Search NASA⌕ Search

SEARCH · Search NASA

Results for “distributed parameter system”

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 145 records · Page 8
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Background Information on the Protection Requirements in IEEE Std 1547-2018

As the DER penetration level has risen significantly in recent times, parameter settings and configuration of installed DERs are having direct impact on local electric distribution utilities as well as bulk power systems during normal and abnormal grid conditions. On the other hand, all grid-connected DERs in the United States of America must conform to the interconnection requirements prescribed in the IEEE Std 1547-2018. Because of these reasons, it is crucial to understand the expected DER performance during abnormal grid conditions considering the requirements of the IEEE Std 1547-2018. The aim of this document is twofold: (1) to summarize requirements in IEEE Std 1547-2018 having direct implications in distribution network protection, and (2) to provide the reader with knowledge and information that will be required for users to apply the requirements specified. Since the focus of this document is on distribution network protection in the presence of DERs, it is assumed that the readers have basic understanding of distribution network protection as well as working principles of DERs. This document starts with a brief introduction on protection systems for distribution network followed by a discussion on the impact of DER on the protection systems. Current practices to provide network protection in the presence of DER is briefly discussed. The document then explores the DER performance requirements in IEEE Std. 1547-2018; especially on those related to distribution network protection. Section 6.1 in IEEE Std. 1547-2018 provides an overview of capabilities and control requirements for DER under abnormal operating conditions. This section also introduces abnormal operating performance categories I, II, III. The DER response to various types of faults and grid conditions such as short-circuit faults and open phase conditions are discussed in section 6.2 in the standard while requirements for coordination with the Area EPS reclosing scheme are provided in section 6.3. Section 6.4 in the IEEE Std 1547-2018 specifies requirements for mandatory voltage tripping and ride-through requirements during low and high voltage disturbances and section 6.5 specifies similar requirements for low and high frequency disturbances. IEEE Std. 1547 requires that the conformance of the DERs to IEEE Std 1547-2018 requirements should be verified in accordance with IEEE Std. 1547. 1. There are several parameters and settings of DERs that need to be properly selected for reliable operation during abnormal grid condition while applying the requirements of IEEE Std 1547-2018. Key decisions for proper selections of parameter and settings are: 1. Determination of required DER abnormal operating performance category, 2. Determination of DER response (shall trip) to abnormal voltages and 3. Determination of DER response (shall trip) to abnormal frequency.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2024

A Diffusion‐Based Uncertainty Quantification Method to Advance E3SM Land Model Calibration

Abstract Calibrating land surface models and accurately quantifying their uncertainty are crucial for improving the reliability of simulations of complex environmental processes. This, in turn, advances our predictive understanding of ecosystems and supports climate‐resilient decision‐making. Traditional calibration methods, however, face challenges of high computational costs and difficulties in accurately quantifying parameter uncertainties. To address these issues, we develop a diffusion‐based uncertainty quantification (DBUQ) method. Unlike conventional generative diffusion methods, which are computationally expensive and memory‐intensive, DBUQ innovates by formulating a parameterized generative model and approximates this model through supervised learning, which enables quick generation of parameter posterior samples to quantify its uncertainty. DBUQ is effective, efficient, and general‐purpose, making it suitable for site‐specific ecosystem model calibration and broadly applicable for parameter uncertainty quantification across various earth system models. In this study, we applied DBUQ to calibrate the Energy Exascale Earth System Model land model at the Missouri Ozark AmeriFlux forest site. Results indicated that DBUQ produced accurate parameter posterior distributions similar to those from Markov Chain Monte Carlo sampling but with 30 times less computing time. This significant improvement in efficiency suggests that DBUQ can enable rapid, site‐level model calibration at a global scale, enhancing our predictive understanding of climate impacts on terrestrial ecosystems.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2024

ARM Aerial Facility (AAF) - Unmanned Aircraft Systems, Cloud Droplet Probe with QC and lat/lon/alt

The Cloud Droplet Probe (CDP) is designed to measure cloud droplet size distribution from 2 µm to 50 µm. The CDP and an appropriate data system can also calculate various other parameters including particle concentrations, effective diameter (ED), Median Volume Diameter (MVD), and Liquid Water Content (LWC). The b1 level adds standard quality control flags, and merges in flight navigation data.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2022

Implementation of a practical Markov chain Monte Carlo sampling algorithm in PyBioNetFit

Abstract Summary Bayesian inference in biological modeling commonly relies on Markov chain Monte Carlo (MCMC) sampling of a multidimensional and non-Gaussian posterior distribution that is not analytically tractable. Here, we present the implementation of a practical MCMC method in the open-source software package PyBioNetFit (PyBNF), which is designed to support parameterization of mathematical models for biological systems. The new MCMC method, am, incorporates an adaptive move proposal distribution. For warm starts, sampling can be initiated at a specified location in parameter space and with a multivariate Gaussian proposal distribution defined initially by a specified covariance matrix. Multiple chains can be generated in parallel using a computer cluster. We demonstrate that am can be used to successfully solve real-world Bayesian inference problems, including forecasting of new Coronavirus Disease 2019 case detection with Bayesian quantification of forecast uncertainty. Availability and implementation PyBNF version 1.1.9, the first stable release with am, is available at PyPI and can be installed using the pip package-management system on platforms that have a working installation of Python 3. PyBNF relies on libRoadRunner and BioNetGen for simulations (e.g. numerical integration of ordinary differential equations defined in SBML or BNGL files) and Dask.Distributed for task scheduling on Linux computer clusters. The Python source code can be freely downloaded/cloned from GitHub and used and modified under terms of the BSD-3 license (https://github.com/lanl/pybnf). Online documentation covering installation/usage is available (https://pybnf.readthedocs.io/en/latest/). A tutorial video is available on YouTube (https://www.youtube.com/watch?v=2aRqpqFOiS4&t=63s). Supplementary information Supplementary data are available at Bioinformatics online.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2025

Seasonal Reconfiguration of Electrical Distribution Systems to Mitigate the Impact of Electric Vehicle Charging

Power grids face challenges in their infrastructure related to the integration of electric vehicles (EV). In particular, EV charging stations may induce instability in key system parameters such as substantial voltage drops, active power losses, and transformer overload due to high demand during charging periods. This article presents a seasonal reconfiguration strategy based on the differential evolution (DE) algorithm, aimed at enhancing system performance under highly variable and stochastic load profiles, particularly those driven by EV charging. The DEA algorithm is hybridized with the find-union (FU) algorithm to efficiently ensure network radiality throughout the optimization process. The proposed methodology is validated on a hybrid distribution system composed of the IEEE 33-bus network, a modified IEEE 13-bus system, and a specific 13-bus microgrid. Results have demonstrated that seasonal reconfiguration significantly reduces active power losses and mitigates transformer loading during critical demand hours, thereby quantifiably increasing the system’s performance. As an integral component of the proposed approach, an analysis of CO2 emissions associated with energy losses is included, allowing a contextualized assessment of the environmental benefits of seasonal reconfiguration in various geographical areas.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2025

Solving disorder in (3D) real space: a comparative study of the three-dimensional difference pair distribution function and atomic resolution holography reconstructions

The quantitative analysis of local ordering principles in disordered crystalline systems has gained much attention over the past few years, as it is often considered crucial for optimizing material functionality. This development has been driven by significant advancements in computational and experimental methods, which have led to the establishment and widespread use of various analytical techniques. In this study, we perform model calculations to compare the effectiveness of atomic resolution holography and three-dimensional difference pair distribution function analysis (3D-ΔPDF). Using Cu 3 Au as a model system, we demonstrate an approach to derive local order parameters quantitatively and show that both techniques are well suited to quantifying chemical short-range order correlations and local bond-distance variations. By evaluating the strengths and limitations of both techniques, we advocate for their combined use to solve complex short-range order problems accurately.

3D-ΔPDF↗
DOE OSTI2025

Versatile system for ion energy measurements generated by pulsed laser ionization: Insights into electron-ion dynamics

Ion energy distributions generated by pulsed laser interactions with materials are essential for applications ranging from materials science to oncology. Ion energy characterization is particularly important for an emerging mass spectrometry technique called virtual-slit cycloidal mass spectrometry (VS-CMS). The ion energy distribution influences the design and performance of VS-CMS instruments, as well as the efficacy of laser-driven ionization methods in various fields. Several established techniques, including the retarding potential method, time-of-flight (TOF) analysis, and electrostatic energy analyzers, have been employed to measure ion energy distributions. The wide range of ion energies reported highlights the strong dependence of ion energy on laser parameters, target materials, and experimental conditions, as well as the necessity of making independent measurements of the ion energy distribution for specific laser systems and materials. This paper presents the design and characterization of a simple TOF-based apparatus for measuring ion energy distributions from pulsed laser ionization without external fields. This approach minimizes perturbation of electron-ion dynamics and enables simultaneous energy measurements at multiple spatial positions. Here, the apparatus was tested using a nanosecond pulsed Nd:YAG laser operating at 1064 nm, 532 nm, and 266 nm on solid copper sheets at various laser fluences. Simultaneous measurements at different distances provide new insights into ion-electron interactions post-ionization and demonstrate the influence of laser wavelength and fluence on ion energy distributions.

Ion energy↗
DOE OSTI2025

A kinetic-based regularization method for data science applications

We propose a physics-based regularization technique for function learning, inspired by statistical mechanics. By drawing an analogy between optimizing the parameters of an interpolator and minimizing the energy of a system, we introduce corrections that impose constraints on the lower-order moments of the data distribution. This minimizes the discrepancy between the discrete and continuum representations of the data, in turn allowing to access more favorable energy landscapes, thus improving the accuracy of the interpolator. Our approach improves performance in both interpolation and regression tasks, even in high-dimensional spaces. Unlike traditional methods, it does not require empirical parameter tuning, making it particularly effective for handling noisy data. We also show that thanks to its local nature, the method offers computational and memory efficiency advantages over Radial Basis Function interpolators, especially for large datasets.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

A Distributed Trust Model Simulator for Energy Grid of Things Distributed Energy Resource Management System

The evolution of networks into more distributed, self-reliant nodes has mitigated single-point failures that plagued traditional centralized networks. Applied to power grids, distributed systems can increase the integrity and availability of grid services while also offering a power management solution. However, while distributed networks provide scalability, security, and sustainability compared to centralized networks, their distributed nature makes them harder for anomaly detection and prevention. Incorporating a Distributed Trust Model (DTM) System into an Energy Grid of Things Distributed Energy Resource Management System (EGOT DERMS) allows grid participants to be characterized and their communication to be analyzed for possible attacks. A Trust Model simulator is needed to evaluate and improve the DTM System.Trustworthiness is calculated using a Trust Model. While many trust models exist, most only consider 2-3 matrices to evaluate trust. The TM proposed in this thesis uses a Metric Vector of Trust (MVoT) monitoring 17 parameters when assessing trust. Moreover, unlike standard trust models, the proposed trust model establishes a method to test the trust between various actors within the network and probe the trust model itself. Using a Trust Model Simulator, MVoT calaculations, initial values, and parameters are fine-tuned to achieve high-confidence message classifications and minimize false positives. The DTM System and Trust Mode Simulation Suite allow for distributed trust evaluation with a real-time classification of EGOT DERMS actors, providing additional security for distributed systems.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2024

Trust-Based Detection and Mitigation of Cyber Attacks in Distributed Cooperative Control of Islanded AC Microgrids

In this study, we address the challenge of detecting and mitigating cyber attacks in the distributed cooperative control of islanded AC microgrids, with a particular focus on detecting False Data Injection Attacks (FDIAs), a significant threat to the Smart Grid (SG). The SG integrates traditional power systems with communication networks, creating a complex system with numerous vulnerable links, making it a prime target for cyber attacks. These attacks can lead to the disclosure of private data, control network failures, and even blackouts. Unlike machine learning-based approaches that require extensive datasets and mathematical models dependent on accurate system modeling, our method is free from such dependencies. To enhance the microgrid’s resilience against these threats, we propose a resilient control algorithm by introducing a novel trustworthiness parameter into the traditional cooperative control algorithm. Our method evaluates the trustworthiness of distributed energy resources (DERs) based on their voltage measurements and exchanged information, using Kullback-Leibler (KL) divergence to dynamically adjust control actions. We validated our approach through simulations on both the IEEE-34 bus feeder system with eight DERs and a larger microgrid with twenty-two DERs. The results demonstrated a detection accuracy of around 100%, with millisecond range mitigation time, ensuring rapid system recovery. Additionally, our method improved system stability by up to almost 100% under attack scenarios, showcasing its effectiveness in promptly detecting attacks and maintaining system resilience. These findings highlight the potential of our approach to enhance the security and stability of microgrid systems in the face of cyber threats.

Computer Science↗
DOE OSTI2025

Solving high-dimensional inverse problems using amortized likelihood-free inference with noisy and incomplete data

Here, we present a likelihood-free probabilistic inversion method based on normalizing flows for high-dimensional inverse problems. The proposed method is composed of two complementary networks: a summary network for data compression and an inference network for parameter estimation. The summary network encodes raw observations into a fixed-size vector of summary features, while the inference network generates samples of the approximate posterior distribution of the model parameters based on these summary features. The posterior samples are produced in a deep generative fashion by sampling from a latent Gaussian distribution and passing these samples through an invertible transformation. We construct this invertible transformation by sequentially alternating conditional invertible neural network and conditional neural spline flow layers. The summary and inference networks are trained simultaneously. We apply the proposed method to an inversion problem in groundwater hydrology to estimate the posterior distribution of the log-conductivity field conditioned on spatially sparse time-series observations of the system’s hydraulic head responses. The conductivity field is represented with 706 degrees of freedom in the considered problem. Comparison with the likelihood-based iterative ensemble smoother PEST-IES method demonstrates that the proposed method accurately estimates the parameter posterior distribution and the observations’ predictive posterior distribution at a fraction of the inference time of PEST-IES.

conditional invertible neural network↗
DOE OSTI2021

Cloud Droplet Measurement System for the ARM Tethered Balloon System (TBS) Field Campaign Report

The Mesa Photonics' cloud droplet measurement system (CDMS) performs in situ measurement of droplet size distribution and droplet number density in clouds. These characteristics are important cloud microphysical properties that are critical input parameters for atmospheric models and are also useful for proper calibration and validation of performance of other atmospheric measurement instrumentation. This small campaign was the first project (out of two) that involved integration of the CDMS into the U.S. Department of Energy (DOE) Atmospheric Radiation Measurement (ARM) user facility’s tethered balloon system (TBS) and its initial testing at the ARM Southern Great Plains (SGP) atmospheric observatory. The follow-up campaign (AFC07016) involved testing of the CDMS under relevant conditions at the third ARM Mobile Facility (AMF3) at Oliktok Point, Alaska and making in situ measurements in clouds. The goal of this small field campaign was to deploy the CDMS on the ARM TBS, demonstrate its integrity and compatibility with the TBS, and test the wireless telecommunication system in preparation for subsequent in-cloud measurements at ARM’s North Slope of Alaska (NSA) observatory at Utqiagvik (formerly Barrow). The specific technical objectives included: (1) Integration of the Mesa Photonics' CDMS into the ARM TBS, (2) testing of the wireless telecommunication system of the CDMS, (3) evaluating the data acquisition software and image processing algorithms especially under high-background-illumination conditions, and (4) testing the ruggedness of the instrument's optical alignment and other performance characteristics. All objectives have been successfully met. The mounting hardware was designed and built mostly at Sandia National Laboratories (SNL), which allowed reliable mounting of the CDMS on the TBS and co-locating the CDMS with other instruments. The field tests were performed by the principal investigator (PI) in collaboration with the TBS operational crew led by D. Dexheimer. The field testing demonstrated good compatibility of the CDMS with the TBS. During the TBS flights, the crew did not experience any operational issues with the CDMS. The instrument demonstrated reliable operational characteristics, including sufficient battery life and good thermal management. The wireless telecommunication system has been successfully tested at TBS flight altitudes of up to 1000 m. During all flights, the CDMS raw data were wirelessly transmitted to the ground station and processed in real time. The ruggedness of the instrument's optical alignment was evaluated by performing calibration of the CDMS (using Mesa Photonics' fixed-size monodisperse droplet generator) before and after the campaign. The calibration was stable within the instrument's measurement precision. In summary, the campaign demonstrated good compatibility of the Mesa Photonics' CDMS with the ARM TBS. The flight tests confirmed the mechanical integrity of the CDMS and stability of its optical layout, as well as reliability of the wireless telecommunication system. Successful completion of this campaign provided a solid basis for the next campaign (AFC07016, November 2020), that involved testing of the CDMS during in-cloud TBS flights at AMF3 at Oliktok Point.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2023

Collisional-energy-cascade model for nonthermal velocity distributions of neutral atoms in plasmas

Nonthermal velocity distributions with much greater tails than a Maxwellian have been observed for radical atoms in plasmas for a long time. Historically, such velocity distributions have been modeled by a two-temperature Maxwell distribution. In this paper, I propose a model based on collisional energy cascade, which has been studied in the field of granular materials. In the collisional energy cascade, a particle ensemble undergoes energy input at the high-energy region, entropy production by elastic collisions among particles, and energy dissipation. For radical atoms, energy input may be caused by the Franck-Condon energy of molecular dissociation or charge-exchange collision with hot ions, and the input energy is eventually dissipated by collisions with the walls. Here, I show that the steady-state velocity distribution in the collisional energy cascade is approximated by the generalized Mittag-Leffler distribution, which is a one-parameter extension of the Maxwell distribution. This parameter indicates the degree of the nonthermality and is related to the relative importance of energy dissipation over entropy production. This model is compared with a direct molecular dynamics simulation for simplified gaseous systems with energy input, as well as some experimentally observed velocity distributions of light radicals in plasmas.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2023

Software-Hardware Co-design of Heterogeneous SmartNIC System for Recommendation Models Inference and Training

Deep Learning Recommendation Models (DLRMs) are critical applications in various domains and have evolved as one of the single largest machine learning applications. Trillions of DLRM parameters exceed the on-chip memory capacity of GPUs. Large-scale multi-node systems are required for distributed DLRM inference and training, which suffer from the all-to-all communication bottleneck, mainly limiting the scalability of ever-growing DLRMs. In recent years, SmartNICs have evolved with coupled computation and communication capabilities providing opportunities for a powerful heterogeneous device in the system. However, there isn't such a distributed system that fully leverages the abundant smartNIC resources that resolve the scalability issue of DLRMs. In this work, we proposed a software-hardware co-design of a heterogeneous smartNIC system that resolves the communication bottleneck of distributed DLRMs, mitigates the memory bandwidth pressure, and improves computation efficiency. We provide a set of smartNIC designs of cache systems (including local cache and remote cache) and smartNIC computation kernels which reduce data movement, relieve memory lookup intensity, and improve the GPU's computation efficiency. In addition, we propose a graph algorithm that improves the data locality of queries within batches which optimizes the overall system performance with higher data reuse. Our evaluation shows that our system achieves 2.1x latency speedup for inference and 1.6x throughput speedup for training.

Guo, Anqi↗
DOE OSTI2021

Microstructure-Based Estimation of Strength and Ductility Distributions for \(\alpha +\beta \) Titanium Alloys

Titanium alloys are processed to develop a wide range of microstructure configurations and therefore material properties. While these properties are typically measured experimentally, a framework for property prediction could greatly enhance alloy design and manufacturing. In this work, a microstructure-sensitive framework is presented for the prediction of strength and ductility as well as estimates of the bounds in variability for these properties. The framework explicitly considers distributions of microstructure via new approaches for instantiation of structure in synthetic samples. The parametric evaluation strategy, including the finite element simulation package FEpX, is used to create and test virtual polycrystalline samples to evaluate the variability bounds of mechanical properties in Ti-6Al-4V. Critical parameters for the property evaluation framework are provided by measurements of single crystal properties and advanced characterization of microstructure and slip system strengths in 2D and 3D. Property distributions for yield strength and ductility are presented, along with the validation and verification steps undertaken. Comparisons between strain localization and slip activity in virtual samples and in experimental grain-scale strain measurements are also discussed.

36 MATERIALS SCIENCE↗
DOE OSTI2023

Parallel Grid-Forming Inverter-Driven Black Start: Power-Hardware-in-the-Loop Validation: Preprint

With power systems encountering increasing deployment levels of inverter-based resources (IBRs), system restoration using grid-forming (GFM) IBRs has gained attention. Engineered to establish grid voltages in the absence of a stiff grid, black-start-capable GFM IBRs are expected to enhance power system resilience by playing a critical role in bottom-up system restoration. This paper experimentally studies the feasibility of the novel approach through power-hardware-in-the-loop (PHIL) testing and validation with a commercial GFM inverter. The PHIL test setup demonstrates an inverter-driven black start of a 5-MW unbalanced distribution feeder where two GFM inverters collectively black start the feeder: one is a commercial hardware GFM inverter interacting through the PHIL interface, and the other is a software GFM emulated by a real-time simulation with full electromagnetic transient (EMT) models. Both GFM inverters are equipped with negative-sequence voltage control to suppress the voltage imbalance resulting from the unbalanced loading, allowing us to study the dynamic interactions between the GFMs with their control parameters unknown. To evaluate the dynamic behavior of the GFM inverters under the entire black-start process, the EMT model of the distribution system details the transformer and motor dynamics to emulate their inrush and startup behaviors. It abstracts conventional grid-connected inverter dynamics, i.e., grid-following inverters. Oscilloscope measurements of the hardware GFM inverter are also presented for critical steps. Takeaways for further study and field deployment are provided.

grid-forming inverter↗
DOE OSTI2021

Pseudorapidity distributions of charged particles in pp($\bar p$), p(d)A and AA collisions using Tsallis thermodynamics

The pseudorapidity distributions of charged particles measured in p + p($\overline{\text{p}}$) collisions for energies ranging from $\sqrt{{s}_{\mathrm{N}\mathrm{N}}}=23.6$ GeV to 13 TeV and A + A collisions at RHIC and LHC are investigated in the fireball model with Tsallis thermodynamics. We assume that the rapidity axis is populated with fireballs following q-Gaussian distribution and the charged particles follow the Tsallis distribution in the fireball. We also extend the fireball model to asymmetric collision systems, i.e. d + Au collisions at $\sqrt{{s}_{\mathrm{N}\mathrm{N}}}=200$ GeV and p + Pb collisions at $\sqrt{{s}_{\mathrm{N}\mathrm{N}}}=5.02$ TeV, by taking into account the asymmetric geometry configuration. The model can fit the experimental data well for all the collision systems and centralities investigated. The collision energy and centrality dependence of the model parameters for the symmetric (asymmetric) collision system, i.e. the central position y 0 (y 0a , y 0A ) and its width σ (σ a , σ A ) of the fireball distribution, are also investigated and discussed. Furthermore, our results suggest that the fireball model with Tsallis thermodynamics can be used as a universal framework for the pseudorapidity distributions of charged particles in high energy collisions at RHIC and LHC.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2022

Distributed deep learning training using silicon photonic switched architectures

The scaling trends of deep learning models and distributed training workloads are challenging network capacities in today’s datacenters and high-performance computing (HPC) systems. We propose a system architecture that leverages silicon photonic (SiP) switch-enabled server regrouping using bandwidth steering to tackle the challenges and accelerate distributed deep learning training. In addition, our proposed system architecture utilizes a highly integrated operating system-based SiP switch control scheme to reduce implementation complexity. To demonstrate the feasibility of our proposal, we built an experimental testbed with a SiP switch-enabled reconfigurable fat tree topology and evaluated the network performance of distributed ring all-reduce and parameter server workloads. The experimental results show up to 3.6× improvements over the static non-reconfigurable fat tree. Our large-scale simulation results show that server regrouping can deliver up to 2.3× flow throughput improvement for a 2× tapered fat tree and a further 11% improvement when higher-layer bandwidth steering is employed. The collective results show the potential of integrating SiP switches into datacenters and HPC systems to accelerate distributed deep learning training.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗