Search NASA⌕ Search

SEARCH · Search NASA

Results for “approximate computing”

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 73 records · Page 4

First-principles calculations and experimental studies of: XYZ 2 thermoelectric compounds: Detailed analysis of van der Waals interactions

First-principles calculations can accelerate the search for novel high-performance thermoelectric materials. However, the prediction of the thermoelectric properties is strongly dependent on the approximations used for the calculations. Here, thermoelectric properties were calculated with different computational approximations ( i.e. , PBE-GGA, HSE06, spin–orbit coupling and DFT-D3) for three layered XYZ 2 compounds (TmAgTe 2 , YAgTe 2 , and YCuTe 2 ). In addition to the computations, the structural, electrical and thermal properties of these compounds were measured experimentally and compared to the computations. Furthermore, an enhanced prediction of the crystal structure and heat capacity was achieved with the inclusion of van der Waals interactions due to more accurate modeling of the interatomic forces. In particular, a large shift of the acoustic phonons and low-frequency optical phonons to lower frequencies was observed from the dispersion-optimized structure. From the phonon dispersion curves of these compounds, the ultralow thermal conductivity in the investigated XYZ 2 compounds could be described by a recent developed minimum thermal conductivity model. Additionally, for the prediction of the electrical conductivity, a temperature-dependent relaxation time was used, and it was limited by acoustic phonons. While HSE06 has only a small influence on the electrical properties due to a computed band gap energy of >0.25 eV, the inclusion of both van der Waals interactions and spin–orbit coupling leads to a more accurate band structure, resulting in better prediction of electrical properties. Furthermore, the experimental thermoelectric properties of YAgTe 2 , TmAg 0.95 Zn 0.05 Te 2 and TmAg 0.95 Mg 0.05 Te 2 were measured, showing an increase in zT of TmAg 0.95 Zn 0.05 Te 2 by more than 35% ( zT = 0.47 ± 0.12) compared to TmAgTe 2 .

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Assimilation of TOPEX Sea Level Measurements with a Reduced-Gravity, Shallow Water Model of the Tropical Pacific Ocean

Sea surface height variability measured by TOPEX is analyzed in the tropical Pacific Ocean by way of assimilation into a wind-driven, reduced-gravity, shallow water model using an approximate Kalman filter and smoother. The analysis results in an optimal fit of the dynamic model to the observations, providing it dynamically consistent interpolation of sea level and estimation of the circulation. Nearly 80% of the expected signal variance is accounted for by the model within 20 deg of the equator, and estimation uncertainty is substantially reduced by the voluminous observation. Notable features resolved by the analysis include seasonal changes associated with the North Equatorial Countercurrent and equatorial Kelvin and Rossby waves. Significant discrepancies are also found between the estimate and TOPEX measurements, especially near the eastern boundary. Improvements in the estimate made by the assimilation are validated by comparisons with independent tide gauge and current meter observations. The employed filter and smoother are based on approximately computed estimation error covariance matrices, utilizing a spatial transformation and an symptotic approximation. The analysis demonstrates the practical utility of a quasi-optimal filter and smoother.

Fukumori, Ichiro↗

Transport of infrared radiation in cuboidal clouds

The transport of infrared radiation in a single cuboidal cloud is modeled using a variable azimuth two-stream approximation. Computations are made at 10 microns for a Deirmendjian (1969) C-1 water cloud where the single scattering albedo is equal to 0.638 and the asymmetry parameter is 0.865. The results indicate that the emittance of the top face of the model cloud is always less than that for a plane parallel cloud of the same optical depth. The hemispheric flux escaping from the cloud top possesses a gradient from the center to the edges which are warmer when the cloud is over warmer ground. Cooling rate calculations in the 8-13.6 micron region demonstrate that there is cooling out of the sides of the cloud at all levels even when there is heating of the core from the ground below. The radiances exiting from model cuboidal clouds are computed by path integration over the source function obtained with the two-stream approximation. Results indicate that the brightness temperature measured from finite clouds will overestimate the cloud-top temperature.

Harshvardhan, MR.↗

Initialization and Restart in Stochastic Local Search: Computing a Most Probable Explanation in Bayesian Networks

For hard computational problems, stochastic local search has proven to be a competitive approach to finding optimal or approximately optimal problem solutions. Two key research questions for stochastic local search algorithms are: Which algorithms are effective for initialization? When should the search process be restarted? In the present work we investigate these research questions in the context of approximate computation of most probable explanations (MPEs) in Bayesian networks (BNs). We introduce a novel approach, based on the Viterbi algorithm, to explanation initialization in BNs. While the Viterbi algorithm works on sequences and trees, our approach works on BNs with arbitrary topologies. We also give a novel formalization of stochastic local search, with focus on initialization and restart, using probability theory and mixture models. Experimentally, we apply our methods to the problem of MPE computation, using a stochastic local search algorithm known as Stochastic Greedy Search. By carefully optimizing both initialization and restart, we reduce the MPE search time for application BNs by several orders of magnitude compared to using uniform at random initialization without restart. On several BNs from applications, the performance of Stochastic Greedy Search is competitive with clique tree clustering, a state-of-the-art exact algorithm used for MPE computation in BNs.

Mengshoel, Ole J.↗

A Scalable O(N) Algorithm for Large-Scale Parallel First-Principles Molecular Dynamics Simulations

Traditional algorithms for first-principles molecular dynamics (FPMD) simulations only gain a modest capability increase from current petascale computers, due to their O(N 3 ) complexity and their heavy use of global communications. To address this issue, we are developing a truly scalable O(N) complexity FPMD algorithm, based on density functional theory (DFT), which avoids global communications. The computational model uses a general nonorthogonal orbital formulation for the DFT energy functional, which requires knowledge of selected elements of the inverse of the associated overlap matrix. We present a scalable algorithm for approximately computing selected entries of the inverse of the overlap matrix, based on an approximate inverse technique, by inverting local blocks corresponding to principal submatrices of the global overlap matrix. The new FPMD algorithm exploits sparsity and uses nearest neighbor communication to provide a computational scheme capable of extreme scalability. Accuracy is controlled by the mesh spacing of the finite difference discretization, the size of the localization regions in which the electronic orbitals are confined, and a cutoff beyond which the entries of the overlap matrix can be omitted when computing selected entries of its inverse. We demonstrate the algorithm's excellent parallel scaling for up to O(100K) atoms on O(100K) processors, with a wall-clock time of O(1) minute per molecular dynamics time step.

36 MATERIALS SCIENCE↗

Verifying quantitative reliability for programs that execute on unreliable hardware

Emerging high-performance architectures are expected to contain unreliable components that may exhibit soft errors, which silently corrupt the results of computations. Full detection and masking of soft errors is challenging, expensive, and, for some applications, unnecessary. For example, approximate computing applications (such as multimedia processing, machine learning, and big data analytics) can often naturally tolerate soft errors. We introduce Rely, a programming language that enables developers to reason about the quantitative reliability of an application—namely, the probability that it produces the correct result when executed on unreliable hardware. Rely allows developers to specify the reliability requirements for each value that a function produces. We perform a static quantitative reliability analysis that verifies quantitative requirements on the reliability of an application, enabling a developer to perform sound and verified reliability engineering. The analysis takes a Rely program with a reliability specification and a hardware specification that characterizes the reliability of the underlying hardware components and verifies that the program satisfies its reliability specification when executed on the underlying unreliable hardware platform. We demonstrate the application of quantitative reliability analysis on six computations implemented in Rely.

97 MATHEMATICS AND COMPUTING↗

Projection techniques for iterative solution of A(bar x) equals (bar b) with successive right-hand sides

Two projection techniques for computing approximate solutions to linear systems of the form A(bar x)(sup n) = (bar b)(sup n), for a sequence n = 1, 2, ..., e.g., such as arises from time discretization of a partial differential equation, are presented. The inexpensive approximate solutions can be used as initial guesses for iterative solution of the system, resulting in significantly reduced computational expense. Examples of two- and three-dimensional incompressible Navier-Stokes calculations are presented in which x represents the pressure, and A is a discrete Poisson operator. In flows containing significant dynamic activity, these projection techniques lead to as much as a two-fold reduction in solution time.

Fischer, Paul F.↗

The second-generation Shifted Boundary Method and its numerical analysis

Recently, the Shifted Boundary Method (SBM) was proposed within the class of unfitted (or immersed, or embedded) finite element methods. By reformulating the original boundary value problem over a surrogate (approximate) computational domain, the SBM avoids integration over cut cells and the associated problematic issues regarding numerical stability and matrix conditioning. Accuracy is maintained by modifying the original boundary conditions using Taylor expansions. Hence the name of the method, that shifts the location and values of the boundary conditions. In this article, we present enhanced variational SBM formulations for the Poisson and Stokes problems with improved flexibility and robustness. These simplified variational forms allow to relax some of the assumptions required by the mathematical proofs of stability and convergence of earlier implementations. First, we show that these new SBM implementations can be proved asymptotically stable and convergent even without the rather restrictive assumption that the inner product between the normals to the true and surrogate boundaries is positive. Second, we show that it is not necessary to introduce a stabilization term involving the tangential derivatives of the solution at Dirichlet boundaries, therefore avoiding the calibration of an additional stabilization parameter. Finally, we prove enhanced L 2 -estimates without the cumbersome assumption – of earlier proofs – that the surrogate domain is convex. Instead we rely on a conventional assumption that the boundary of the true domain is smooth, which can also be replaced by requiring convexity of the true domain. The aforementioned improvements open the way to a more general and efficient implementation of the Shifted Boundary Method, particularly in complex three-dimensional geometries. We complement these theoretical developments with numerical experiments in two and three dimensions.

42 ENGINEERING↗

Ball bearing heat analysis program (BABHAP)

The Ball Bearing Heat Analysis Program (BABHAP) is an attempt to assemble a series of equations, some of which are non-linear algebraic systems, in a logical order, which when solved, provide a complex analysis of load distribution among the balls, ball velocities, heat generation resulting from friction, applied load, and ball spinning, minimum lubricant film thickness, and many additional characteristics of ball bearing systems. Although initial design requirements for BABHAP were dictated by the core limitations of the PDP 11/45 computer, (approximately 8K of real words with limited number of instructions) the program dimensions can easily be expanded for large core computers such as the UNIVAC 1108. The PDP version of BABHAP is also operational on the UNIVAC system with the exception that the PDP uses 029 punch and the UNIVAC uses 026. A conversion program was written to allow transfer between machines.

Source record↗

Probabilistic resource allocation system with self-adaptive capability

A probabilistic resource allocation system is disclosed containing a low capacity computational module (Short Term Memory or STM) and a self-organizing associative network (Long Term Memory or LTM) where nodes represent elementary resources, terminal end nodes represent goals, and weighted links represent the order of resource association in different allocation episodes. Goals and their priorities are indicated by the user, and allocation decisions are made in the STM, while candidate associations of resources are supplied by the LTM based on the association strength (reliability). Weights are automatically assigned to the network links based on the frequency and relative success of exercising those links in the previous allocation decisions. Accumulation of allocation history in the form of an associative network in the LTM reduces computational demands on subsequent allocations. For this purpose, the network automatically partitions itself into strongly associated high reliability packets, allowing fast approximate computation and display of allocation solutions satisfying the overall reliability and other user-imposed constraints. System performance improves in time due to modification of network parameters and partitioning criteria based on the performance feedback.

Yufik, Yan M.↗

Probabilistic resource allocation system with self-adaptive capability

A probabilistic resource allocation system is disclosed containing a low capacity computational module (Short Term Memory or STM) and a self-organizing associative network (Long Term Memory or LTM) where nodes represent elementary resources, terminal end nodes represent goals, and directed links represent the order of resource association in different allocation episodes. Goals and their priorities are indicated by the user, and allocation decisions are made in the STM, while candidate associations of resources are supplied by the LTM based on the association strength (reliability). Reliability values are automatically assigned to the network links based on the frequency and relative success of exercising those links in the previous allocation decisions. Accumulation of allocation history in the form of an associative network in the LTM reduces computational demands on subsequent allocations. For this purpose, the network automatically partitions itself into strongly associated high reliability packets, allowing fast approximate computation and display of allocation solutions satisfying the overall reliability and other user-imposed constraints. System performance improves in time due to modification of network parameters and partitioning criteria based on the performance feedback.

Yufik, Yan M.↗

Randomized Sketching Algorithms for Low-Memory Dynamic Optimization

This paper develops a novel limited-memory method to solve dynamic optimization problems. The memory requirements for such problems often present a major obstacle, particularly for problems with PDE constraints such as optimal flow control, full waveform inversion, and optical tomography. In these problems, PDE constraints uniquely determine the state of a physical system for a given control; the goal is to find the value of the control that minimizes an objective. While the control is often low dimensional, the state is typically more expensive to store. This paper suggests using randomized matrix approximation to compress the state as it is generated and shows how to use the compressed state to reliably solve the original dynamic optimization problem. Concretely, the compressed state is used to compute approximate gradients and to apply the Hessian to vectors. The approximation error in these quantities is controlled by the target rank of the sketch. This approximate first- and second-order information can readily be used in any optimization algorithm. As an example, we develop a sketched trust-region method that adaptively chooses the target rank using a posteriori error information and provably converges to a stationary point of the original problem. Numerical experiments with the sketched trust-region method show promising performance on challenging problems such as the optimal control of an advection-reaction-diffusion equation and the optimal control of fluid flow past a cylinder.

97 MATHEMATICS AND COMPUTING↗

Frozen Gaussian series representation of the imaginary time propagator theory and numerical tests

Thawed Gaussian wavepackets have been used in recent years to compute approximations to the thermal density matrix. From a numerical point of view, it is cheaper to employ frozen Gaussian wavepackets. In this paper, we provide the formalism for the computation of thermal densities using frozen Gaussian wavepackets. We show that the exact density may be given in terms of a series, in which the zeroth order term is the frozen Gaussian. A numerical test of the methodology is presented for deep tunneling in the quartic double well potential. In all cases, the series is observed to converge. The convergence of the diagonal density matrix element is much faster than that of the antidiagonal one, suggesting that the methodology should be especially useful for the computation of partition functions. As a by product of this study, we find that the density matrix in configuration space can have more than two saddle points at low temperatures. This has implications for the use of the quantum instanton theory.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Model selection and parameter estimation in tumor growth models using approximate Bayesian computation-ABC

Cancer is one of the most fatal diseases in the world. Governments and researchers from various areas have continuously concentrated efforts to better understand the disease and propose diagnostic and treatment techniques. The use of mathematical models of tumor growth is of great importance for the development of such techniques. Due to the variety of models nowadays available in the literature, the problems of model selection and parameter estimation come into picture, aiming at suitably predicting the patient’s status of the disease. As the available data on dependent variables of existing models might not justify the use of common likelihood functions, approximate Bayesian computation (ABC) becomes a very attractive tool for model selection and model calibration (parameter estimation) in tumor growth models. In the present study, a Monte Carlo approximate Bayesian computation (ABC) algorithm is applied to select among competing models of tumor growth, with and without chemotherapy treatment. Simulated measurements are used in this work. The results obtained show that the algorithm correctly selects the model and estimates the parameters used to generate the simulated measurements.

60 APPLIED LIFE SCIENCES↗

HD-Bind: Encoding of Molecular Structure with Low Precision, Hyperdimensional Binary Representations

Publicly available collections of drug-like molecules have grown to comprise tens of billions of compounds due to advances in combinatorial chemistry. Traditional methods for identifying "hit" molecules from a large collection of potential drug-like candidates have relied on biophysical theory to compute approximations to the Gibbs free energy of the binding interaction between the drug and its protein target. These approaches have the major drawback that they require exceptional computing capabilities for even relatively small collections of molecules. Hyperdimensional Computing (HDC) is a recently-proposed learning paradigm that represents data with high-dimension binary vectors; this allows the use of low-precision binary vector arithmetic to create models of the data that can be learned without the need for the gradient-based optimization required in many conventional machine learning and deep learning methods. This algorithmic simplicity allows for acceleration in hardware that has been previously demonstrated in a range of application areas. We consider existing HDC approaches for molecular property classification and introduce two novel encodings of a commonly-used molecular representation, the extended connectivity fingerprint (ECFP). We show that HDC-based inference methods are as much as 91 times more efficient than traditional machine learning methods, and achieve an acceleration of nearly nine orders of magnitude compared to molecular docking. Our results show that HDC accelerated methods retain competitive accuracy on a number of well-studied tasks such as molecular property predictions using the MoleculeNet dataset, and bind/no-bind activity classification using the DUD-E and LIT-PCBA datasets. Our work thus motivates further investigation into molecular representation learning to develop ultraefficient pre-screening tools.

Jones, William↗

Verification of floating-point software

Floating point computation presents a number of problems for formal verification. Should one treat the actual details of floating point operations, or accept them as imprecisely defined, or should one ignore round-off error altogether and behave as if floating point operations are perfectly accurate. There is the further problem that a numerical algorithm usually only approximately computes some mathematical function, and we often do not know just how good the approximation is, even in the absence of round-off error. ORA has developed a theory of asymptotic correctness which allows one to verify floating point software with a minimum entanglement in these problems. This theory and its implementation in the Ariel C verification system are described. The theory is illustrated using a simple program which finds a zero of a given function by bisection. This paper is presented in viewgraph form.

Hoover, Doug N.↗

HDBind: encoding of molecular structure with hyperdimensional binary representations

Traditional methods for identifying “hit” molecules from a large collection of potential drug-like candidates rely on biophysical theory to compute approximations to the Gibbs free energy of the binding interaction between the drug and its protein target. These approaches have a significant limitation in that they require exceptional computing capabilities for even relatively small collections of molecules. Increasingly large and complex state-of-the-art deep learning approaches have gained popularity with the promise to improve the productivity of drug design, notorious for its numerous failures. However, as deep learning models increase in their size and complexity, their acceleration at the hardware level becomes more challenging. Hyperdimensional Computing (HDC) has recently gained attention in the computer hardware community due to its algorithmic simplicity relative to deep learning approaches. The HDC learning paradigm, which represents data with high-dimension binary vectors, allows the use of low-precision binary vector arithmetic to create models of the data that can be learned without the need for the gradient-based optimization required in many conventional machine learning and deep learning methods. This algorithmic simplicity allows for acceleration in hardware that has been previously demonstrated in a range of application areas (computer vision, bioinformatics, mass spectrometery, remote sensing, edge devices, etc.). To the best of our knowledge, our work is the first to consider HDC for the task of fast and efficient screening of modern drug-like compound libraries. We also propose the first HDC graph-based encoding methods for molecular data, demonstrating consistent and substantial improvement over previous work. We compare our approaches to alternative approaches on the well-studied MoleculeNet dataset and the recently proposed LIT-PCBA dataset derived from high quality PubChem assays. We demonstrate our methods on multiple target hardware platforms, including Graphics Processing Units (GPUs) and Field Programmable Gate Arrays (FPGAs), showing at least an order of magnitude improvement in energy efficiency versus even our smallest neural network baseline model with a single hidden layer. Our work thus motivates further investigation into molecular representation learning to develop ultra-efficient pre-screening tools. We make our code publicly available at https://github.com/LLNL/hdbind.

59 BASIC BIOLOGICAL SCIENCES↗