efforts to remove and reduce these approximations, show the resulting improvement in performance prediction in comparison to a reference production code, Shift, and provide some basic profiling analysis of the resulting application.

Shriver, Forrest↗

A Framework for Error-Bounded Approximate Computing, with an Application to Dot Products

Approximate computing techniques, which trade off the computation accuracy of an algorithm for better performance and energy efficiency, have been successful in reducing computation and power costs in several domains. However, error sensitive applications in high-performance computing are unable to benefit from existing approximate computing strategies that are not developed with guaranteed error bounds. While approximate computing techniques can be developed for individual high-performance computing applications by domain specialists, this often requires additional theoretical analysis and potentially extensive software modification. Hence, the development of low-level error-bounded approximate computing strategies that can be introduced into any high-performance computing application without requiring additional analysis or significant software alterations is desirable. In this paper, we provide a contribution in this direction by proposing a general framework for designing error-bounded approximate computing strategies and apply it to the dot product kernel to develop \bf qdot---an error-bounded approximate dot product kernel. Following the introduction of qdot, here we perform a theoretical analysis that yields a deterministic bound on the relative approximation error introduced by qdot. Empirical tests are performed to illustrate the tightness of the derived error bound and to demonstrate the effectiveness of qdot on a synthetic dataset, as well as two scientific benchmarks---the conjugate gradient (CG) and power methods. In some instances, using qdot for the dot products in CG can result in many components being quantized to half precision without increasing the iteration count required for convergence to the same solution as CG using a double precision dot product.

97 MATHEMATICS AND COMPUTING↗

Assessing barriers and research challenges for automated fault detection and diagnosis technology for small commercial buildings in the United States

Commercial buildings often experience faults that waste energy, decrease occupant comfort, and increase operating costs. For medium and larger commercial buildings (buildings with more than approximately 1000 m 2 [approximately 10,000 ft 2 ] of floor area), studies have shown that automated fault detection and diagnosis (AFDD) tools can help building owners and operators identify and correct faults, improving building performance and producing up to 10% energy savings. However, the existing state of the art in AFDD tools and algorithms poorly serves the needs of commercial buildings less than approximately 1000 m2 (approximately 10,000 ft 2 ). Using the United States market and building stock as a case study, this article characterizes AFDD needs for small commercial buildings, surveys the types of AFDD tools presently available in the market, identifies gaps and barriers to widespread adoption of AFDD technology in small commercial buildings, and makes recommendations for the future research and development of small buildings AFDD technology.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Effective matrix model for nuclear physics on a quantum computer

In this project we studied a simplified potential from [1] describing nuclear physics from quark and gluon fields, where spatial degrees of freedom have been integrated over. This potential deals with the phase structure of the strong interaction. We created a matrix representation of the Hamiltonian for this theory using discrete quantum mechanics [2]. We then feed this Hamiltonian to IBM's Qiskit package [3]. We run on IBM's 32-qubit quantum simulator, which approximates the ground state energy using the variational quantum eigensolver (VQE) algorithm [4]. This result is then compared with the exact ground state energy, which we can calculate classically since the matrices we use are still small enough. The bosonic Hamiltonian for a 6-qubit representation took over 1000 Pauli factors to represent fully. To improve our performance, we can look for a more efficient representation of this model's Hamiltonian. We will develop a better variational trial wave function and initial state, as well as include the effect of fermions and heavy quarks on the potential.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Three-dimensional imaging using coherent x rays at grazing incidence geometry

We have developed a three-dimensional coherent diffraction imaging algorithm to retrieve phases of diffraction patterns of samples in grazing incidence small angle x-ray scattering experiments. The algorithm interprets the diffraction patterns using the distorted-wave Born approximation instead of the Born approximation, as in this case, the existence of a reflected beam from the substrate causes the diffraction pattern to deviate significantly from the simple Fourier transform of the object. Detailed computer simulations show that the algorithm works. Verification with real experiments is planned.

Yang, Yi↗

Global Identification of Previously Undetected Pre-Satellite-Era Tropical Cyclone Candidates in NOAA/CIRES Twentieth-Century Reanalysis Data

Prior to the satellite era, limited synoptic observation networks led to an indefinite number of tropical cyclones (TCs) remaining undetected. This period of decreased confidence in the TC climatological record includes the first two-thirds of the twentieth century. While prior studies found that this undersampling exists, disagreement regarding its magnitude has caused difficulties in interpreting multidecadal changes in TC activity. Previous research also demonstrated that reanalyses can be used to extend TC climatology, utilizing the NOAA/Cooperative Institute for Research in Environmental Sciences (CIRES) Twentieth-Century Reanalysis to manually identify previously unknown Atlantic Ocean basin potential TCs. This study expands the spatiotemporal scope of the earlier work by presenting a filtering algorithm that dramatically improves the efficiency with which candidate events are identified in the reanalysis. This algorithm was applied to all tropical basins for the years 1871–1979, resulting in the first quantitative and objective global TC candidate event counts for the decades prior to formal recordkeeping. Observational verification performed on a subset of these events indicates that the algorithm identifies potential missing TCs at a success rate approximating that of earlier work with a significant decrease in the amount of time required. Extrapolating these proportions to all of the candidate events identified suggests that this method may help to locate hundreds of previously unknown TCs worldwide for future study and cataloging. As such, the dataset produced by this research is a source of independent guidance for use in ongoing and future TC climatology revision efforts to produce a more complete historical record more quickly than with current methods.

54 ENVIRONMENTAL SCIENCES↗

Constructing diabatic representations using adiabatic and approximate diabatic data – Coping with diabolical singularities

We have recently introduced a diabatization scheme, which simultaneously fits and diabatizes adiabatic ab initio electronic wave functions, Zhu and Yarkony J. Chem. Phys. 140, 024112 (2014). The algorithm uses derivative couplings in the defining equations for the diabatic Hamiltonian, H{sup d}, and fits all its matrix elements simultaneously to adiabatic state data. This procedure ultimately provides an accurate, quantifiably diabatic, representation of the adiabatic electronic structure data. However, optimizing the large number of nonlinear parameters in the basis functions and adjusting the number and kind of basis functions from which the fit is built, which provide the essential flexibility, has proved challenging. In this work, we introduce a procedure that combines adiabatic state and diabatic state data to efficiently optimize the nonlinear parameters and basis function expansion. Further, we consider using direct properties based diabatizations to initialize the fitting procedure. To address this issue, we introduce a systematic method for eliminating the debilitating (diabolical) singularities in the defining equations of properties based diabatizations. We exploit the observation that if approximate diabatic data are available, the commonly used approach of fitting each matrix element of H{sup d} individually provides a starting point (seed) from which convergence of the full H{sup d} construction algorithm is rapid. The optimization of nonlinear parameters and basis functions and the elimination of debilitating singularities are, respectively, illustrated using the 1,2,3,4{sup 1}A states of phenol and the 1,2{sup 1}A states of NH{sub 3}, states which are coupled by conical intersections.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Mapping ephemeral stream networks in desert environments using very-high-spatial-resolution multispectral remote sensing

In this paper, understanding the spatial patterns of ephemeral streams is crucial for understanding how hydrologic processes influence the abundance and distribution of wildlife habitats in desert regions. Available methods for mapping ephemeral streams at the watershed scale typically underestimate the size of channel networks. Although remote sensing is an effective means of collecting data and obtaining information on large, inaccessible areas, conventional techniques for extracting channel features are not sufficient in regions that have small topographic gradients and subtle target-background spectral contrast. By using very high resolution multispectral imagery, we developed a new algorithm that applies landscape information to map ephemeral channels in desert regions of the Southwestern United States where utility-scale solar energy development is occurring. Knowledge about landscape features and structures was integrated into the algorithm using a series of spectral transformation and spatial statistical operations to integrate information about landscape features and structures. The algorithm extracted ephemeral stream channels at a local scale, with the result that approximately 900% more ephemeral streams was identified than what were identified by using the U.S. Geological Survey’s National Hydrography Dataset. The accuracy of the algorithm in detecting channel areas was as high as 92%, and its accuracy in delineating channel center lines was 91% when compared to a subset of channel networks that were digitized by using the very high resolution imagery. Although the algorithm captured stream channels in desert landscapes across various channel sizes and forms, it often underestimated stream headwaters and channels obscured by bright soils and sparse vegetation. While further improvement is warranted, the algorithm provides an effective means of obtaining detailed information about ephemeral streams, and it could make a significant contribution toward improving the hydrological modelling of desert environments.

47 OTHER INSTRUMENTATION↗

MULTIRESOLUTION REPRESENTATION OF OPERATORS WITH BOUNDARY CONDITIONS ON SIMPLE DOMAINS

We develop a multiresolution representation of a class of integral operators satisfying boundary conditions on simple domains in order to construct fast algorithms for their application. We also elucidate some delicate theoretical issues related to the construction of periodic Green s functions for Poisson s equation. By applying the method of images to the non-standard form of the free space operator, we obtain lattice sums that converge absolutely on all scales, except possibly on the coarsest scale. On the coarsest scale the lattice sums may be only conditionally convergent and, thus, allow for some freedom in their definition. We use the limit of square partial sums as a definition of the limit and obtain a systematic, simple approach to the construction (in any dimension) of periodized operators with sparse non-standard forms. We illustrate the results on several examples in dimensions one and three: the Hilbert transform, the projector on divergence free functions, the non-oscillatory Helmholtz Green s function and the Poisson operator. Remarkably, the limit of square partial sums yields a periodic Poisson Green s function which is not a convolution. Using a short sum of decaying Gaussians to approximate periodic Green s functions, we arrive at fast algorithms for their application. We further show that the results obtained for operators with periodic boundary conditions extend to operators with Dirichlet, Neumann, or mixed boundary conditions.

99 GENERAL AND MISCELLANEOUS↗

Linear-scaling implementation of the direct random-phase approximation

We report the linear-scaling implementation of the direct random-phase approximation (dRPA) for closed-shell molecular systems. As a bonus, linear-scaling algorithms are also presented for the second-order screened exchange extension of dRPA as well as for the second-order Møller–Plesset (MP2) method and its spin-scaled variants. Our approach is based on an incremental scheme which is an extension of our previous local correlation method [Rolik et al., J. Chem. Phys. 139, 094105 (2013)]. The approach extensively uses local natural orbitals to reduce the size of the molecular orbital basis of local correlation domains. In addition, we also demonstrate that using natural auxiliary functions [M. Kállay, J. Chem. Phys. 141, 244113 (2014)], the size of the auxiliary basis of the domains and thus that of the three-center Coulomb integral lists can be reduced by an order of magnitude, which results in significant savings in computation time. The new approach is validated by extensive test calculations for energies and energy differences. Our benchmark calculations also demonstrate that the new method enables dRPA calculations for molecules with more than 1000 atoms and 10 000 basis functions on a single processor.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Experimental investigation of a moving averaging algorithm for motion perpendicular to the leaf travel direction in dynamic MLC target tracking

Purpose: In dynamic multileaf collimator (MLC) motion tracking with complex intensity-modulated radiation therapy (IMRT) fields, target motion perpendicular to the MLC leaf travel direction can cause beam holds, which increase beam delivery time by up to a factor of 4. As a means to balance delivery efficiency and accuracy, a moving average algorithm was incorporated into a dynamic MLC motion tracking system (i.e., moving average tracking) to account for target motion perpendicular to the MLC leaf travel direction. The experimental investigation of the moving average algorithm compared with real-time tracking and no compensation beam delivery is described. Methods: The properties of the moving average algorithm were measured and compared with those of real-time tracking (dynamic MLC motion tracking accounting for both target motion parallel and perpendicular to the leaf travel direction) and no compensation beam delivery. The algorithm was investigated using a synthetic motion trace with a baseline drift and four patient-measured 3D tumor motion traces representing regular and irregular motions with varying baseline drifts. Each motion trace was reproduced by a moving platform. The delivery efficiency, geometric accuracy, and dosimetric accuracy were evaluated for conformal, step-and-shoot IMRT, and dynamic sliding window IMRT treatment plans using the synthetic and patient motion traces. The dosimetric accuracy was quantified via a {gamma}-test with a 3%/3 mm criterion. Results: The delivery efficiency ranged from 89 to 100% for moving average tracking, 26%-100% for real-time tracking, and 100% (by definition) for no compensation. The root-mean-square geometric error ranged from 3.2 to 4.0 mm for moving average tracking, 0.7-1.1 mm for real-time tracking, and 3.7-7.2 mm for no compensation. The percentage of dosimetric points failing the {gamma}-test ranged from 4 to 30% for moving average tracking, 0%-23% for real-time tracking, and 10%-47% for no compensation. Conclusions: The delivery efficiency of moving average tracking was up to four times higher than that of real-time tracking and approached the efficiency of no compensation for all cases. The geometric accuracy and dosimetric accuracy of the moving average algorithm was between real-time tracking and no compensation, approximately half the percentage of dosimetric points failing the {gamma}-test compared with no compensation.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Expressed sequence tags from Peromyscus testis and placenta tissue: Analysis, annotation, and utility for mapping

Background: Mice of the genus Peromyscus are found in nearly every habitat from Alaska to Central America and from the Atlantic to the Pacific. They provide an evolutionary outgroup to the Mus/Rattus lineage and serve as an intermediary between that lineage and humans. Although Peromyscus has been studied extensively under both field and laboratory conditions, research has been limited by the lack of molecular resources. Genes associated with reproduction typically evolve rapidly and thus are excellent sources of evolutionary information. In this study we describe the generation of two cDNA libraries, one from placenta and one from testis, characterize the resulting ESTs, and describe their utility for mapping the Peromyscus genome. Results: The 5' ends of 1,510 placenta and 4,798 testis clones were sequenced. Low quality sequences were removed and after clustering and contig assembly, 904 unique placenta and 2,002 unique testis sequences remained. Average lengths of placenta and testis ESTs were 711 bp and 826 bp, respectively. Approximately 82% of all ESTs were identified using the BLASTX algorithm to Mus and Rattus, and 34 – 54% of all ESTs could be assigned to a biological process gene ontology category in either Mus or Rattus. Because the Peromyscus genome organization resembles the Rattus genome more closely than Mus we examined the distribution of the Peromyscus ESTs across the rat genome finding markers on all rat chromosomes except the Y. Approximately 40% of all ESTs were specific to only one location in the Mus genome and spanned introns of an appropriate size for sequencing and SNP detection. Of the primers that were tried 54% provided useful assays for genotyping on interspecific backcross and whole-genome radiation hybrid cell panels. Conclusion: The 2,906 Peromyscus placenta and testis ESTs described here significantly expands the molecular resources available for the genus. These ESTs allow for specific PCR amplification and broad coverage across the genome, creating an excellent genetic marker resource for the generation of a medium-density genomic map. Thus, this resource will significantly aid research of a genus that is uniquely well-suited to both laboratory and field research.

59 BASIC BIOLOGICAL SCIENCES↗

Optical galaxy cluster detection across a wide redshift range

The past decade is one of the most exciting period in the history of physics and astronomy. The discovery of cosmic acceleration dramatically changed our understanding about the evolution and constituents of the Universe. To accommodate the new acceleration phase into our well established Big Bang cosmological scenario under the frame work of General Relativity, there must exist a very special substance that has negative pressure and make up about 73% of the total energy density in our Universe. It is called Dark Energy. For the first time people realized that the vast majority of our Universe is made of things that are totally different from the things we are made of. Therefore, one of the major endeavors in physics and astronomy in the coming years is trying to understand, if we can, the nature of dark energy. Understanding dark energy cannot be achieved from pure logic. We need empirical evidence to finally determine about what is dark energy. The better we can constrain the energy density and evolution of the dark energy, the closer we will get to the answer. There are many ways to constrain the energy density and evolution of dark energy, each of which leads to degeneracy in certain directions in the parameter space. Therefore, a combination of complimentary methods will help to reduce the degeneracies and give tighter constraints. Dark energy became dominate over matter in the Universe only very recently (at about z ~ 1.5) and will affect both the cosmological geometry and large scale structure formation. Among the various experiments, some of them constrain the dark energy mainly via geometry (such as CMB, Supernovae) while some others provides constraints from both structures and geometry (such as BAO, Galaxy Clusters) Galaxy clusters can be used as a sensitive probe for cosmology. A large cluster catalog that extends to high redshift with well measured masses is indispensable for precisely constraining cosmological parameters. Detecting clusters in optical bands is very efficient. Multi-color CCD photometry allows combined detection and redshift estimation for clusters across broad redshift ranges. However, the lack of precise information about galaxy positions along the line of sight leads to contamination by projection, which plagues both cluster detection and the measurement of their properties. The dominance of red sequence galaxies, tightly clustered along the E/S0 ridgeline, provides a powerful method for de-projecting field galaxies. We developed an Error Corrected Gaussian Mixture Model to fit the galaxies color distribution around clusters by taking into account the measurement errors. By this technique, we can detect the red sequence color clustering and extract unbiased information about the evolution of the red sequence ridgeline and its width. Precision measurements of ridgelines yields better estimates of cluster richness and possibly their dynamical state, leading to better estimates of cluster mass. By using the red sequence clustering in color space identified from the Error Corrected Gaussian Mixture Model, together with the projected NFW filter in the projected RA/DEC plane, we developed a new and efficient cluster finding algorithm that can reliably detect galaxy clusters across the redshift range from 0.1 to 1.4. We have also run the cluster finder on legacy SDSS DR7 data and assembled an approximately volume limited cluster catalog across redshifts from 0.1 to 0.5. The algorithm has been tested against a Monte Carlo mock catalog, showing the identified clusters are highly complete and pure. With the completion of this thesis, we build the first and essential step towards precision cluster cosmology. Meanwhile, the large optical cluster catalog across a wide redshift range makes possible the systematic and detailed investigation of cluster formation and evolution.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

An Explicitly Correlated Wavelet Method for the Electronic Schroedinger Equation

A discretization for an explicitly correlated formulation of the electronic Schroedinger equation based on hyperbolic wavelets and exponential sum approximations of potentials is described, covering mathematical results as well as algorithmic realization, and discussing in particular the potential of methods of this type for parallel computing.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

A high-order time-parallel scheme for solving wave propagation problems via the direct construction of an approximate time-evolution operator

Our manuscript demonstrates a technique for efficiently solving the classical wave equation, the shallow water equations, and, more generally, equations of the form ∂u/∂t=Lu∂u/∂t=Lu, where LL is a skew-Hermitian differential operator. The idea is to explicitly construct an approximation to the time-evolution operator exp(τL)exp(τL) for a relatively large time-step ττ. Recently developed techniques for approximating oscillatory scalar functions by rational functions, and accelerated algorithms for computing functions of discretized differential operators are exploited. Principal advantages of the proposed method include: stability even for large time-steps, the possibility to parallelize in time over many characteristic wavelengths and large speed-ups over existing methods in situations where simulation over long times are required. Numerical examples involving the 2D rotating shallow water equations and the 2D wave equation in an inhomogenous medium are presented, and the method is compared to the 4th order Runge–Kutta (RK4) method and to the use of Chebyshev polynomials. The new method achieved high accuracy over long-time intervals, and with speeds that are orders of magnitude faster than both RK4 and the use of Chebyshev polynomials.

97 MATHEMATICS AND COMPUTING↗

Beyond Limber: efficient computation of angular power spectra for galaxy clustering and weak lensing

Angular two-point statistics of large-scale structure observables are important cosmological probes. To reach the high accuracy required by the statistical precision of future surveys, some of these statistics may need to be computed without the commonly employed Limber approximation; the exact computation however requires integration over Bessel functions, and a brute-force evaluation is slow to converge. Here, we present a new method based on our generalized FFTLog algorithm for the efficient computation of angular power spectra beyond the Limber approximation. The new method significantly simplifies the calculation and improves the numerical speed and stability. It is easily extended to handle integrals involving derivatives of Bessel functions, making it equally applicable to numerically more challenging cases such as contributions from redshift-space distortions and Doppler effects. We implement our method for galaxy clustering and galaxy-galaxy lensing power spectra. We find that using the Limber approximation for galaxy clustering in future analyses like LSST Year 1 and DES Year 6 may cause significant biases in cosmological parameters, indicating that going beyond the Limber approximation is necessary for these analyses.

weak gravitational lensing↗

Efficient Neural Network Approaches for Conditional Optimal Transport with Applications in Bayesian Inference

In this work, we present two neural network approaches that approximate the solutions of static and dynamic conditional optimal transport (COT) problems. Both approaches enable conditional sampling and conditional density estimation, which are core tasks in Bayesian inference—particularly in the simulation-based (“likelihood-free”) setting. Our methods represent the target conditional distribution as a transformation of a tractable reference distribution. Obtaining such a transformation, chosen here to be an approximation of the COT map, is computationally challenging even in moderate dimensions. To improve scalability, our numerical algorithms use neural networks to parameterize candidate maps and further exploit the structure of the COT problem. Our static approach approximates the map as the gradient of a partially input convex neural network. It uses a novel numerical implementation to increase computational efficiency compared to state-of-the-art alternatives. Our dynamic approach approximates the conditional optimal transport via the flow map of a regularized neural ODE; compared to the static approach, it is slower to train but offers more modeling choices and can lead to faster sampling. We demonstrate both algorithms numerically, comparing them with competing state-of-the-art approaches, using benchmark datasets and simulation-based Bayesian inverse problems.

97 MATHEMATICS AND COMPUTING↗