Neural Learning Using Orthogonal Arrays
The paper proposes the use of the orthogonal arrays for neural network learning. Learning can be seen as a search for the neural weights that give an optimal network performance.
SEARCH · Search NASA
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.
The paper proposes the use of the orthogonal arrays for neural network learning. Learning can be seen as a search for the neural weights that give an optimal network performance.
Currently, airport surface communications are fragmented across multiple types of systems. These communication systems for airport operations at most airports today are based dedicated and separate architectures that cannot support system-wide interoperability and information sharing. The requirements placed upon the Communications, Navigation, and Surveillance (CNS) systems in airports are rapidly growing and integration is urgently needed if the future vision of the National Airspace System (NAS) and the Next Generation Air Transportation System (NGATS) 2025 concept are to be realized. To address this and other problems such as airport surface congestion, the Space Based Technologies Project s Surface ICNS Network Architecture team at NASA Glenn Research Center has assessed airport surface communications requirements, analyzed existing and future surface applications, and defined a set of architecture functions that will help design a scalable, reliable and flexible surface network architecture to meet the current and future needs of airport operations. This paper describes the systems approach or methodology to networking that was employed to assess airport surface communications requirements, analyze applications, and to define the surface network architecture functions as the building blocks or components of the network. The systems approach used for defining these functions is relatively new to networking. It is viewing the surface network, along with its environment (everything that the surface network interacts with or impacts), as a system. Associated with this system are sets of services that are offered by the network to the rest of the system. Therefore, the surface network is considered as part of the larger system (such as the NAS), with interactions and dependencies between the surface network and its users, applications, and devices. The surface network architecture includes components such as addressing/routing, network management, network performance and security.
This slide presentation reviews the testing of the Delay/Disruption Tolerant Network (DTN) designed for use with Lunar Surface applications. This is being done through the DTN experimental Network (DEN), that permit access and testing by other NASA centers, DTN team members and protocol developers. The objective of this work is to demonstrate DTN for high return applications in lunar scenarios, provide DEN connectivity with analogs of Constellation elements, emulators, and other resources from DTN Team Members, serve as a wireless communications staging ground for remote analog excursions and enable testing of detailed communication scenarios and evaluation of network performance. Three scenarios for DTN on the Lunar surface are reviewed: Motion imagery, Voice and sensor telemetry, and Navigation telemetry.
An experiment was performed to determine the degradation in the bit-error-rate (BER) in the high-data-rate cables chosen for the Orion Service Module due to extreme launch conditions of vibrations with a magnitude of 60g. The cable type chosen for the Orion Service Module was no. 8 quadrax cable. The increase in electrical noise induced on these no. 8 quadrax cables was measured at the NASA Glenn vibration facility in the Structural Dynamics Laboratory. The intensity of the vibrations was set at 32g, which was the maximum available level at the facility. The cable lengths used during measurements were 1, 4, and 8 m. The noise measurements were done in an analog fashion using a performance network analyzer (PNA) by recording the standard deviation of the transmission scattering parameter S(sub 21) over the frequency range of 100 to 900 MHz. The standard deviation of S(sub 210 was measured before, during, and after the vibration of the cables at the vibration facility. We observed an increase in noise by a factor of 2 to 6. From these measurements we estimated the increase expected in the BER for a cable length of 25 m and concluded that these findings are large enough that the noise increase due to vibration must be taken in to account for the design of the communication system for a BER of 10(exp -8).
The International Space Station (ISS) is in an operational configuration. To fully utilize the ISS and take advantage of the modern protocols and updated Ku-band access, the Huntsville Operations Support Center (HOSC) has designed an approach to extend the Kuband forward link access for payload investigators to their on-orbit payloads. This dramatically increases the ground to ISS communications for those users. This access also enables the ISS flight controllers operating in the Payload Operations and Integration Center to have more direct control over the systems they are responsible for managing and operating. To extend the Ku-band forward link to the payload user community the development of a new command server is necessary. The HOSC subsystems were updated to process the Internet Protocol Encapsulated packets, enable users to use the service based on their approved services, and perform network address translation to insure that the packets are forwarded from the user to the correct payload repeating that process in reverse from ISS to the payload user. This paper presents the architecture, implementation, and lessons learned. This will include the integration of COTS hardware and software as well as how the device is incorporated into the operational mission of the ISS. Thus, this paper also discusses how this technology can be applicable to payload users of the ISS.
We introduce the Precipitation Probability DISTribution (PPDIST) dataset, a collection of global high-resolution (0.1°) observation-based climatologies (1979–2018) of the occurrence and peak intensity of precipitation (P) at daily and 3-hourly time-scales. The climatologies were produced using neural networks trained with daily P observations from 93,138 gauges and hourly P observations (resampled to 3-hourly) from 11,881 gauges worldwide. Mean validation coefficient of determination (R^(2)) values ranged from 0.76 to 0.80 for the daily P occurrence indices, and from 0.44 to 0.84 for the daily peak P intensity indices. The neural networks performed significantly better than current state-of-the-art reanalysis (ERA5) and satellite (IMERG) products for all P indices. Using a 0.1 mm 3 per h threshold, P was estimated to occur 12.2%, 7.4%, and 14.3% of the time, on average, over the global, land, and ocean domains, respectively. The highest P intensities were found over parts of Central America, India, and Southeast Asia, along the western equatorial coast of Africa, and in the intertropical convergence zone.
Computer centers such as NERSC and OLCF have traditionally focused on delivering computational capability that enables breakthrough innovation in a wide range of science domains. Accessing that computational power has required services and tools to move the data from input and output to computation and storage. A ''pivot to data'' is occurring in HPC. Data transfer tools and services that were previously peripheral are becoming integral to scientific workflows. Emerging requirements from high-bandwidth detectors, high-throughput screening techniques, highly concur- rent simulations, increased focus on uncertainty quantification, and an emerging open-data policy posture toward published research are among the data-drivers shaping the networks, file systems, databases, and overall compute and data environment. In this paper we explain the pivot to data in HPC through user requirements and the changing resources provided by HPC with particular focus on data movement. For WAN data transfers we present the results of a study of network performance between centers.
Avalaunch implements a tree-based process launcher. It first bootstraps itself on to a set of compute nodes by launching children processes, which immediately connect back to the parent process to acquire info needed t launch their own children. Once the tree is established, user processes are started by broadcasting commands and application binaries through the tree. All communication flows over high-performance network protocols via spawnnet. The goal is to start MPI jobs having hundreds of thousands of processes within seconds.
Systems and methods of building massively parallel computing systems using low power computing complexes in accordance with embodiments of the invention are disclosed. A massively parallel computing system in accordance with one embodiment of the invention includes at least one Solid State Blade configured to communicate via a high performance network fabric. In addition, each Solid State Blade includes a processor configured to communicate with a plurality of low power computing complexes interconnected by a router, and each low power computing complex includes at least one general processing core, an accelerator, an I/O interface, and cache memory and is configured to communicate with non-volatile solid state memory.
As a representative of Partitioned Global Address Space models, OpenSHMEM provides a variety of functionalities including one-sided communication, atomic operations, and collective routines. The communication layer of OpenSHMEM-X plays a crucial role for its functionalities. OFI Libfabric is an open-source network library that supports portable low-latency interfaces from different fabric providers while minimizing the semantic gap across API endpoints. In this paper, we present the design and implementation of OpenSHMEM-X communication conduit using Libfabric. This Libfabric conduit is designed to support a broad range of network providers while achieving excellent network performance and scalability. We have performed an extensive set of experiments to validate the performance of our implementation, and compared with the Sandia OpenSHMEM implementation. Our results show that the Libfabric conduit improves the communication bandwidth on the socket provider by up to 42% and 11%, compared to an alternative OpenSHMEM implementation for put and get operations, respectively. In addition, our implementation of atomic operations has achieved similar latency to that of the Sandia implementation.
Systems and methods of building massively parallel computing systems using low power computing complexes in accordance with embodiments of the invention are disclosed. A massively parallel computing system in accordance with one embodiment of the invention includes at least one Solid State Blade configured to communicate via a high performance network fabric. In addition, each Solid State Blade includes a processor configured to communicate with a plurality of low power computing complexes interconnected by a router, and each low power computing complex includes at least one general processing core, an accelerator, an I/O interface, and cache memory and is configured to communicate with non-volatile solid state memory.
Systems and methods of building massively parallel computing systems using low power computing complexes in accordance with embodiments of the invention are disclosed. A massively parallel computing system in accordance with one embodiment of the invention includes at least one Solid State Blade configured to communicate via a high performance network fabric. In addition, each Solid State Blade includes a processor configured to communicate with a plurality of low power computing complexes interconnected by a router, and each low power computing complex includes at least one general processing core, an accelerator, an I/O interface, and cache memory and is configured to communicate with non-volatile solid state memory.
Microencapsulated phase change materials are being studied for applications for thermal energy storage in concentrated solar fields. During fabrication, the thickness of the encapsulation cannot be readily measured for real-time control. Therefore, a machine learning network, specifically a Long Short-Term Memory network, is being developed to estimate the ratio of the shell radius to core radius based on a one second temperature history. The mini-batch size determines how often the algorithm weights are updated during network training, and shuffle indicates whether the training data is shuffled during training. A general factorial design is used to analyze the effects of varying mini-batch size and shuffle, along with the core-to-shell ratio, on the RMSE of the response from the Long Short-Term Memory network. It was found that the network performed better for smaller core to shell ratios (less than 0.6) and had the lowest RMSE when the minibatch size was 128. The minimum RMSE found was 0.00501.
Many fields within scientific computing have embraced advances in big-data analysis and machine learning, which often requires the deployment of large, distributed and complicated workflows that may combine training neural networks, performing simulations, running inference, and performing database queries and data analysis in asynchronous, parallel and pipelined execution frameworks. Such a shift has brought into focus the need for scalable, efficient workflow management solutions with reproducibility, error and provenance handling, traceability, and checkpoint-restart capabilities, among other needs. Here, we discuss challenges and best-practices for deploying exascale-generation computational science workflows on resources at the Oak Ridge Leadership Computing Facility (OLCF). We present our experiences with large-scale deployment of distributed workflows on the Summit supercomputer, including for bioinformatics and computational biophysics, materials science, and deep learning model optimization. We also present problems and solutions created by working within a Python-centric software base on traditional HPC systems, and discuss steps that will be required before the convergence of HPC, AI, and data science can be fully realized. Our results point to a wealth of exciting new possibilities for harnessing this convergence to tackle new scientific challenges.
High-rate Delay Tolerant Networking (HDTN) is a performance-optimized Delay Tolerant Networking (DTN) implementation that can provide multigigabit per second data rates in disruptive and high-latency space networks. Routing, especially in large-scale space networks, remains challenging due to network topologies that evolve over time. This paper presents a simulation tool that enables HDTN implementation testing at accelerated speeds, which is key for routing in large-scale space networks.
High-performance polymer made by new synthesis in which one or more easy-to-process, but brittle, thermosetting polyimides combined with one or more tough, but difficult-to-process, linear thermoplastics to yield semi-interpenetrating polymer network (semi-IPN) having combination of easy processability and high tolerance to damage. Two commercially available resins combined to form tough, semi-IPN called "LaRC-RP49." Displays improvements in toughness and resistance to microcracking. LaRC-RP49 has potential as high-temperature matrix resin, adhesive, and molding resin. Useful in aerospace, automotive, and electronic industries.
Not Available