Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “network acceleration”

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 37 records · Page 2
All sourcesNASA NTRSDOE OSTI
DOE OSTI2025

A Grassroots Network and Community Roadmap for Interconnected Autonomous Science Laboratories for Accelerated Discovery

Scientific discovery is being revolutionized by AI and autonomous systems, yet current autonomous laboratories remain isolated islands unable to collaborate across institutions. We present the Autonomous Interconnected Science Lab Ecosystem (AISLE), a grassroots network transforming fragmented capabilities into a unified system that shorten the path from ideation to innovation to impact and accelerates discovery from decades to months. AISLE addresses five critical dimensions: (1) cross-institutional equipment orchestration, (2) intelligent data management with FAIR compliance, (3) AI-agent driven orchestration grounded in scientific principles, (4) interoperable agent communication interfaces, and (5) AI/ML-integrated scientific education. By connecting autonomous agents across institutional boundaries, autonomous science can unlock research spaces inaccessible to traditional approaches while democratizing cutting-edge technologies. This paradigm shift toward collaborative autonomous science promises breakthroughs in sustainable energy, materials development, and public health.

Ferreira da Silva, Rafael [Oak Ridge National Labo↗
DOE OSTI2024

Approximating accelerator impedances with resonator networks

It is common in the accelerator community to use the impedance of accelerator components to describe wake interactions in the frequency domain. However, it is often desirable to understand such wake interactions in the time domain in a general manner for excitations that are not necessarily Gaussian in nature. Here, the conventional method for doing this involves taking the inverse Fourier Transform of the component impedance, obtaining the Green's Function, and then convolving it with the desired excitation distribution. This method can prove numerically cumbersome, for a convolution integral must be evaluated for each individual point in time when the wake function is desired. An alternative to this method would be to compute the wake function analytically, which would sidestep the need for repetitive integration. Only a handful of cases, however, are simple enough for this method to be tenable. One of these cases is the case where the component in question is an RLC resonator, which has a closed-form analytical wake function solution. This means that a component which can be represented in terms of resonators can leverage this solution. As it happens, common network synthesis techniques may be used to map arbitrary impedance profiles to RLC resonator networks in a manner the accelerator community has yet to take advantage of. In this work, we will use Foster Canonical Resonator Networks and partial derivative descent optimization to develop a technique for synthesizing resonator networks that well approximate the impedances of real-world accelerator components. We will link this synthesis to the closed-form resonator wake function solution, giving rise to a powerful workflow that may be used to streamline beam dynamics simulations.

43 PARTICLE ACCELERATORS↗
NASA NTRS1998

Feasibility of Using Neural Network Models to Accelerate the Testing of Mechanical Systems

Verification testing is an important aspect of the design process for mechanical mechanisms, and full-scale, full-length life testing is typically used to qualify any new component for use in space. However, as the required life specification is increased, full-length life tests become more costly and lengthen the development time. At the NASA Lewis Research Center, we theorized that neural network systems may be able to model the operation of a mechanical device. If so, the resulting neural network models could simulate long-term mechanical testing with data from a short-term test. This combination of computer modeling and short-term mechanical testing could then be used to verify the reliability of mechanical systems, thereby eliminating the costs associated with long-term testing. Neural network models could also enable designers to predict the performance of mechanisms at the conceptual design stage by entering the critical parameters as input and running the model to predict performance. The purpose of this study was to assess the potential of using neural networks to predict the performance and life of mechanical systems. To do this, we generated a neural network system to model wear obtained from three accelerated testing devices: 1) A pin-on-disk tribometer; 2) A line-contact rub-shoe tribometer; 3) A four-ball tribometer.

Fusaro, Robert L.↗
DOE OSTI2023

cuAlign: Scalable Network Alignment on GPU Accelerators

Given two graphs, the objective of network alignment is to find the best one-to-one mapping of vertices in one graph (??) to vertices in the other (??), such that the number of overlaps is maximized. We say that edges(??, ??) ???and(??', ??') ??? are overlapped if ?? is mapped to ??' and ?? is mapped to??'. Network alignment is an important optimization problem with several applications in bioinformatics, computer vision and ontology matching. Since it is an NP-hard problem, efficient heuristics and scalable implementations are necessary. In this work, we introduce a new framework that combines the concepts of intra-network proximity using vertex embedding,Belief Propagation (BP) and approximate weighted matching, and provides qualitative improvements up to22%over state-of-the-art approaches. We also provide scalable implementations on GPU accelerators, demonstrating up to19×speedup for Belief Propagation and 3× speedup for approximate weighted matching relative to previous multithreaded implementation. A combination of combinatorial and algebraic kernels within the network alignment algorithm poses significant hurdles for parallelization. Load imbalance and irregular DRAM traffic limit achievable performance on GPUs. Our novel approach identifies and exploits unique structural proper-ties of the BP-based algorithm and employs code fusion to reduce data movement between different steps of the algorithm. Using a diverse set of inputs, we demonstrate qualitative improvements of our algorithms, and performance gains of our GPU-accelerated implementation. We believe that our work will enable algorithmic improvements and practical applications of network alignment.

Xiang, Lizhi↗
DOE OSTI2025

HEPOM: Using Graph Neural Networks for the Accelerated Predictions of Hydrolysis Free Energies in Different pH Conditions

Hydrolysis is a fundamental family of chemical reactions where water facilitates the cleavage of bonds. The process is ubiquitous in biological and chemical systems, owing to water’s remarkable versatility as a solvent. However, accurately predicting the feasibility of hydrolysis through computational techniques is a difficult task, as subtle changes in reactant structure like heteroatom substitutions or neighboring functional groups can influence the reaction outcome. Furthermore, hydrolysis is sensitive to the pH of the aqueous medium, and the same reaction can have different reaction properties at different pH conditions. In this work, we have combined reaction templates and high-throughput ab initio calculations to construct a diverse data set of hydrolysis free energies. The developed framework automatically identifies reaction centers, generates hydrolysis products, and utilizes a trained graph neural network (GNN) model to predict ΔG values for all potential hydrolysis reactions in a given molecule. The long-term goal of the work is to develop a data-driven, computational tool for high-throughput screening of pH-specific hydrolytic stability and the rapid prediction of reaction products, which can then be applied in a wide array of applications including chemical recycling of polymers and ion-conducting membranes for clean energy generation and storage.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2024

Acceleration of Graph Neural Network-Based Prediction Models in Chemistry via Co-Design Optimization on Intelligence Processing Units

Atomic structure prediction and associated property calculations are the bedrock of chemical physics. Since high-fidelity ab initio modeling techniques for computing the structure and properties can be prohibitively expensive, this motivates the development of machine-learning (ML) models that make these predictions more efficiently. Training graph neural networks over large atomistic databases introduces unique computational challenges such as the need to process millions of small graphs with variable size and support communication patterns that are distinct from learning over large graphs such as social networks. We demonstrate a novel hardware-software co-design approach to scale up the training of atomistic graph neural networks (GNN) for structure and property prediction. First, to eliminate redundant computation and memory associated with alternative padding techniques and to improve throughput via minimizing communication, we formulate the effective coalescing of the batches of variable-size atomistic graphs as the bin packing problem and introduce a hardware-agnostic algorithm to pack these batches. In addition, we propose hardware-specific optimizations including a planner and vectorization for the gather-scatter operations targeted for Graphcore’s Intelligence Processing Unit (IPU), as well as model-specific optimizations such as merged communication collectives and optimized softplus. Putting these all together, we demonstrate the effectiveness of the proposed co-design approach by providing an implementation of a well-established atomistic GNN on the Graphcore IPUs. We evaluate the training performance on multiple atomistic graph databases with varying degrees of graph counts, sizes and sparsity. Here, we demonstrate that such a co-design approach can reduce the training time of atomistic GNNs and can improve the performance by up to 1.5× compared to the baseline implementation of the model on the IPUs. Additionally, we compare our IPU implementation with a Nvidia GPU-based implementation and show that our atomistic GNN implementation on the IPUs can run 1.8× faster on average compared to the execution time on the GPUs.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
NASA NTRS1993

An accelerated training method for back propagation networks

The principal objective is to provide a training procedure for a feed forward, back propagation neural network which greatly accelerates the training process. A set of orthogonal singular vectors are determined from the input matrix such that the standard deviations of the projections of the input vectors along these singular vectors, as a set, are substantially maximized, thus providing an optimal means of presenting the input data. Novelty exists in the method of extracting from the set of input data, a set of features which can serve to represent the input data in a simplified manner, thus greatly reducing the time/expense to training the system.

Shelton, Robert O.↗
DOE OSTI2025

ChemComp: A Compilation Framework for Computing with Chemical Reaction Networks

The acceleration of scientific computation, data analytics, and artificial intelligence is driving a surge in computational requirements. Yet, state-of-the-art high-performance computing systems are approaching physical limitations that impede further significant improvements in energy efficiency. As we move towards post-exascale computing systems, innovative approaches are necessary to overcome this barrier in power consumption. Novel analog and hybrid digital-analog architectures hold promise for enhancing energy efficiency by several orders of magnitude. Biochemical computation stands out among the various solutions being explored due to its potential to enable new classes of devices with immense computational capabilities. These devices can capitalize on the inherent efficacy of biological cells in solving optimization problems and are scalable through increasing reaction system size or vessel capacity, potentially satisfying scientific computing's high-performance requirements. Nonetheless, several theoretical and practical limitations persist, including problem formulation and mapping to chemical reaction networks (CRNs) and implementation of actual CRN devices. In this paper, we propose a framework for biochemical computation using systems chemistry. We present the initial components of our approach: an abstract chemical reaction dialect implemented as a multi-level intermediate representation (MLIR) compiler extension and a pathway to represent mathematical problems with CRNs. To showcase the potential of this approach, we emulate a simplified chemical reservoir device. This work lays the groundwork for leveraging chemistry's computing potential in creating energy-efficient, high-performance computing systems tailored to contemporary computational needs.

artificial intelligence↗
NASA NTRSDate not supplied

Launching Liquid Slugs Through Tube Networks: Investigating Multiphase Liquid Slug Acceleration Behaviors in Spacecraft Propulsion Systems using 3D Multiphase VOF CFD Simulations and High-Speed Flow-Visualization Validation Experiments

Propellant management in spacecraft propulsion systems is complex. In many current and future crewed and robotic spacecraft propulsion systems, a number of unintended or unavoidable scenarios can occasionally lead to liquid “slugs” of propellant accumulating in unwanted locations and thereafter being rapidly accelerated through complex and arbitrary tubing networks otherwise filled with vapor or gas. To develop better understanding of the multiphase fluid dynamics associated with these “slug launch” scenarios, NASA Marshal Space Flight Center’s (MSFC) ER42 branch – the propulsion fluid dynamics research branch – conducted 3D multiphase transient Volume of Fluid (VOF) CFD simulations using an in-house tool called LOCI-Stream VOF on semi-canonical tube networks with straight sections, bends, and bend complexes. In addition, high-speed flow visualization validation experiments using propellants were conducted on similar tube network geometries at the NASA White Sands Test Facility (WSTF). These simulations and flow-visualization experiments found these scenarios lead to a variety of unique multiphase fluid dynamic effects involving the leading and trailing edges of the slug in both straight sections and curved bends as well as the progressive erosion, acceleration, and eventual “punch-through” of the driving gas through the liquid slug. Understanding these multiphase fluid dynamic effects has led to unique and often counter-intuitive observations which have been of critical relevance to important engineering concerns associated with these “slug launch” scenarios.

Multiphase Fluid Dynamics↗
DOE OSTI2025

TorchBraid: High-Performance Layer-Parallel Training of Deep Neural Networks with MPI and GPU Acceleration

TorchBraid is a high-performance implementation of layer-parallel training for deep neural networks (DNNs) supporting MPI-based parallelism and GPU acceleration. Layer-parallel training has been developed to overcome the serialization inherent in forward and backward propagation of DNNs that limits utilization of computational resources in the strong scaling limit. To achieve this, TorchBraid integrates the PyTorch neural network framework with the state-of-the-art XBraid time-parallel library. Furthermore, this article presents the use and performance of TorchBraid, in addition to solutions for overcoming the algorithmic challenges inherent in combining automatic differentiation with layer-parallel. Results are presented with and without GPU acceleration for the Tiny ImageNet and MNIST image classification data sets, as well as recurrent neural networks. Overall, TorchBraid enables fast training of DNNs, both in a strong and weak scaling context. In addition to the TorchBraid software, several new advances in applying layer-parallel algorithms are detailed. Integration of layer-parallel with data-parallel algorithms is presented for the first time, showing the computational advantages of the combination. Standard deep learning techniques, like batch-normalization, are developed for layer-parallel training. Finally, a new approach combining layer-parallel with spatial coarsening in order to accelerate training for 3D image classification shows roughly a 10× speedup over serial execution.

Layer-parallel↗
DOE OSTI2023

Phase Space Reconstruction from Accelerator Beam Measurements Using Neural Networks and Differentiable Simulations

Characterizing the phase space distribution of particle beams in accelerators is a central part of accelerator understanding and performance optimization. However, conventional reconstruction-based techniques either use simplifying assumptions or require specialized diagnostics to infer high-dimensional (> $2D$) beam properties. In this Letter, we introduce a general-purpose algorithm that combines neural networks with differentiable particle tracking to efficiently reconstruct high-dimensional phase space distributions without using specialized beam diagnostics or beam manipulations. Furthermore, we demonstrate that our algorithm accurately reconstructs detailed 4D phase space distributions with corresponding confidence intervals in both simulation and experiment using a single focusing quadrupole and diagnostic screen. This technique allows for the measurement of multiple correlated phase spaces simultaneously, which will enable simplified 6D phase space distribution reconstructions in the future.

47 OTHER INSTRUMENTATION↗
DOE OSTI2023

Laser Wakefield Accelerator modelling with Variational Neural Networks

A machine learning model was created to predict the electron spectrum generated by a GeV-class laser wakefield accelerator. The model was constructed from variational convolutional neural networks, which mapped the results of secondary laser and plasma diagnostics to the generated electron spectrum. An ensemble of trained networks was used to predict the electron spectrum and to provide an estimation of the uncertainty of that prediction. It is anticipated that this approach will be useful for inferring the electron spectrum prior to undergoing any process that can alter or destroy the beam. In addition, the model provides insight into the scaling of electron beam properties due to stochastic fluctuations in the laser energy and plasma electron density.

43 PARTICLE ACCELERATORS↗
NASA NTRS1992

Accelerating Learning By Neural Networks

Electronic neural networks made to learn faster by use of terminal teacher forcing. Method of supervised learning involves addition of teacher forcing functions to excitations fed as inputs to output neurons. Initially, teacher forcing functions are strong enough to force outputs to desired values; subsequently, these functions decay with time. When learning successfully completed, terminal teacher forcing vanishes, and dynamics or neural network become equivalent to those of conventional neural network. Simulated neural network with terminal teacher forcing learned to produce close approximation of circular trajectory in 400 iterations.

Toomarian, Nikzad↗
NASA NTRS2011

IONAC-Lite

The Interplanetary Overlay Net - working Protocol Accelerator (IONAC) described previously in The Inter - planetary Overlay Networking Protocol Accelerator (NPO-45584), NASA Tech Briefs, Vol. 32, No. 10, (October 2008) p. 106 (http://www.techbriefs.com/component/ content/article/3317) provides functions that implement the Delay Tolerant Networking (DTN) bundle protocol. New missions that require high-speed downlink-only use of DTN can now be accommodated by the unidirectional IONAC-Lite to support high data rate downlink mission applications. Due to constrained energy resources, a conventional software implementation of the DTN protocol can provide only limited throughput for any given reasonable energy consumption rate. The IONAC-Lite DTN Protocol Accelerator is able to reduce this energy consumption by an order of magnitude and increase the throughput capability by two orders of magnitude. In addition, a conventional DTN implementation requires a bundle database with a considerable storage requirement. In very high downlink datarate missions such as near-Earth radar science missions, the storage space utilization needs to be maximized for science data and minimized for communications protocol-related storage needs. The IONAC-Lite DTN Protocol Accelerator is implemented in a reconfigurable hardware device to accomplish exactly what s needed for high-throughput DTN downlink-only scenarios. The following are salient features of the IONAC-Lite implementation: An implementation of the Bundle Protocol for an environment that requires a very high rate bundle egress data rate. The C&DH (command and data handling) subsystem is also expected to be very constrained so the interaction with the C&DH processor and the temporary storage are minimized. Fully pipelined design so that bundle processing database is not required. Implements a lookup table-based approach to eliminate multi-pass processing requirement imposed by the Bundle Protocol header s length field structure and the SDNV (self-delimiting numeric value) data field formatting. 8-bit parallel datapath to support high data-rate missions. Reduced resource utilization implementation for missions that do not require custody transfer features. There was no known implementation of the DTN protocol in a field programmable gate array (FPGA) device prior to the current implementation. The combination of energy and performance optimization that embodies this design makes the work novel.

Torgerson, Jordan L.↗
DOE OSTI2022

Neuromorphic Accelerator for Deep Spiking Neural Networks with NVM Crossbar Arrays

In this paper, we present a scalable digital hardware accelerator based on non-volatile memory arrays capable of realizing deep convolutional spiking neural networks (SNNs). Our design studies are conducted using a compact model for spin-transfer torque random access memory (STT-RAM) devices. Large networks are realized by tiling multiple cores which communicate by transmitting spike packets via an on-chip routing network. Compared to an equivalent SRAM based core design, we show that the STT-RAM based design achieves nearly 15X higher GSOPS (Synaptic Operations per Second) per Watt per mm 2 making it a promising platform for realizing systems with significant area and power limitations.

Kulkarni, Shruti↗
DOE OSTI2023

Control Systems Design for STS Accelerator

The Second Target Station (STS) Project will expand the capabilities of the existing Spallation Neutron Source (SNS), with a suite of neutron instruments optimized for long wavelengths. A new accelerator transport line will be built to deliver one out of four SNS pulses to the new target station. The Integrated Control Systems (ICS) will provide remote control, monitoring, OPI, alarms, and archivers for the accelerator systems, such as magnets power supply, vacuum devices, and beam instrumentation. The ICS will upgrade the existing Linac LLRF controls to allow independent operation of the FTS and STS and support different power levels of the FTS and STS proton beam. The ICS accelerator controls are in the phase of preliminary design for the control systems of magnet power supply, vacuum, LLRF, Timing, Machine protection system (MPS), and computing and machine network. The accelerator control systems build upon the existing SNS Machine Control systems, use the SNS standard hardware and EPICS software, and take full advantage of the performance gains delivered by the PPU Project at SNS.

Yan, Jay↗
DOE OSTI2023

Differentiable Neural Architecture, Mixed Precision and Accelerator Co-Search

Quantization, effective Neural Network architecture, and efficient accelerator hardware are three important design paradigms to maximize accuracy and efficiency. Mixed Precision Quantization is a process of assigning different precision to different Neural Network layers for optimized inference. Neural Architecture Search (NAS) is a process of automatically designing the neural network for a task and can also be extended to search for the precision of each weight and activation matrix. In this paper, we develop the following three methods: (i) Fast Differentiable Hardware-aware Mixed Precision Quantization Search method to find optimal precision, (ii) Joint Differentiable hardware-aware Architecture and Mixed Precision Quantization Co-search, (iii) Joint Accelerator, Architecture, and Precision triple co-search to find best possibilities in all the three worlds. We demonstrate the effectiveness of our proposed methods targeting Bitfusion accelerator by searching mixed precision models on MobilenetV2. We achieve better accuracy-latency trade-off models than the manually designed and previously proposed search methods.

97 MATHEMATICS AND COMPUTING↗