Search NASA⌕ Search

SEARCH · Search NASA

Results for “Parallel optimization”

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 1,495 records · Page 83

Intrinsic plasma rotation and Reynolds stress at the plasma edge in the HSX stellarator

Using multi-tipped Langmuir probes in the edge of the HSX stellarator, the radial electric field and parallel flows are found to deviate from the values calculated by the neoclassical transport code PENTA for the optimized quasi-helically symmetric (QHS) configuration. To understand whether Reynolds stress might explain the discrepancy, fluctuating floating potential measurements are made at two locations in the torus corresponding to the low field and high field sides of the device. The measurements at the two locations show clear evidence of a gradient in the Reynolds stress. However, the resulting flow due to the gradient in the stress is found to be large and in opposite directions for the two locations. This makes an estimation of the flux surface average using a small number of measurement locations impractical from an experimental perspective. These results neither confirm nor rule out whether Reynolds stress plays an important role for the QHS configuration. Measurements made in configurations with the quasi-symmetry degraded show even larger flows and greater deviations from the neoclassically calculated velocity profiles than the QHS configuration while the fluctuation magnitudes are reduced. Lastly, for these configurations in particular, the Reynolds stress is most likely not responsible for the additional momentum.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Supercharging simulation-based inference for Bayesian optimal experimental design

Abstract Bayesian optimal experimental design (BOED) seeks to maximize the expected information gain (EIG) of experiments. This requires a likelihood estimate, which in many settings is intractable. Simulation-based inference (SBI) provides powerful tools for this regime. However, existing work explicitly connecting SBI and BOED is restricted to a single contrastive EIG bound. We show that the EIG admits multiple formulations which can directly leverage modern SBI density estimators, encompassing neural posterior, likelihood, and ratio estimation. Building on this perspective, we define a novel EIG estimator using neural likelihood estimation. Further, we identify optimization as a key bottleneck of gradient based EIG maximization and show that a simple multi-start parallel gradient ascent procedure can substantially improve reliability and performance. With these innovations, our SBI-based BOED methods are able to match or outperform by up to 22% existing state-of-the-art approaches across standard BOED benchmarks.

97 MATHEMATICS AND COMPUTING↗

Algorithms for Efficient Reproducible Floating Point Summation

We define “reproducibility” as getting bitwise identical results from multiple runs of the same program, perhaps with different hardware resources or other changes that should not affect the answer. Many users depend on reproducibility for debugging or correctness. However, dynamic scheduling of parallel computing resources, combined with nonassociative floating point addition, makes reproducibility challenging even for summation, or operations like the BLAS. We describe a “reproducible accumulator” data structure (the “binned number”) and associated algorithms to reproducibly sum binary floating point numbers, independent of summation order. We use a subset of the IEEE Floating Point Standard 754-2008 and bitwise operations on the standard representations in memory. Our approach requires only one read-only pass over the data, and one reduction in parallel, using a 6-word reproducible accumulator (more words can be used for higher accuracy), enabling standard tiling optimization techniques. Summing n words with a 6-word reproducible accumulator requires approximately 9 n floating point operations (arithmetic, comparison, and absolute value) and approximately 3 n bitwise operations. The final error bound with a 6-word reproducible accumulator and our default settings can be up to 2 29 times smaller than the error bound for conventional (recursive) summation on ill-conditioned double-precision inputs.

Computer Science↗

Development of Advanced Membranes Technology Platform for Hydrocarbon Separations

Virtually all natural gas is dehydrated during its production, transmission and storage, mostly by absorption processes. Membranes offer many potential advantages over absorption, including smaller footprints, lighter-weight packages, packaging flexibility, minimal electrical power duty, amenability to expansion due to system modularity, reduced maintenance costs, reduced emissions of heavy hydrocarbons, no liquid waste streams, and amenability to unmanned operation. The latter is particularly valuable because new natural gas sources are generally located in remote onshore and offshore sites. Most commercially-available membranes for natural gas upgrading involve high capital costs, high methane loss and performance degradation from operational upsets – all of which are barriers to their widespread adoption by the industry. The original focus of the project was to develop and demonstrate robust, high-performance membranes for natural gas dehydration. The first task completed was a user needs-and-wants study to 1) clarify the expectations of system fabricators and end users of the new separations equipment, and 2) establish the required technical and commercial targets for the membrane products. Following this, membrane system modeling and membrane development in the lab proceeded in parallel. Membrane module diameter and length, as well as and the fiber outer and inner fiber diameter, were optimized from a mathematical model that accounts for the relevant fluid dynamics and permeation phenomena. Module design was evaluated in the context of overall system design, capital costs and energy consumption, including the process scheme (particularly sweep generation), feed pretreatment, system layout, and process control. This study provided targets for membrane permeation coefficients and membrane geometry in a commercial offering that would be competitive with absorption systems. A commercially-available polymer with good tensile strength and chemical resistance was selected for membrane development. A novel dope composition and spinning process were developed, which provide a new approach to controlling membrane porosity and wall and skin morphology. A hollow-fiber membrane with an external dense “skin” was produced that has a high water vapor permeation coefficient and selectivity, durability when in operation at 1000 psig and 70°C, and the ability to withstand aromatic and aliphatic hydrocarbon vapors for an extended period. The fiber meets the technical requirements for a commercial product offering in gas dehydration. It can be readily manufactured with some changes in process equipment and process conditions, and is an excellent candidate for scale-up to full-size membrane modules.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Parallel File System I/O Performance Testing On LANL Clusters

These are slides from a presentation on parallel file system I/O performance testing on LANL clusters. I/O is a known bottleneck for HPC applications. Performance optimization of I/O is often required. This summer project entailed integrating IOR under Pavilion and automating the results analysis. The slides cover the following topics: scope of the work, tools utilized, IOR-Pavilion test workflow, build script, IOR parameters, how parameters are passed to IOR, *run_ior: functionality, Python IOR-Output Parser, Splunk data format, Splunk dashboard and features, and future work.

97 MATHEMATICS AND COMPUTING↗

ARPA E GO Competition

A Nonlinear Programming SC-ACOPF Framework with Parallel Computing Capabilities

24 POWER TRANSMISSION AND DISTRIBUTION↗

Issues in the digital implementation of control compensators

Techniques developed for the finite-precision implementation of digital filters were used, adapted, and extended for digital feedback compensators, with particular emphasis on steady state, linear-quadratic-Gaussian compensators. Topics covered include: (1) the linear-quadratic-Gaussian problem; (2) compensator structures; (3) architectural issues: serialism, parallelism, and pipelining; (4) finite wordlength effects: quantization noise, quantizing the coefficients, and limit cycles; and (5) the optimization of structures.

Moroney, P.↗

Application of advanced multidisciplinary analysis and optimization methods to vehicle design synthesis

Advanced multidisciplinary analysis and optimization methods, namely system sensitivity analysis and non-hierarchical system decomposition, are applied to reduce the cost and improve the visibility of an automated vehicle design synthesis process. This process is inherently complex due to the large number of functional disciplines and associated interdisciplinary couplings. Recent developments in system sensitivity analysis as applied to complex non-hierarchic multidisciplinary design optimization problems enable the decomposition of these complex interactions into sub-processes that can be evaluated in parallel. The application of these techniques results in significant cost, accuracy, and visibility benefits for the entire design synthesis process.

Consoli, Robert David↗

The interaction of collisionless shocks in astrophysical plasmas

The interaction between collisionless shocks, such as may arise in a number of situations in space plasmas, is examined by means of hybrid numerical simulations. The production of energetic particles by shock collisions is investigated. When the shocks are quasi-perpendicular, ions with energies greater than 10E(0) are produced, where E(0) is the kinetic energy of the unshocked ions. The effect is optimized when the two colliding shocks have roughly equal strengths. As the shocks become closer to quasi-parallel, the number and energy of the accelerated ions increases. Energies in excess of 30E(0) are obtained in the quasi-parallel regime, with a small fall-off when the shocks are exactly parallel. For quasi-parallel collisions, the ion energization can be attributed to the interaction of hot ions with both strong electromagnetic waves and the motional electric field located at the shocks. These results are applied to shock collisions in the distant solar wind, at planetary bow shocks, as well as to the impulsive phase of solar flares.

Cargill, P. J.↗

Self-organization in neural networks - Applications in structural optimization

The present paper discusses the applicability of ART (Adaptive Resonance Theory) networks, and the Hopfield and Elastic networks, in problems of structural analysis and design. A characteristic of these network architectures is the ability to classify patterns presented as inputs into specific categories. The categories may themselves represent distinct procedural solution strategies. The paper shows how this property can be adapted in the structural analysis and design problem. A second application is the use of Hopfield and Elastic networks in optimization problems. Of particular interest are problems characterized by the presence of discrete and integer design variables. The parallel computing architecture that is typical of neural networks is shown to be effective in such problems. Results of preliminary implementations in structural design problems are also included in the paper.

Hajela, Prabhat↗

Stereo imaging based particle velocimeter

Three dimensional coordinates of an object are determined from it's two dimensional images for a class of points on the object. Two dimensional images are first filtered by a Laplacian of Gaussian (LOG) filter in order to detect a set of feature points on the object. The feature points on the left and the right images are then matched using a Hopfield type optimization network. The performance index of the Hopfield network contains both local and global properties of the images. Parallel computing in stereo matching can be achieved by the proposed methodology.

Batur, Celal↗

Data parallel sorting for particle simulation

Sorting on a parallel architecture is a communications intensive event which can incur a high penalty in applications where it is required. In the case of particle simulation, only integer sorting is necessary, and sequential implementations easily attain the minimum performance bound of O (N) for N particles. Parallel implementations, however, have to cope with the parallel sorting problem which, in addition to incurring a heavy communications cost, can make the minimun performance bound difficult to attain. This paper demonstrates how the sorting problem in a particle simulation can be reduced to a merging problem, and describes an efficient data parallel algorithm to solve this merging problem in a particle simulation. The new algorithm is shown to be optimal under conditions usual for particle simulation, and its fieldwise implementation on the Connection Machine is analyzed in detail. The new algorithm is about four times faster than a fieldwise implementation of radix sort on the Connection Machine.

Dagum, Leonardo↗

HPF Implementation of ARC3D

We present an HPF (High Performance Fortran) implementation of ARC3D code along with the profiling and performance data on SGI Origin 2000. Advantages and limitations of HPF as a parallel programming language for CFD applications are discussed. For achieving good performance results we used the data distributions optimized for implementation of implicit and explicit operators of the solver and boundary conditions. We compare the results with MPI and directive based implementations.

Frumkin, Michael↗

A Portable 3D FFT Package for Distributed-Memory Parallel Architectures

A parallel algorithm for 3D FFTs is implemented as a series of local 1D FFTs combined with data transposes. This allows the use of vendor supplied (often fully optimized) sequential 1D FFTs. The FFTs are carried out in-place by using an in-place data transpose across the processors.

fast Fourrier transform FFT parallel processing↗

Software Aids Visualization of Computed Unsteady Flow

Unsteady Flow Analysis Toolkit (UFAT) is a computer program that synthesizes motions of time-dependent flows represented by very large sets of data generated in computational fluid dynamics simulations. Prior to the development of UFAT, it was necessary to rely on static, single-snapshot depictions of time-dependent flows generated by flow-visualization software designed for steady flows. Whereas it typically takes weeks to analyze the results of a largescale unsteady-flow simulation by use of steady-flow visualization software, the analysis time is reduced to hours when UFAT is used. UFAT can be used to generate graphical objects of flow visualization results using multi-block curvilinear grids in the format of a previously developed NASA data-visualization program, PLOT3D. These graphical objects can be rendered using FAST, another popular flow visualization software developed at NASA. Flow-visualization techniques that can be exploited by use of UFAT include time-dependent tracking of particles, detection of vortex cores, extractions of stream ribbons and surfaces, and tetrahedral decomposition for optimal particle tracking. Unique computational features of UFAT include capabilities for automatic (batch) processing, restart, memory mapping, and parallel processing. These capabilities significantly reduce analysis time and storage requirements, relative to those of prior flow-visualization software. UFAT can be executed on a variety of supercomputers.

Kao, David↗

Experience in environmental remediation of radioactively contaminated site in waste disposal area in Istanbul - 15011

In this paper specific environmental remediation examples of method applications are given. Radiological contaminated site was created by non-nuclear industry in Turkey. This is designated as a contaminated site and can ultimately lead to undesired health effects for residents. Appropriate actions have been taken to protect human and environment from contaminated site. As every site has its own characteristics, contaminated area was examined by sampling and surveying works. Before taken into action, remediation actions need to be justified and optimized. Radiation levels and homogeneity of contamination was investigated. Remediation program was established and taken into action in parallel two main works. First one covers all actions to the contamination itself. The site was isolated properly and all equipment was deployed. Working methods for segregation of contaminated material from the soil were determined. Besides, risk evaluation was carried out. Pathways between the contaminated soil and people were taken into account. Groundwater samples were taken and analyzed. According to these results, evacuation, area isolation or changing land use options were evaluated. Isolation of the site and removal of radioactive materials were determined as appropriate actions. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Optimal sun-alignment techniques of large solar arrays in electric propulsion spacecraft

Optimum sun-alignment of large solar arrays in electric propulsion spacecraft operating in earth orbit requires periodic roll motions around the thrust axis, synchronized with the apparent conical motion of the sun line. This oscillation is sustained effectively with the aid of gravity gradient torques while only a small share of the total torque is being contributed by the attitude control system. Tuning the system for resonance requires an appropriate choice of moment-of-inertia characteristics. To minimize atmospheric drag at low orbital altitudes the solar array is oriented parallel, or nearly parallel, to the flight direction. This can increase the thrust-to-drag ratio by as much as an order of magnitude. Coupled with optimal roll orientation, this feathering technique will permit use of electric propulsion effectively at low altitudes in support of space shuttle or space station activities and in spiral ascent missions.

Meissinger, H. F.↗

Advances in Multi-Physics Modeling for Parallel Finite Element Code Suite ACE3P

SLAC developed ACE3P code suite is a comprehensive set of parallel finite-element codes for multi-physics modeling of particle accelerators. Running on massively parallel computer platforms for high fidelity and high accuracy simulation, ACE3P enables rapid virtual prototyping of accelerator and RF component design, optimization and analysis. Furthermore, recent advances of ACE3P have been focused on the implementation of advanced numerical algorithms, enhancement of multi-physics capabilities, and improvement of code performance on state-of-the-art high performance computing (HPC) platforms for large-scale RF modeling in accelerator applications and beyond. In this paper, we will present a nonlinear electromagnetic (EM) eigensolver for damping calculation of resonant modes for cavities coupled with waveguides, the thermal solver for accurate evaluation of heat loads in superconducting RF (SRF) cavity power coupler, the elastic solver for investigation of SRF cavity cryomodule (CM) bowing due to thermal effect, the frequency domain mechanical solver for studying SRF cavity mechanical oscillation modes, the particle tracking code for dark current simulation in SRF CM using a hybrid MPI+OpenMP parallel programing model, the time domain solver for high fidelity modeling of EM wave propagation for medical device in human body.

97 MATHEMATICS AND COMPUTING↗