Durango: Scalable Synthetic Workloard Generation for Extreme-Scale Application Performance Modeling & Simulation
Explore the source record for details and available documents.
Carothers, Christopher D.↗
SEARCH · Search NASA
Results for “Network simulation”
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.
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↗
A FD/DAMA network architecture for the first generation land mobile satellite services
A frequency division/demand assigned multiple access (FD/DAMA) network architecture for the first-generation land mobile satellite services is presented. Rationales and technical approaches are described. In this architecture, each mobile subscriber must follow a channel access protocol to make a service request to the network management center before transmission for either open-end or closed-end services. Open-end service requests will be processed on a blocked call cleared basis, while closed-end requests will be processed on a first-come-first-served basis. Two channel access protocols are investigated, namely, a recently proposed multiple channel collision resolution scheme which provides a significantly higher useful throughput, and the traditional slotted Aloha scheme. The number of channels allocated for either open-end or closed-end services can be adaptively changed according to aggregated traffic requests. Both theoretical and simulation results are presented. Theoretical results have been verified by simulation on the JPL network testbed.
Yan, T.-Y.↗
Streaming data analytics via message passing with application to graph algorithms
The need to process streaming data, which arrives continuously at high-volume in real-time, arises in a variety of contexts including data produced by experiments, collections of environmental or network sensors, and running simulations. Streaming data can also be formulated as queries or transactions which operate on a large dynamic data store, e.g. a distributed database. We describe a lightweight, portable framework named PHISH which enables a set of independent processes to compute on a stream of data in a distributed-memory parallel manner. Datums are routed between processes in patterns defined by the application. PHISH can run on top of either message-passing via MPI or sockets via ZMQ. The former means streaming computations can be run on any parallel machine which supports MPI; the latter allows them to run on a heterogeneous, geographically dispersed network of machines. We illustrate how PHISH can support streaming MapReduce operations, and describe streaming versions of three algorithms for large, sparse graph analytics: triangle enumeration, subgraph isomorphism matching, and connected component finding. Lastly, we also provide benchmark timings for MPI versus socket performance of several kernel operations useful in streaming algorithms.
97 MATHEMATICS AND COMPUTING↗
Population dynamics, information transfer, and spatial organization in a chemical reaction network under spatial confinement and crowding conditions
Here, we investigate, via Brownian dynamics simulations, the reaction dynamics of a generic, nonlinear chemical network under spatial confinement and crowding conditions. In detail, the Willamowski-Rossler chemical reaction system has been “extended” and considered as a prototype reaction-diffusion system. These results are potentially relevant to a number of open problems in biophysics and biochemistry, such as the synthesis of primitive cellular units (protocells) and the definition of their role in the chemical origin of life and the characterization of vesicle-mediated drug delivery processes. More generally, the computational approach presented in this work makes the case for the use of spatial stochastic simulation methods for the study of biochemical networks in vivo where the “well-mixed” approximation is invalid and both thermal and intrinsic fluctuations linked to the possible presence of molecular species in low number copies cannot be averaged out.
37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
Understanding Proton Movement in [Fe-Fe] Hydrogenases
Nature uses specialized metalloenzymes to carry out small molecule activation reactions, including CO 2 fixation, O 2 activation, and proton reduction, with unparalleled efficiency, rates, and selectivity. The latter reactions are performed by hydrogenases, protein metallo-complexes that interconverts H 2 to protons and electrons (H 2 oxidation) and the reverse reaction (H 2 production) with incredibly low energy input and amazingly fast kinetics. Reproducing both the activity and efficiency of metalloenzymes in sustainable anthropogenic systems remains one of the “holy grails” of inorganic chemistry. However, identifying the precise molecular components responsible for these desirable properties has been challenging in the natural metalloenzymes, hindering efforts to develop analogous processes in synthetic compounds. Considering the inherent complexity of a metalloenzyme and the many interactions, both strong and weak, that contribute to the function of an enzyme, we have elected to model natural metalloenzymes on a biochemical platform. Towards this end, we have a developed structural, functional, and mechanistic mimic of the [Ni-Fe] hydrogenases within a robust protein scaffold, rubredoxin, to understand the influence of the secondary coordination environment on the metal center. This involved preparing a series of three rubredoxin constructs containing a single point mutation at Val positions and making the NMR chemical shift assignments for the paramagnetic (nickel-substituted) and non-paramagnetic (zinc-substituted) form. These physical studies were complemented with in silico molecular dynamic studies on proteins with [Fe-S] clusters to engineer new proton channels to test in vitro . To assist our search for new natural metalloprotein scaffolds within the vast number of sequenced genomes, we created a neural network-based program to identify proteins with specific metal-binding sites. These computational and physical studies with metalloenzymes provide direct insight into the fundamental chemical principles driving the natural systems and offer design principles for developing catalysts that utilize analogous principles.
37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
Optimizing Signal Timing Control for Large Urban Traffic Networks Using an Adaptive Linear Quadratic Regulator Control Strategy
We discuss how traffic signal control is important for intersection safety and efficiency. However, most traffic signal control methods are designed for individual intersections or corridors. Although some adaptive control systems have been developed, the methods used are often proprietary and not published, making it difficult to evaluate their effectiveness. This study proposes an adaptive multi-input and multi-output traffic signal control method that not only can improve network-wide traffic operations in terms of reduced traffic delay and energy consumption, but also is more computationally feasible than existing centralized signal control methods. Considering intersection interactions, a linear dynamic traffic system model was built and adaptively updated to reflect how the signal control input of each intersection affects network-wide vehicle travel delay. Based on the system model, an adaptive linear-quadratic regulator (LQR) was designed to minimize both traffic delay and incremental changes in the control input. The proposed control method was evaluated in a microscopic traffic simulation environment with a 35-intersection network of Bellevue City, Washington. Simulation results show that the proposed method had shorter average traffic delays in the network when compared with the traffic delays controlled by the state-of-the-art max-pressure, self-organizing traffic lights, and independent deep Q network methods.
42 ENGINEERING↗
Deep-Learning Electron Diffractive Imaging
Here, we report the development of deep-learning coherent electron diffractive imaging at subangstrom resolution using convolutional neural networks (CNNs) trained with only simulated data. We experimentally demonstrate this method by applying the trained CNNs to recover the phase images from electron diffraction patterns of twisted hexagonal boron nitride, monolayer graphene, and a gold nanoparticle with comparable quality to those reconstructed by a conventional ptychographic algorithm. Fourier ring correlation between the CNN and ptychographic images indicates the achievement of a resolution in the range of 0.70 and 0.55 Å. We further develop CNNs to recover the probe function from the experimental data. The ability to replace iterative algorithms with CNNs and perform real-time atomic imaging from coherent diffraction patterns is expected to find applications in the physical and biological sciences.
47 OTHER INSTRUMENTATION↗
Pilot Workload Rating Predictions Using Image Data and Recurrent Neural Networks
In this work, we augmented existing methods for estimating pilot workload ratings with deep neural networks trained using data from simulated flight tests in the Vertical Motion Simulator (VMS). We used an existing method, Spare Capacity Operations Estimator (SCOPE), along with a recurrent neural network and conducted comparison studies between the two methods individually, and when used together. We found that using both methods together can improve the result over using either approach alone. In our first test case, we achieved an improved linear correlation coefficient of 0.409 over that of SCOPE alone at 0.352 on the training dataset. Through cross validation, we also found that the results may be dependent on the split of training vs. validation data, and that further investigation should be conducted to understand what additional inputs to the neural network model should be made.
Image Data↗
Explosive synchronization coexists with classical synchronization in the Kuramoto model
Explosive synchronization has recently been reported in a system of adaptively coupled Kuramoto oscillators, without any conditions on the frequency or degree of the nodes. Here, we find that, in fact, the explosive phase coexists with the standard phase of the Kuramoto oscillators. We determine this by extending the mean-field theory of adaptively coupled oscillators with full coupling to the case with partial coupling of a fraction f. This analysis shows that a metastable region exists for all finite values of f > 0, and therefore explosive synchronization is expected for any perturbation of adaptively coupling added to the standard Kuramoto model. We verify this theory with GPU-accelerated simulations on very large networks (N ∼ 10{sup 6}) and find that, in fact, an explosive transition with hysteresis is observed for all finite couplings. By demonstrating that explosive transitions coexist with standard transitions in the limit of f → 0, we show that this behavior is far more likely to occur naturally than was previously believed.
71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
Prediction of carbon nanostructure mechanical properties and the role of defects using machine learning
Graphene-based nanostructures hold immense potential as strong and lightweight materials, however, their mechanical properties such as modulus and strength are difficult to fully exploit due to challenges in atomic-scale engineering. This study presents a database of over 2,000 pristine and defective nanoscale CNT bundles and other graphitic assemblies, inspired by microscopy, with associated stress–strain curves from reactive molecular dynamics (MD) simulations using the reactive INTERFACE force field (IFF-R). These 3D structures, containing up to 80,000 atoms, enable detailed analyses of structure-stiffness-failure relationships. By leveraging the database and physics- and chemistry-informed machine learning (ML), accurate predictions of elastic moduli and tensile strength are demonstrated at speeds 1,000 to 10,000 times faster than efficient MD simulations. Hierarchical Graph Neural Networks with Spatial Information (HS-GNNs) are introduced, which integrate chemistry knowledge. HS-GNNs as well as extreme gradient boosted trees (XGBoost) achieve forecasts of mechanical properties of arbitrary carbon nanostructures with only 3 to 6% mean relative error. The reliability equals experimental accuracy and is up to 20 times higher than other ML methods. Predictions maintain 8 to 18% accuracy for large CNT bundles, CNT junctions, and carbon fiber cross-sections outside the training distribution. The physics- and chemistry-informed HS-GNN works remarkably well for data outside the training range while XGBoost works well with limited training data inside the training range. The carbon nanostructure database is designed for integration with multimodal experimental and simulation data, scalable beyond 100 nm size, and extendable to chemically similar compounds and broader property ranges. The ML approaches have potential for applications in structural materials, nanoelectronics, and carbon-based catalysts.
Winetrout, Jordan J.↗
NASA Tech Briefs, August 2005
Topics include: Hidden Identification on Parts: Magnetic Machine-Readable Matrix Symbols; System for Processing Coded OFDM Under Doppler and Fading; Multipurpose Hyperspectral Imaging System; Magnetic-Flux-Compensated Voltage Divider; High-Performance Satellite/Terrestrial-Network Gateway; Internet-Based System for Voice Communication With the ISS; Stripline/Microstrip Transition in Multilayer Circuit Board; Dual-Band Feed for a Microwave Reflector Antenna; Quadratic Programming for Allocating Control Effort; Range Process Simulation Tool; Simulator of Space Communication Networks; Computing Q-D Relationships for Storage of Rocket Fuels; Contour Error Map Algorithm; Portfolio Analysis Tool; Glass Frit Filters for Collecting Metal Oxide Nanoparticles; Anhydrous Proton-Conducting Membranes for Fuel Cells; Portable Electron-Beam Free-Form Fabrication System; Miniature Laboratory for Detecting Sparse Biomolecules; Multicompartment Liquid-Cooling/Warming Protective Garments; Laser Metrology for an Optical-Path-Length Modulator; PCM Passive Cooling System Containing Active Subsystems; Automated Electrostatics Environmental Chamber; Estimating Aeroheating of a 3D Body Using a 2D Flow Solver; Artificial Immune System for Recognizing Patterns; Computing the Thermodynamic State of a Cryogenic Fluid; Safety and Mission Assurance Performance Metric; Magnetic Control of Concentration Gradient in Microgravity; Avionics for a Small Robotic Inspection Spacecraft; and Simulation of Dynamics of a Flexible Miniature Airplane.
Source record↗
A framework for stochastic simulations and visualization of biological electron-transfer dynamics
Electron transfer (ET) dictates a wide variety of energy-conversion processes in biological systems. Visualizing ET dynamics could provide key insight into understanding and possibly controlling these processes. We present a computational framework named VizBET to visualize biological ET dynamics, using an outer-membrane Mtr–Omc cytochrome complex in Shewanella oneidensis MR-1 as an example. Starting from X-ray crystal structures of the constituent cytochromes, molecular dynamics simulations are combined with homology modeling, protein docking, and binding free energy computations to sample the configuration of the complex as well as the change of the free energy associated with ET. This information, along with quantum-mechanical calculations of the electronic coupling, provides inputs to kinetic Monte Carlo (KMC) simulations of ET dynamics in a network of heme groups within the complex. Visualization of the KMC simulation results has been implemented as a plugin to the Visual Molecular Dynamics (VMD) software. Finally, VizBET has been used to reveal the nature of ET dynamics associated with novel nonequilibrium phase transitions in a candidate configuration of the Mtr–Omc complex due to electron–electron interactions.
74 ATOMIC AND MOLECULAR PHYSICS↗
The Corrected Distortion model for Lagrangian spray simulation of transcritical fuel injection
In this work, we present a detailed implementation and validation of the droplet modeling framework proposed by Dahms and Oefelein (2016) into the engine commercial CFD software CONVERGE using the User Defined Function (UDF) interface. The model accounts for the nonlinear deformation and oscillation experienced by liquid spray droplet injected into high pressure and temperature. Lagrangian spray simulations of Engine Combustion Network (ECN) Spray A are performed. Model validation against standard experimental measurements of liquid velocity, vapor mixture fraction is conducted. To perform more rigorous model validation, new experimental measurements based on Diffused Back Illumination (DBI) are introduced. The new measurements are processed for Projected Liquid Volume (PLV), which offers as close as possible one-to-one model validation for liquid penetration while offering new insights into the spray physics. Comparison with a One-D model based on adiabatic mixing theory by Siebers (1999) and Desantes et al. (2007) are also conducted. Through these model validation exercises, it is shown that the new framework improves liquid-phase penetration predictions, following a tendency for enhanced evaporation, compared to the standard approach for both Reynolds Average Navier Stokes (RANS) and Large Eddy Simulation (LES). Additionally, at the liquid length, maximum mixture fraction values predicted by the new approach are in good agreement those of an adiabatic mixing model. Qualitative analysis of the spray behaviors during the early stage of the injection process reveals that the proposed framework predicts significant increase in droplet evaporation rate with lower drop drag compared to the current standard approach.
42 ENGINEERING↗
Theoretical and Computational Modeling of Magnetically Ordered Molecules & Electronic Nano-Transport of Spins: State of Art and Unanswered Questions : Workshop
As a culmination of a five-year Nordforsk Network project entitled “Nanospintronics: Theory and Simulations”, Professor Carlo Canali (Linneaus University, Sweden) and members of the network, and Vincenza Benza (Local Organizer Milan) organized a summer workshop in conjunction with the A. Volta Scientific Cultural Exchange program. This workshop took place 24-30 August 2013. Several Basic Energy Scientists from the US conducted lectures and also provided hands-on tutorials to developing materials and chemicals scientists. We have received a total of $10,000 of support to offset the travel expenses of US-based participants for this workshop.
36 MATERIALS SCIENCE↗
Comparison between analytical models and finite-difference simulations in transmission-line tap resistors and L-type cross-Kelvin resistors
Approximate analytical models of the transmission-line tap resistor and the cross-Kelvin resistor are compared with computer-simulated pseudo-three-dimensional resistor network models. The analytical formulas are in good agreement with the simulations over a useful range of parameters and are readily applied to the extraction of the contact resistivity and the sheet resistances of the semiconducting layer under and outside the contacts. The extraction procedure, which is easily implemented on a personal computer, is carried out for the case of alloyed AuGeNi/GaAs contacts, illustrating the importance of (1) distinguishing between layer sheet resistance under and outside the contacts and (2) considering two-dimensional current flow in the semiconducting layer.
Scorzoni, Andrea↗
A Multi-Branch Decoder Network Approach to Adaptive Temporal Data Selection and Reconstruction for Big Scientific Simulation Data
A key challenge in scientific simulation is that the simulation outputs often require intensive I/O and storage space to store the results for effective post hoc analysis. This article focuses on a quality-aware adaptive temporal data selection and reconstruction problem where the goal is to adaptively select simulation data samples at certain key timesteps in situ and reconstruct the discarded samples with quality assurance during post hoc analysis. This problem is motivated by the limitation of current solutions that a significant amount of simulation data samples are either discarded or aggregated during the sampling process, leading to inaccurate modeling of the simulated phenomena. Two unique challenges exist: 1) the sampling decisions have to be made in situ and adapted to the dynamics of the complex scientific simulation data; 2) the reconstruction error must be strictly bounded to meet the application requirement. To address the above challenges, we develop DeepSample , an error-controlled convolutional neural network framework, that jointly integrates a set of coherent multi-branch deep decoders to effectively reconstruct the simulation data with rigorous quality assurance. The results on two real-world scientific simulation applications show that DeepSample significantly outperforms other state-of-the-art methods on both sampling efficiency and reconstructed simulation data quality.
Zhang, Yang↗