Search NASA⌕ Search

SEARCH · Search NASA

Results for “complex 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 1,405 records · Page 78

A Study of Complex Deep Learning Networks on High Performance, Neuromorphic, and Quantum Computers

Current Deep Learning models use highly optimized convolutional neural networks (CNN) trained on large graphical processing units (GPU)-based computers with a fairly simple layered network topology, i.e., highly connected layers, without intra-layer connections. Complex topologies have been proposed, but are intractable to train on current systems. Building the topologies of the deep learning network requires hand tuning, and implementing the network in hardware is expensive in both cost and power. In this paper, we evaluate deep learning models using three different computing architectures to address these problems: quantum computing to train complex topologies, high performance computing (HPC) to automatically determine network topology, and neuromorphic computing for a low-power hardware implementation. Due to input size limitations of current quantum computers we use the MNIST dataset for our evaluation. The results show the possibility of using the three architectures in tandem to explore complex deep learning networks that are untrainable using a von Neumann architecture. We show that a quantum computer can find high quality values of intra-layer connections and weights, while yielding a tractable time result as the complexity of the network increases; a high performance computer can find optimal layer-based topologies; and a neuromorphic computer can represent the complex topology and weights derived from the other architectures in low power memristive hardware. This represents a new capability that is not feasible with current von Neumann architecture. It potentially enables the ability to solve very complicated problems unsolvable with current computing technologies.

Potok, Thomas E↗

Discrete Fourier Transform Analysis in a Complex Vector Space

Alternative computational strategies for the Discrete Fourier Transform (DFT) have been developed using analysis of geometric manifolds. This approach provides a general framework for performing DFT calculations, and suggests a more efficient implementation of the DFT for applications using iterative transform methods, particularly phase retrieval. The DFT can thus be implemented using fewer operations when compared to the usual DFT counterpart. The software decreases the run time of the DFT in certain applications such as phase retrieval that iteratively call the DFT function. The algorithm exploits a special computational approach based on analysis of the DFT as a transformation in a complex vector space. As such, this approach has the potential to realize a DFT computation that approaches N operations versus Nlog(N) operations for the equivalent Fast Fourier Transform (FFT) calculation.

Dean, Bruce H.↗

Deactivation of Co-Schiff base catalysts in the oxidation of para-substituted lignin models for the production of benzoquinones

The effect of quinones on the deactivation of four- and five-coordinate Co-Schiff base catalysts used for the oxidation of lignin models is systematically studied. 2,6-Dimethoxy-1,4-benzoquinone does not affect the catalytic activity of any of the studied Co-Schiff base catalysts, but 1,4-benzoquinone and 2-methoxy-1,4-benzoquinone have a strong effect on the catalytic activity. Quinone solubility in the reaction solvent does not correlate with catalyst deactivation, but added pyridine (a basic axial ligand) promotes catalyst deactivation by quinone. The synthesis and characterization of a catalytically inactive Co-Schiff base-quinone complex is presented and preliminary computational analysis of this complex in comparison to a dimeric Co-Schiff base peroxo complex is also discussed. Quinone and the Co-Schiff base redox potentials are found to correlate with catalyst deactivation. Thus, catalysts with a lower redox potential were more susceptible to deactivation, and quinones with a higher redox potential deactivate the catalysts. Based on these results, two mechanisms for deactivation of the catalyst are proposed. The first mechanism describes how the formation a Co-Schiff base-quinone complex prevents formation of the key catalytically active Co-superoxo complex. The second proposed mechanism suggests that quinones inhibit the Co-Schiff base catalyst by scavenging intermediate Co-superoxo radicals.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

First-Order-hold interpolation digital-to-analog converter with application to aircraft simulation

Those who design piloted aircraft simulations must contend with the finite size and speed of the available digital computer and the requirement for simulation reality. With a fixed computational plant, the more complex the model, the more computing cycle time is required. While increasing the cycle time may not degrade the fidelity of the simulated aircraft dynamics, the larger steps in the pilot cue feedback variables (such as the visual scene cues), may be disconcerting to the pilot. The first-order-hold interpolation (FOHI) digital-to-analog converter (DAC) is presented as a device which offers smooth output, regardless of cycle time. The Laplace transforms of these three conversion types are developed and their frequency response characteristics and output smoothness are compared. The FOHI DAC exhibits a pure one-cycle delay. Whenever the FOHI DAC input comes from a second-order (or higher) system, a simple computer software technique can be used to compensate for the DAC phase lag. When so compensated, the FOHI DAC has (1) an output signal that is very smooth, (2) a flat frequency response in frequency ranges of interest, and (3) no phase error. When the input comes from a first-order system, software compensation may cause the FOHI DAC to perform as an FOHE DAC, which, although its output is not as smooth as that of the FOHI DAC, has a smoother output than that of the ZOH DAC.

Cleveland, W. B.↗

Density functional theory study of the interaction of vinyl radical, ethyne, and ethene with benzene, aimed to define an affordable computational level to investigate stability trends in large van der Waals complexes

Our purpose is to identify a computational level sufficiently dependable and affordable to assess trends in the interaction of a variety of radical or closed shell unsaturated hydro-carbons A adsorbed on soot platelet models B. These systems, of environmental interest, would unavoidably have rather large sizes, thus prompting to explore in this paper the performances of relatively low-level computational methods and compare them with higher-level reference results. To this end, the interaction of three complexes between non-polar species, vinyl radical, ethyne, or ethene (A) with benzene (B) is studied, since these species, involved themselves in growth processes of polycyclic aromatic hydrocarbons (PAHs) and soot particles, are small enough to allow high-level reference calculations of the interaction energy ΔE{sub AB}. Counterpoise-corrected interaction energies ΔE{sub AB} are used at all stages. (1) Density Functional Theory (DFT) unconstrained optimizations of the A−B complexes are carried out, using the B3LYP-D, ωB97X-D, and M06-2X functionals, with six basis sets: 6-31G(d), 6-311 (2d,p), and 6-311++G(3df,3pd); aug-cc-pVDZ and aug-cc-pVTZ; N07T. (2) Then, unconstrained optimizations by Møller-Plesset second order Perturbation Theory (MP2), with each basis set, allow subsequent single point Coupled Cluster Singles Doubles and perturbative estimate of the Triples energy computations with the same basis sets [CCSD(T)//MP2]. (3) Based on an additivity assumption of (i) the estimated MP2 energy at the complete basis set limit [E{sub MP2/CBS}] and (ii) the higher-order correlation energy effects in passing from MP2 to CCSD(T) at the aug-cc-pVTZ basis set, ΔE{sub CC-MP}, a CCSD(T)/CBS estimate is obtained and taken as a computational energy reference. At DFT, variations in ΔE{sub AB} with basis set are not large for the title molecules, and the three functionals perform rather satisfactorily even with rather small basis sets [6-31G(d) and N07T], exhibiting deviation from the computational reference of less than 1 kcal mol{sup −1}. The zero-point vibrational energy corrected estimates Δ(E{sub AB}+ZPE), obtained with the three functionals and the 6-31G(d) and N07T basis sets, are compared with experimental D{sub 0} measures, when available. In particular, this comparison is finally extended to the naphthalene and coronene dimers and to three π−π associations of different PAHs (R, made by 10, 16, or 24 C atoms) and P (80 C atoms)

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Grid generation for computational CFD of full aerodynamic configurations

It is now possible to compute flow about very complex three-dimensional aerodynamic configurations. Grid generation is an essential part of such computations and a basic capability has been demonstrated. The present problems are: (1) choosing among the competing approaches (structured, unstructured or rectangular); (2) automating and packaging software for grid generation; (3) assessing and assuring grid quality; (4) solid boundary-surface representation and grid generation on solid boundaries; (5) automated domain decomposition (structured grids); (6) grid adaption during the flow solution process. This paper discusses some of the capabilities in the computation of complex three-dimensional grids and addresses some of he current challenges.

Smith, Robert E.↗

Cray XMT Brings New Energy to High-Performance Computing

The ability to solve our nation’s most challenging problems—whether it’s cleaning up the environment, finding alternative forms of energy or improving public health and safety—requires new scientific discoveries. High performance experimental and computational technologies from the past decade are helping to accelerate these scientific discoveries, but they introduce challenges of their own. The vastly increasing volumes and complexities of experimental and computational data pose significant challenges to traditional high-performance computing (HPC) platforms as terabytes to petabytes of data must be processed and analyzed. And the growing complexity of computer models that incorporate dynamic multiscale and multiphysics phenomena place enormous demands on high-performance computer architectures. Just as these new challenges are arising, the computer architecture world is experiencing a renaissance of innovation. The continuing march of Moore’s law has provided the opportunity to put more functionality on a chip, enabling the achievement of performance in new ways. Power limitations, however, will severely limit future growth in clock rates. The challenge will be to obtain greater utilization via some form of on-chip parallelism, but the complexities of emerging applications will require significant innovation in high-performance architectures. The Cray XMT, the successor to the Tera/Cray MTA, provides an alternative platform for addressing computations that stymie current HPC systems, holding the potential to substantially accelerate data analysis and predictive analytics for many complex challenges in energy, national security and fundamental science that traditional computing cannot do.

97 MATHEMATICS AND COMPUTING↗

Computational Tools for Predictive Modeling of Properties in Complex Actinide Systems

In this chapter we focus on methodological and computational aspects that are key to accurately modeling the spectroscopic and thermodynamic properties of molecular systems containing actinides within the density functional theory (DFT) framework. Our focus is on properties that require either an accurate relativistic all-electron description or an accurate description of the dynamical behavior of actinide species in an environment at finite temperature, or both. The implementation of the methods and the calculations discussed in this chapter were done with the NWChem software suite (Valiev et al. 2010). In the first two sections we discuss two methods that account for relativistic effects, the ZORA and the X2C Hamiltonian. Section 1.2.1 discusses the implementation of the approximate relativistic ZORA Hamiltonian and its extension to magnetic properties. Section 1.3 focuses on the exact X2C Hamiltonian and the application of this methodology to obtain accurate molecular properties. In Section 1.4 we examine the role of a dynamical environment at finite temperature as well as the presence of other ions on the thermodynamics of hydrolysis and exchange reaction mechanisms. Finally, Section 1.5 discusses the modeling of XAS (EXAFS, XANES) properties in realistic environments accounting for both the dynamics of the system and (for XANES) the relativistic effects.

actinide↗

(TAML)Fe**IV=O Complex in Aqueous Solution: Synthesis And Spectroscopic And Computational Characterization

Recently, we reported the characterization of the S = 1/2 complex [Fe{sup V}(O)B*]{sup -}, where B* belongs to a family of tetraamido macrocyclic ligands (TAMLs) whose iron complexes activate peroxides for environmentally useful applications. The corresponding one-electron reduced species, [Fe{sup IV}(O)B*]{sup 2-} (2), has now been prepared in >95% yield in aqueous solution at pH > 12 by oxidation of [Fe{sup III}(H{sub 2}O)B*]{sup -} (1), with tert-butyl hydroperoxide. At room temperature, the monomeric species 2 is in a reversible, pH-dependent equilibrium with dimeric species [B*Fe{sup IV}?O?Fe{sup IV}B*]{sup 2-} (3), with a pK{sub a} near 10. In zero field, the Moessbauer spectrum of 2 exhibits a quadrupole doublet with {Delta}E{sub Q} = 3.95(3) mm/s and {delta} = ?0.19(2) mm/s, parameters consistent with a S = 1 Fe{sup IV} state. Studies in applied magnetic fields yielded the zero-field splitting parameter D = 24(3) cm{sup -1} together with the magnetic hyperfine tensor A/g{sub n}{beta}{sub n} = (?27, ?27, +2) T. Fe K-edge EXAFS analysis of 2 shows a scatterer at 1.69 (2) {angstrom}, a distance consistent with a Fe{sup IV} = O bond. DFT calculations for [Fe{sup IV}(O)B*]{sup 2-} reproduce the experimental data quite well. Further significant improvement was achieved by introducing hydrogen bonding of the axial oxygen with two solvent-water molecules. It is shown, using DFT, that the {sup 57}Fe hyperfine parameters of complex 2 give evidence for strong electron donation from B* to iron.

08 HYDROGEN↗

SGD-Net: Efficient Model-Based Deep Learning with Theoretical Guarantees

Deep unfolding networks have recently gained popularity for solving imaging inverse problems. However, the computational and memory complexity of data-consistency layers within traditional deep unfolding networks scales with the number of measurements, limiting their applicability to large-scale imaging inverse problems. We propose SGD-Net as a new methodology for improving the efficiency of deep unfolding through stochastic approximations of the data-consistency layers. Our theoretical analysis shows that SGD-Net can be trained to approximate batch deep unfolding networks to an arbitrary precision. Our simulations on intensity diffraction tomography and sparse-view computed tomography show that SGD-Net can match the performance of the traditional batch network at a fraction of training and testing complexity.Deep unfolding networks have recently gained popularity for solving imaging inverse problems. However, the computational and memory complexity of data-consistency layers within traditional deep unfolding networks scales with the number of measurements, limiting their applicability to large-scale imaging inverse problems. We propose SGD-Net as a new methodology for improving the efficiency of deep unfolding through stochastic approximations of the data-consistency layers. Our theoretical analysis shows that SGD-Net can be trained to approximate batch deep unfolding networks to an arbitrary precision. Our simulations on intensity diffraction tomography and sparse-view computed tomography show that SGD-Net can match the performance of the traditional batch network at a fraction of training and testing complexity.

97 MATHEMATICS AND COMPUTING↗

Multiscale method based on coupled lattice‐Boltzmann and Langevin‐dynamics for direct simulation of nanoscale particle/polymer suspensions in complex flows

Summary A hybrid computational method coupling the lattice‐Boltzmann (LB) method and a Langevin‐dynamics (LD) method is developed to simulate nanoscale particle and polymer (NPP) suspensions in the presence of both thermal fluctuation and long‐range many‐body hydrodynamic interactions (HIs). Brownian motion of the NPP is explicitly captured by a stochastic forcing term in the LD method. The LD method is two‐way coupled to the nonfluctuating LB fluid through a discrete LB forcing source distribution to capture the long‐range HI. To ensure intrinsically linear scalability with respect to the number of particles, a Eulerian‐host algorithm for short‐distance particle neighbor search and interaction is developed and embedded to LB‐LD framework. The validity and accuracy of the LB‐LD approach are demonstrated through several sample problems. The simulation results show good agreements with theory and experiment. The LB‐LD approach can be favorably incorporated into complex multiscale computational frameworks for efficiently simulating multiscale multicomponent particulate suspension systems such as complex blood suspensions.

Liu, Zixiang↗

Development of an explicit multiblock/multigrid flow solver for viscous flows in complex geometries

A new computer program is being developed for doing accurate simulations of compressible viscous flows in complex geometries. The code employs the full compressible Navier-Stokes equations. The eddy viscosity model of Baldwin and Lomax is used to model the effects of turbulence on the flow. A cell centered finite volume discretization is used for all terms in the governing equations. The Advection Upwind Splitting Method (AUSM) is used to compute the inviscid fluxes, while central differencing is used for the diffusive fluxes. A four-stage Runge-Kutta time integration scheme is used to march solutions to steady state, while convergence is enhanced by a multigrid scheme, local time-stepping and implicit residual smoothing. To enable simulations of flows in complex geometries, the code uses composite structured grid systems where all grid lines are continuous at block boundaries (multiblock grids). Example results are shown a flow in a linear cascade, a flow around a circular pin extending between the main walls in a high aspect-ratio channel, and a flow of air in a radial turbine coolant passage.

Steinthorsson, E.↗

Development of an explicit multiblock/multigrid flow solver for viscous flows in complex geometries

A new computer program is being developed for doing accurate simulations of compressible viscous flows in complex geometries. The code employs the full compressible Navier-Stokes equations. The eddy viscosity model of Baldwin and Lomax is used to model the effects of turbulence on the flow. A cell centered finite volume discretization is used for all terms in the governing equations. The Advection Upwind Splitting Method (AUSM) is used to compute the inviscid fluxes, while central differencing is used for the diffusive fluxes. A four-stage Runge-Kutta time integration scheme is used to march solutions to steady state, while convergence is enhanced by a multigrid scheme, local time-stepping, and implicit residual smoothing. To enable simulations of flows in complex geometries, the code uses composite structured grid systems where all grid lines are continuous at block boundaries (multiblock grids). Example results shown are a flow in a linear cascade, a flow around a circular pin extending between the main walls in a high aspect-ratio channel, and a flow of air in a radial turbine coolant passage.

Steinthorsson, E.↗

A coarse-grained computational model of the nuclear pore complex predicts Phe-Gly nucleoporin dynamics

The phenylalanine-glycine–repeat nucleoporins (FG-Nups), which occupy the lumen of the nuclear pore complex (NPC), are critical for transport between the nucleus and cytosol. Although NPCs differ in composition across species, they are largely conserved in organization and function. Transport through the pore is on the millisecond timescale. Here, to explore the dynamics of nucleoporins on this timescale, we use coarse-grained computational simulations. These simulations generate predictions that can be experimentally tested to distinguish between proposed mechanisms of transport. Our model reflects the conserved structure of the NPC, in which FG-Nup filaments extend into the lumen and anchor along the interior of the channel. The lengths of the filaments in our model are based on the known characteristics of yeast FG-Nups. The FG-repeat sites also bind to each other, and we vary this association over several orders of magnitude and run 100-ms simulations for each value. The autocorrelation functions of the orientation of the simulated FG-Nups are compared with in vivo anisotropy data. We observe that FG-Nups reptate back and forth through the NPC at timescales commensurate with experimental measurements of the speed of cargo transport through the NPC. Our results are consistent with models of transport where FG-Nup filaments are free to move across the central channel of the NPC, possibly informing how cargo might transverse the NPC.

59 BASIC BIOLOGICAL SCIENCES↗

Complex matrix multiplication operations with data pre-conditioning in a high performance computing architecture

Mechanisms for performing a complex matrix multiplication operation are provided. A vector load operation is performed to load a first vector operand of the complex matrix multiplication operation to a first target vector register. The first vector operand comprises a real and imaginary part of a first complex vector value. A complex load and splat operation is performed to load a second complex vector value of a second vector operand and replicate the second complex vector value within a second target vector register. The second complex vector value has a real and imaginary part. A cross multiply add operation is performed on elements of the first target vector register and elements of the second target vector register to generate a partial product of the complex matrix multiplication operation. The partial product is accumulated with other partial products and a resulting accumulated partial product is stored in a result vector register.

97 MATHEMATICS AND COMPUTING↗

A comparison of Gaussian processes and neural networks for computer model emulation and calibration

The Department of Energy relies on complex physics simulations for prediction in domains like cosmology, nuclear theory, and materials science. These simulations are often extremely computationally intensive, with some requiring days or weeks for a single simulation. In order to assure their accuracy, these models are calibrated against observational data in order to estimate inputs and systematic biases. Because of their great computational complexity, this process typically requires the construction of an emulator, a fast approximation to the simulation. In this paper, two emulator approaches are compared: Gaussian process regression and neural networks. Their emulation accuracy and calibration performance on three real problems of Department of Energy interest is considered. On these problems, the Gaussian process emulator tends to be more accurate with narrower, but still well-calibrated uncertainty estimates. The neural network emulator is accurate, but tends to have large uncertainty on its predictions. Finally, as a result, calibration with the Gaussian process emulator produces more constrained posteriors that still perform well in prediction.

97 MATHEMATICS AND COMPUTING↗

Computational engine structural analysis

A significant research activity at the NASA Lewis Research Center is the computational simulation of complex multidisciplinary engine structural problems. This simulation is performed using computational engine structural analysis (CESA) which consists of integrated multidisciplinary computer codes in conjunction with computer post-processing for problem-specific application. A variety of the computational simulations of specific cases are described in some detail in this paper. These case studies include: (1) aeroelastic behavior of bladed rotors, (2) high velocity impact of fan blades, (3) blade-loss transient response, (4) rotor/stator/squeeze-film/bearing interaction, (5) blade-fragment/rotor-burst containment, and (6) structural behavior of advanced swept turboprops. These representative case studies are selected to demonstrate the breath of the problems analyzed and the role of the computer including post-processing and graphical display of voluminous output data.

Chamis, C. C.↗