Search NASA⌕ Search

SEARCH · Search NASA

Results for “exploration techniques”

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 199 records · Page 11

Towards Low-Overhead Resilience for Data Parallel Deep Learning

Data parallel techniques have been widely adopted both in academia and industry as a tool to enable scalable training of deep learning models. At scale, DL training jobs can fail due to software or hardware bugs, may need to be preempted or terminated due to unexpected events, or may perform suboptimally because they were misconfigured. Under such circumstances, there is a need to recover and/or reconfigure data-parallel DL training jobs on-the-fly, while minimizing the impact on the accuracy of the DNN model and the runtime overhead. In this regard, state-of-art techniques adopted by the HPC community mostly rely on checkpoint-restart, which inevitably leads to loss of progress, thus increasing the runtime overhead. In this paper we explore alternative techniques that exploit the properties of modern deep learning frameworks (overlapping of gradient averaging and weight updates with local gradient computations through pipeline parallelism) to reduce the overhead of resilience/elasticity. To this end we introduce a failure simulation framework and two resilience strategies (immediate mini-batch rollback and lossy forward recovery), which we study compared with checkpoint-restart approaches in a variety of settings in order to understand the trade-offs between the accuracy loss of the DNN model and the runtime overhead.

data-parallel training↗

Analysis of the Thermal Stability of Very Thin Surface Layers of Corrosion Inhibitors by Time-of-Flight Secondary Ion Mass Spectrometry

The powerful nature of the secondary ion mass spectrometry (SIMS) technique was explored in order to analyse very thin surface layers that were self-assembled on steel material from acidic solution. These surface layers are adsorbed corrosion inhibitors. The SIMS technique proved useful to characterise the molecular structure and homogeneity of thin surface layers in the nanometre range of specific analytes on the metallic substrate. Using SIMS, the thermal stability of these layers was further investigated and the desorption energy at a certain temperature was determined, where special attention was devoted to the method’s static limit. In order to compare, and for certain cases emphasise, the benefits gained by using SIMS in such surface analysis compared with the X-ray photoelectron spectroscopy (XPS) method, the same samples were also analysed by means of the latter. XPS is usually considered to be the most powerful analytical tool in surface analysis studies, but, as shown herein, it has certain limitations compared to SIMS. Finally, the surface topography was investigated by employing atomic force microscopy (AFM) in order to carry out a comprehensive surface analysis. .

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Application Deconfliction Characterization and Alternatives Analysis

This report provides an overview of the domain space and solution techniques that could be used to create a robust, flexible app deconfliction service. Three approaches are reviewed with summaries of the characteristics, elements, and results from preliminary demonstrations of solution techniques based on each approach: 1) rules and heuristics, 2) cooperation, and 3) optimization. The strengths and weaknesses of each solution technique were explored through a set of numerical demonstrations on modified IEEE 123 node and 9500 node test feeders. An alternatives analysis of individual deconfliction elements was performed with each solution technique element evaluated against criteria reflecting the dynamic app environment, need to balance app objectives, and scalability issues versus the number of applications, setpoints, and distributed control areas. It is anticipated that a combined solution for a GridAPPS-D Deconfliction Service can be formulated using a combination of elements from each solution technique. The combined solution would combine 1) device control budgets to reduce the size of the solution space by constraining system setpoints to those will not result in accelerated degradation of physical assets, 2) system operations rules to constrain the solution space by eliminating setpoints that result in violations of system limits or operational best practices, 3) contextual status signals shared with or among apps such that they could update their desired setpoints based on the evolving context, 4) a mediator that incentivizes apps to come to a cooperative solution, and 5) Setpoint-informed optimization as a fallback mechanism if a cooperative solution cannot be agreed upon by applications.

24 POWER TRANSMISSION AND DISTRIBUTION↗

OpenCRUMS USA: An Open Machine Learning Framework for Characterizing Variability in Aerosol Reanalysis Data

Advances in artificial intelligence (AI) have called for exploring how these techniques can be used for exploring patterns in large climate datasets. To that regard, the U.S. Department of Energy AI for Earth System Predictability (AI4ESP) supported a pilot initiative called the Open Classification of Regimes in the Southeast USA (OpenCRUMS USA) project to explore how AI can be used to characterize modes of spatial variability in large climate datasets. For this study, we focus on comparing two methods for characterizing the modes of spatial variability of surface aerosol concentration over the Houston region: empirical orthogonal functions (EOFs) and layerwise relevance propagation (LRP) applied to a convolutional neural network (CNN) classifier. We show that EOF analysis typically attributes spatial variability modes that span all of southeast Texas, prohibiting the attribution of spatial variability to localized regions. However, using LRP on the CNN classifier resolves the explanatory parameters at a finer spatial resolution than EOFs. This allows for the attribution of the spatial variability of surface aerosols to local regions of organic carbon which was not possible using EOFs. In addition, the LRP analysis also suggests that synoptic-scale transport of dust is most prevalent during anticyclonic and pretrough synoptic conditions as categorized by self-organizing maps.

54 ENVIRONMENTAL SCIENCES↗

Spatial filtering with photonic crystals

Photonic crystals are well known for their celebrated photonic band-gaps—the forbidden frequency ranges, for which the light waves cannot propagate through the structure. The frequency (or chromatic) band-gaps of photonic crystals can be utilized for frequency filtering. In analogy to the chromatic band-gaps and the frequency filtering, the angular band-gaps and the angular (spatial) filtering are also possible in photonic crystals. In this article, we review the recent advances of the spatial filtering using the photonic crystals in different propagation regimes and for different geometries. We review the most evident configuration of filtering in Bragg regime (with the back-reflection—i.e., in the configuration with band-gaps) as well as in Laue regime (with forward deflection—i.e., in the configuration without band-gaps). We explore the spatial filtering in crystals with different symmetries, including axisymmetric crystals; we discuss the role of chirping, i.e., the dependence of the longitudinal period along the structure. We also review the experimental techniques to fabricate the photonic crystals and numerical techniques to explore the spatial filtering. Finally, we discuss several implementations of such filters for intracavity spatial filtering.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Computed Tomography of Transverse Phase Space

Two computed tomography techniques are explored to reconstruct beam transverse phase space using both simulated beam and multi-wire profile data in the Fermilab Muon Test Area ("MTA") beamline. Both Filtered Back-Projection ("FBP") and Simultaneous Algebraic Reconstruction Technique ("SART") algorithms [2] are considered and compared. Errors and artifacts are compared as a function of each algorithm’s free parameters, and it is shown through simulation and MTA beamline profiles that SART is advantageous for reconstructions with limited profile data.

43 PARTICLE ACCELERATORS↗

A Global Data Analysis for Representing Sediment and Particulate Organic Carbon Yield in Earth System Models

Although sediment yield (SY) from water erosion is ubiquitous and its environmental consequences are well recognized, its impacts on the global carbon cycle remain largely uncertain. This knowledge gap is partly due to the lack of soil erosion modeling in Earth System Models (ESMs), which are important tools used to understand the global carbon cycle and explore its changes. As such, this study analyzed sediment and particulate organic carbon yield (CY) data from 1081 and 38 small catchments (0.1-200 km 2 ), respectively, in different environments across the globe. Using multiple statistical analysis techniques, we explored environmental factors and hydrological processes important for SY and CY modeling in ESMs. Our results show clear correlations of high SY with traditional agriculture, seismicity and heavy storms, as well as strong correlations between SY and annual peak runoff. These highlight the potential limitation of SY models that represent only interrill and rill erosion because shallow overland flow and rill flow have limited transport capacity due to their hydraulic geometry to produce high SY. Further, our results suggest that SY modeling in ESMs should be implemented at the event scale to produce the catastrophic mass transport during episodic events. Several environmental factors such as seismicity and land management that are often not considered in current catchment-scale SY models can be important in controlling global SY. Our analyses show that SY is likely the primary control on CY in small catchments and a statistically significant empirical relationship is established to calculate SY and CY jointly in ESMs.

54 ENVIRONMENTAL SCIENCES↗

Optical emission and nanoparticle generation in Al plasmas using ultrashort laser pulses temporally optimized by real-time spectroscopic feedback

With an interest in pulsed laser deposition and remote spectroscopy techniques, we explore here the potential of laser pulses temporally tailored on ultrafast time scales to control the expansion and the excitation degree of various ablation products including atomic species and nanoparticulates. Taking advantage of automated pulse-shaping techniques, an adaptive procedure based on spectroscopic feedback is applied to regulate the irradiance and enhance the optical emission of monocharged aluminum ions with respect to the neutral signal. This leads to optimized pulses usually consisting in a series of femtosecond peaks distributed on a longer picosecond sequence. The ablation features induced by the optimized pulse are compared with those determined by picosecond pulses generated by imposed second-order dispersion or by double pulse sequences with adjustable picosecond separation. This allows to analyze the influence of fast- and slow-varying envelope features on the material heating and the resulting plasma excitation degree. Using various optimal pulse forms including designed asymmetric shapes, we analyze the establishment of surface pre-excitation that enables conditions of enhanced radiation coupling. Thin films elaborated by unshaped femtosecond laser pulses and by optimized, stretched, or double pulse sequences are compared, indicating that the nanoparticles generation efficiency is strongly influenced by the temporal shaping of the laser irradiation. A thermodynamic scenario involving supercritical heating is proposed to explain enhanced ionization rates and lower particulates density for optimal pulses. Numerical one-dimensional hydrodynamic simulations for the excited matter support the interpretation of the experimental results in terms of relative efficiency of various relaxation paths for excited matter above or below the thermodynamic stability limits. The calculation results underline the role of the temperature and density gradients along the ablated plasma plume which lead to the spatial distinct locations of excited species. Moreover, the nanoparticles sizes are computed based on liquid layer ejection followed by a Rayleigh and Taylor instability decomposition, in good agreement with the experimental findings.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Membrane Structure Studies by Means of Small-Angle Neutron Scattering (SANS)

The basic model for membrane structure--a lipid bilayer with imbedded proteins--was formulated 35 years ago, however the detailed structure is still under active investigation using a variety of physical, chemical and computational techniques. Every biologically active cell is encapsulated by a plasma membrane with most cells also equipped with an extensive intracellular membrane system. The plasma membrane is an important boundary between the cytoplasm of the cell and the external environment, and selectively isolates the cell from that environment. Passive diffusion and/or active transport mechanisms are provided for water, ions, substrates etc. which are vital for cell metabolism and viability. Membranes also facilitate excretion of substances either as useful cellular products or as waste. Despite their complexity and diverse function, plasma membranes from quite different cells have surprisingly similar compositions. A typical membrane structure consists of a phospholipid bilayer with a number of proteins scattered throughout, along with carbohydrates (glycoproteins), glycolipids and sterols. The plasma membranes of most eukaryotic cells contain approximately equal weights of lipid and protein, which corresponds to about 100 lipid molecules per protein molecule. Clearly, lipids are a major constituent and the study of their structure and function in isolation provides valuable insight into the more complex intact multicomponent membrane. The membrane bound protein is the other major constituent and is a very active area of research for a number of reasons including the fact that over 60% of modern drugs act on their receptor sites. The interaction between the protein and the supporting lipid bilayer is clearly of major importance. Neutron scattering is a powerful technique for exploring the structure of membranes, either as reconstituted membranes formed from well characterised lipids, or as intact membranes isolated from selected biological systems. A brief summary of membrane structure will be followed by an outline of the neutron scattering techniques used to understand membrane structure and dynamics. The emphasis will be on the small angle neutron scattering technique since there is a very powerful instrument at Serpong, however brief mention of other techniques will be included to demonstrate how a multidisciplinary approach is usually required.

60 APPLIED LIFE SCIENCES↗

Explorative Visualization for Traffic Safety using Adaptive Study Areas

The pressing need to improve traffic safety has become a societal concern in many cities around the world. Many traffic accidents are not occurring as stand-alone events but as consequences of other road incidents and hazards. To capture the traffic safety indications from a holistic aspect, this paper presents a suite of visualization techniques to explore large traffic safety datasets collected from different sources using adaptive study areas which include the whole region (Hamilton County, Ohio, U.S.) as well as smaller sub-areas. In the present study, these data source include (1) Hamilton County’s 911 emergency response data, which includes traffic incidents as well as other types of incidents throughout the county, and (2) Tennessee crash data, which contains only vehicle crashes with more detail on the circumstances of each crash. We find that both abstract and spatial visualization techniques are used to derive a better understanding of traffic safety patterns for different traffic participants in various urban environments. In addition to the entire region of Hamilton County, safety is examined on the highways, in the downtown area, and in a shopping district east of the city center. It is possible to characterize incidents in the different areas, gain a better understanding of common incident patterns, and identify outliers in the data. Finally, a textured tile calendar is presented to compare spatiotemporal patterns.

99 GENERAL AND MISCELLANEOUS↗

Understanding PV Polymer Backsheet Degradation through X-Ray Scattering

Understanding how photovoltaic (PV) module backsheet polymers age and degrade in response to environmental stresses is important for designing polymers that can maintain their structural integrity after decades of outdoor use. X-ray scattering is a powerful technique for exploring backsheet polymer structure at the angstrom- (wide-angle, WAXS) and nanometer- (small-angle, SAXS) length-scales. We present the use of SAXS and WAXS to study pristine and aged polymer backsheets. The structural insight from these techniques can be used to compare the degradation induced by accelerated testing with the degradation seen in field-aged PV materials.

14 SOLAR ENERGY↗

Emerging Functionality in Transition-Metal Compounds Driven by Spatial Confinement and Broken Symmetry

This research project investigates the emerging functionality in transition-metal-compounds (TMCs) driven by spatial confinement and broken symmetry. It combines advanced growth capabilities with cutting-edge characterization and first principles theory to probe and control the properties of TMC interfaces, including the utilization and development of state-of-the-art atomically resolved electron microscopy and spectroscopy to determine the structure, composition, and bonding at TMC interfaces. The proposed research will focus on four challenging areas: 1) manipulate interfaces to design new material phases such as magnetic metals with unique polar structure (net dipole) to achieve multiple functionality; 2) explore electronic mismatch or screening at interfaces of insulating/poor metal TMCs to produce novel electronic and magnetic properties; 3) elucidate and exploit the role of defects, both point and extended, on the functionality of interfaces; 4) develop advanced electron microscopy/spectroscopy techniques to explore temperature dependent phase transitions and couple these structural tools with new nonlinear optical probes of the electronic structure. The research team aims to close the materials-by-design loop of make, measure, model, and modify. The program promises to enhance our ability to engineer the desired physical properties at interfaces, superlattices (periodic arrays of films of different compounds), and heterostructures of TMCs.

36 MATERIALS SCIENCE↗

On the Impact of Execution Models: A Case Study in Computational Chemistry

Efficient utilization of high-performance computing (HPC) platforms is an important and complex problem. Execution models, abstract descriptions of the dynamic runtime behavior of the execution stack, have significant impact on the utilization of HPC systems. Using a computational chemistry kernel as a case study and a wide variety of execution models combined with load balancing techniques, we explore the impact of execution models on the utilization of an HPC system. We demonstrate a 50 percent improvement in performance by using work stealing relative to a more traditional static scheduling approach. We also use a novel semi-matching technique for load balancing that has comparable performance to a traditional hypergraph-based partitioning implementation, which is computationally expensive. Using this study, we found that execution model design choices and assumptions can limit critical optimizations such as global, dynamic load balancing and finding the correct balance between available work units and different system and runtime overheads. With the emergence of multi- and many-core architectures and the consequent growth in the complexity of HPC platforms, we believe that these lessons will be beneficial to researchers tuning diverse applications on modern HPC platforms, especially on emerging dynamic platforms with energy-induced performance variability.

Execution models, computational chemistry, MPI and↗

Novel Probes Project: Tests of gravity on astrophysical scales

The Novel Probes Project, an initiative to advance the field of astrophysical tests of the dark sector by creating a forum that connects observers and theorists, is introduced. This review focuses on tests of gravity and is intended to be of use primarily to observers, as well as theorists with an interest in the development of experimental tests. It is twinned with a separate upcoming review on dark matter self-interactions. The review focuses on astrophysical tests of gravity in the weak-field regime, ranging from stars to quasilinear cosmological scales. This regime is complementary to both strong-field tests of gravity and background and linear probes in cosmology. In particular, the nonlinear screening mechanisms that are an integral part of viable modified-gravity models lead to characteristic signatures, specifically on astrophysical scales. The potential of these probes is not limited by cosmic variance but comes with the challenge of building robust theoretical models of the nonlinear dynamics of stars, galaxies, and large-scale structure. The groundwork is laid for a thorough exploration of the weak-field, nonlinear regime, with an eye to using the current and next generation of observations for tests of gravity. The scene is set by showing how gravitational theories beyond general relativity are expected to behave, focusing primarily on screening mechanisms. Analytic and numerical techniques for exploring the relevant astrophysical regime are described, as are the pertinent observational signals. With these in hand a range of astrophysical tests of gravity are presented, and prospects for future measurements and theoretical developments are discussed.

79 ASTRONOMY AND ASTROPHYSICS↗

Micro-Kelvin cold molecules.

We have developed a novel experimental technique for direct production of cold molecules using a combination of techniques from atomic optical and molecular physics and physical chemistry. The ability to produce samples of cold molecules has application in a broad spectrum of technical fields high-resolution spectroscopy, remote sensing, quantum computing, materials simulation, and understanding fundamental chemical dynamics. Researchers around the world are currently exploring many techniques for producing samples of cold molecules, but to-date these attempts have offered only limited success achieving milli-Kelvin temperatures with low densities. This Laboratory Directed Research and Development project is to develops a new experimental technique for producing micro-Kelvin temperature molecules via collisions with laser cooled samples of trapped atoms. The technique relies on near mass degenerate collisions between the molecule of interest and a laser cooled (micro-Kelvin) atom. A subset of collisions will transfer all (nearly all) of the kinetic energy from the 'hot' molecule, cooling the molecule at the expense of heating the atom. Further collisions with the remaining laser cooled atoms will thermally equilibrate the molecules to the micro-Kelvin temperature of the laser-cooled atoms.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

A Bandwidth-Optimized Multi-Core Architecture for Irregular Applications

This paper presents an architecture template for next-generation high performance computing systems specifically targeted to irregular applications. We start our work by considering that future generation interconnection and memory bandwidth full-system numbers are expected to grow by a factor of 10. In order to keep up with such a communication capacity, while still resorting to fine-grained multithreading as the main way to tolerate unpredictable memory access latencies of irregular applications, we show how overall performance scaling can benefit from the multi-core paradigm. At the same time, we also show how such an architecture template must be coupled with specific techniques in order to optimize bandwidth utilization and achieve the maximum scalability. We propose a technique based on memory references aggregation, together with the related hardware implementation, as one of such optimization techniques. We explore the proposed architecture template by focusing on the Cray XMT architecture and, using a dedicated simulation infrastructure, validate the performance of our template with two typical irregular applications. Our experimental results prove the benefits provided by both the multi-core approach and the bandwidth optimization reference aggregation technique.

bandwidth optimization↗

The integration of system specifications and program coding

Experience in maintaining up-to-date documentation for one module of the large-scale Medical Literature Analysis and Retrieval System 2 (MEDLARS 2) is described. Several innovative techniques were explored in the development of this system's data management environment, particularly those that use PL/I as an automatic documenter. The PL/I data description section can provide automatic documentation by means of a master description of data elements that has long and highly meaningful mnemonic names and a formalized technique for the production of descriptive commentary. The techniques discussed are practical methods that employ the computer during system development in a manner that assists system implementation, provides interim documentation for customer review, and satisfies some of the deliverable documentation requirements.

Luebke, W. R.↗

Correlative analysis of hard and soft x ray observations of solar flares

We have developed a promising new technique for jointly analyzing BATSE hard X-ray observations of solar flares with simultaneous soft X-ray observations. The technique is based upon a model in which electric currents and associated electric fields are responsible for the respective heating and particle acceleration that occur in solar flares. A useful by-product of this technique is the strength and evolution of the coronal electric field. The latter permits one to derive important flare parameters such as the current density, the number of current filaments composing the loop, and ultimately the hard X-ray spectrum produced by the runaway electrons. We are continuing to explore the technique by applying it to additional flares for which we have joint BATSE/Yohkoh observations. A central assumption of our analysis is the constant of proportionality alpha relating the hard X-ray flux above 50 keV and the rate of electron acceleration. For a thick-target model of hard X-ray production, it can be shown that cv is in fact related to the spectral index and low-energy cutoff of precipitating electrons. The next step in our analysis is to place observational constraints on the latter parameters using the joint BATSE/Yohkoh data.

Zarro, Dominic M.↗