Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “extensible architecture”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 199 records · Page 11

Towards Efficient Convolutional Neural Networks Through Low-Error Filter Saliency Estimation

Filter saliency based channel pruning is a state-of-the-art method for deep convolutional neural network compression and acceleration. This channel pruning method ranks the importance of individual filter by estimating its impact of each filter’s removal on the training loss, and then remove the least important filters and fine-tune the remnant network. In this work, we propose a systematic channel pruning method that significantly reduces the estimation error of filter saliency. Different from existing approaches, our method largely reduces the magnitude of parameters in a network by introducing alternating direction method of multipliers (ADMM) into the pre-training procedure. Therefore, the estimation of filter saliency based on Taylor expansion is significantly improved. Extensive experiments with various benchmark network architectures and datasets demonstrate that the proposed method has a much improved unimportant filter selection capability and outperform state-of-the-art channel pruning method.

Wang, Zi↗

Deep neural operators can predict the real-time response of floating offshore structures under irregular waves

The use of neural operators in a digital twin model of an offshore floating structure holds the potential for a significant shift in the prediction of structural responses and health monitoring, offering valuable real-time control insights. In this work, we investigate the effectiveness of three neural operators, namely the deep operator network (DeepONet), the Fourier neural operator (FNO), and the Wavelet neural operator (WNO), to accurately capture the responses of a floating structure under six different sea state codes (3 − 8) based on the wave characteristics described by the World Meteorological Organization (WMO). To further enhance the accuracy of the vanilla architecture of the neural operators, novel extensions, such as wavelet-DeepONet and self-adaptive WNO, are proposed in this paper. The results demonstrate that these high-precision neural operators can deliver structural responses more efficiently, up to two orders of magnitude faster than a dynamic analysis using conventional numerical solvers. Additionally, compared to gated recurrent units (GRUs), a commonly used recurrent neural network for time-series estimation, neural operators are both more accurate and efficient, especially in situations with limited data availability. Taken together, our study shows that FNO outperforms all other operators for approximating the mapping of one input functional space to the output space as well as for responses that have small bandwidth of the frequency spectrum. Conversely, DeepONet, with historical states, proves most accurate in learning the mapping of multiple input functions to the output space and capturing responses within a broad frequency spectrum.

97 MATHEMATICS AND COMPUTING↗

Experimental Demonstration of Frequency Regulation by Commercial Buildings – Part II: Results and Performance Evaluation

This paper is the second part of a two-part series presenting the results from an experimental demonstration of frequency regulation in a commercial building test facility. We developed relevant building models and designed a hierarchical controller for reserve scheduling, building climate control and frequency regulation in Part I. In Part II, we introduce the communication architecture and experiment settings, and present extensive experimental results under frequency regulation. More specifically, we compute the day-ahead reserve capacity of the test facility under different assumptions and conditions. Furthermore, we demonstrate the ability of model predictive control to satisfy comfort constraints under frequency regulation, and show that fan speed control can track the fast-moving RegD signal of the Pennsylvania, Jersey, and Maryland Power Market (PJM) very accurately. In addition, we discuss potential effects of frequency regulation on building operation (e.g., increase in energy consumption, oscillations in supply air temperature, and effect on chiller cycling), and provide suggestions for real-world implementation projects. Lastly, our results show that hierarchical control is appropriate for frequency regulation from commercial buildings.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Archparse

Archparse is a Python package that holds the purpose and capability of converting the contents of a text file to a functioning neural network based on the Tensorflow 2.X framework. This is able to be done with minimal written Python code and next to no knowledge of how to build models with Tensorflow directly. More specifically Archparse parses a text file with extension ".arch" which contains neural network architecture information that corresponds to either the Tensorflow 2.X API or custom code written with the Tensorflow 2.X API. Included in the initial version is the capacity to easily produce sequential autoencoders and sequential neural networks.

Vander Wal, MichaelD.↗

Towards a data-driven model of hadronization using normalizing flows

We introduce a model of hadronization based on invertible neural networks that faithfully reproduces a simplified version of the Lund string model for meson hadronization. Additionally, we introduce a new training method for normalizing flows, termed MAGIC, that improves the agreement between simulated and experimental distributions of high-level (macroscopic) observables by adjusting single-emission (microscopic) dynamics. Our results constitute an important step toward realizing a machine-learning based model of hadronization that utilizes experimental data during training. Finally, we demonstrate how a Bayesian extension to this normalizing-flow architecture can be used to provide analysis of statistical and modeling uncertainties on the generated observable distributions.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Quantum Monte Carlo Endstation for Petascale Computing

NCSU research group has been focused on accomplising the key goals of this initiative: establishing new generation of quantum Monte Carlo (QMC) computational tools as a part of Endstation petaflop initiative for use at the DOE ORNL computational facilities and for use by computational electronic structure community at large; carrying out high accuracy quantum Monte Carlo demonstration projects in application of these tools to the forefront electronic structure problems in molecular and solid systems; expanding the impact of QMC methods and approaches; explaining and enhancing the impact of these advanced computational approaches. In particular, we have developed quantum Monte Carlo code (QWalk, www.qwalk.org) which was significantly expanded and optimized using funds from this support and at present became an actively used tool in the petascale regime by ORNL researchers and beyond. These developments have been built upon efforts undertaken by the PI's group and collaborators over the period of the last decade. The code was optimized and tested extensively on a number of parallel architectures including petaflop ORNL Jaguar machine. We have developed and redesigned a number of code modules such as evaluation of wave functions and orbitals, calculations of pfaffians and introduction of backflow coordinates together with overall organization of the code and random walker distribution over multicore architectures. We have addressed several bottlenecks such as load balancing and verified efficiency and accuracy of the calculations with the other groups of the Endstation team. The QWalk package contains about 50,000 lines of high quality object-oriented C++ and includes also interfaces to data files from other conventional electronic structure codes such as Gamess, Gaussian, Crystal and others. This grant supported PI for one month during summers, a full-time postdoc and partially three graduate students over the period of the grant duration, it has resulted in 13 published papers, 15 invited talks and lectures nationally and internationally. My former graduate student and postdoc Dr. Michal Bajdich, who was supported byt this grant, is currently a postdoc with ORNL in the group of Dr. F. Reboredo and Dr. P. Kent and is using the developed tools in a number of DOE projects. The QWalk package has become a truly important research tool used by the electronic structure community and has attracted several new developers in other research groups. Our tools use several types of correlated wavefunction approaches, variational, diffusion and reptation methods, large-scale optimization methods for wavefunctions and enables to calculate energy differences such as cohesion, electronic gaps, but also densities and other properties, using multiple runs one can obtain equations of state for given structures and beyond. Our codes use efficient numerical and Monte Carlo strategies (high accuracy numerical orbitals, multi-reference wave functions, highly accurate correlation factors, pairing orbitals, force biased and correlated sampling Monte Carlo), are robustly parallelized and enable to run on tens of thousands cores very efficiently. Our demonstration applications were focused on the challenging research problems in several fields of materials science such as transition metal solids. We note that our study of FeO solid was the first QMC calculation of transition metal oxides at high pressures.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

The connectome of the adult Drosophila mushroom body provides insights into function

Making inferences about the computations performed by neuronal circuits from synapse-level connectivity maps is an emerging opportunity in neuroscience. The mushroom body (MB) is well positioned for developing and testing such an approach due to its conserved neuronal architecture, recently completed dense connectome, and extensive prior experimental studies of its roles in learning, memory, and activity regulation. Here, we identify new components of the MB circuit in Drosophila , including extensive visual input and MB output neurons (MBONs) with direct connections to descending neurons. We find unexpected structure in sensory inputs, in the transfer of information about different sensory modalities to MBONs, and in the modulation of that transfer by dopaminergic neurons (DANs). We provide insights into the circuitry used to integrate MB outputs, connectivity between the MB and the central complex and inputs to DANs, including feedback from MBONs. Our results provide a foundation for further theoretical and experimental work.

Li, Feng↗

Laser transmitter for Lidar In-Space Technology Experiment

The Lidar In-Space Technology Experiment (LITE) Laser Transmitter Module (LTM) flight laser optical architecture has been space qualified by extensive testing at the system, subsystem and component level. The projected system output performance has been verified using an optically and electrically similar breadboard version of the laser. Parasitic lasing was closely examined and completely suppressed after design changes were implemented and tested. Oscillator and amplifier type heads were separately tested to 150 million shots. Critical subassemblies have undergone environmental testing to Shuttle qualification levels. A superior three color anti-reflection coating was developed and tested for use on 14 surfaces after the final amplifier.

Chang, John↗

Eigensolver for a Sparse, Large Hermitian Matrix

A parallel-processing computer program finds a few eigenvalues in a sparse Hermitian matrix that contains as many as 100 million diagonal elements. This program finds the eigenvalues faster, using less memory, than do other, comparable eigensolver programs. This program implements a Lanczos algorithm in the American National Standards Institute/ International Organization for Standardization (ANSI/ISO) C computing language, using the Message Passing Interface (MPI) standard to complement an eigensolver in PARPACK. [PARPACK (Parallel Arnoldi Package) is an extension, to parallel-processing computer architectures, of ARPACK (Arnoldi Package), which is a collection of Fortran 77 subroutines that solve large-scale eigenvalue problems.] The eigensolver runs on Beowulf clusters of computers at the Jet Propulsion Laboratory (JPL).

Tisdale, E. Robert↗

AutoBayes Program Synthesis System System Internals

This lecture combines the theoretical background of schema based program synthesis with the hands-on study of a powerful, open-source program synthesis system (Auto-Bayes). Schema-based program synthesis is a popular approach toward program synthesis. The lecture will provide an introduction into this topic and discuss how this technology can be used to generate customized algorithms. The synthesis of advanced numerical algorithms requires the availability of a powerful symbolic (algebra) system. Its task is to symbolically solve equations, simplify expressions, or to symbolically calculate derivatives (among others) such that the synthesized algorithms become as efficient as possible. We will discuss the use and importance of the symbolic system for synthesis. Any synthesis system is a large and complex piece of code. In this lecture, we will study Autobayes in detail. AutoBayes has been developed at NASA Ames and has been made open source. It takes a compact statistical specification and generates a customized data analysis algorithm (in C/C++) from it. AutoBayes is written in SWI Prolog and many concepts from rewriting, logic, functional, and symbolic programming. We will discuss the system architecture, the schema libary and the extensive support infra-structure. Practical hands-on experiments and exercises will enable the student to get insight into a realistic program synthesis system and provides knowledge to use, modify, and extend Autobayes.

Statistical Algorithms↗

Julia Programming Language Benchmark Using a Flight Simulation

Julia is a relatively new computer language that aims to reduce the challenge for mathmodelers to develop fast computer tools and simulations. It potentially combines the ease-of-coding feature of scripting languages (like Python) with the performance of compiled languages (like C++). A key question for Julia application to the simulation domain is, “Can Julia, with its obvious coding simplicity, provide runtime speeds comparable to conventional compiled languages for flight simulation?” A unique combination of existing elements can be employed to address the previous question: Extensively documented object-oriented simulation architecture, Industry standard rocket flight simulation, Separate versions (C++, Java, and Python) already benchmarked.

Sells, Ray↗

Supporting the Development of Resilient Message Passing Applications using Simulation

An emerging aspect of high-performance computing (HPC) hardware/software co-design is investigating performance under failure. The work in this paper extends the Extreme-scale Simulator (xSim), which was designed for evaluating the performance of message passing interface (MPI) applications on future HPC architectures, with fault-tolerant MPI extensions proposed by the MPI Fault Tolerance Working Group. xSim permits running MPI applications with millions of concurrent MPI ranks, while observing application performance in a simulated extreme-scale system using a lightweight parallel discrete event simulation. The newly added features offer user-level failure mitigation (ULFM) extensions at the simulated MPI layer to support algorithm-based fault tolerance (ABFT). The presented solution permits investigating performance under failure and failure handling of ABFT solutions. The newly enhanced xSim is the very first performance tool that supports ULFM and ABFT.

Engelmann, Christian↗

An Overview of the Layered and Extensible Aircraft Performance System (LEAPS) Development

The Layered and Extensible Aircraft Performance System (LEAPS) is a new sizing and synthesis tool being developed within the Aeronautics Systems Analysis Branch (ASAB) at NASA Langley Research Center. It is a modular, multidisciplinary, multi- fidelity sizing and synthesis tool for modeling advanced aircraft concepts and architectures such as electric/hybrid-electric propulsion, unconventional propulsion airframe integration, and non-traditional mission trajectories. The development of LEAPS is motivated by the lack of existing tools that meet the needs of ASAB. The Flight Optimization System (FLOPS) has been the primary sizing and synthesis tool of ASAB for three decades. However, FLOPS has a number of limitations that make it dicult to use for unconventional aircraft designs. Three high-level goals have been adopted to guide the LEAPS development pro- cess. LEAPS is being developed in Python with an architecture built to enable a exible and extensible analysis capability using the concept of an aircraft object that combines data and analysis models. Five challenge problems for LEAPS have been identi ed to measure progress: analysis of a conventional tube-and-wing aircraft using legacy methods, coupled aeroelastic analysis for weight estimation of a conventional tube-and-wing aircraft, analysis of an advanced hybrid-electric concept, analysis of the X-57 Maxwell distributed electric propulsion aircraft, and optimization of the trajectory of a supersonic vehicle to minimize sonic boom. LEAPS will be a publicly available capability of exceptional quality with modularity and extensibility that makes it a robust tool for design and analysis of current and future unconventional aircraft concepts.

Welstead, Jason R.↗

GridPACK: Library Architecture and Usability Assessment

The LLNL team has carried out extensive analysis of GridPACK code, including two major code reviews (each being a 2-day event) and a vendor emulation exercise, where the team integrated GridPACK library within a third-party application. The code analysis identified a number of areas for improvement of GridPACK code architecture, but it did not find fundamental flaws in the GridPACK design. The team found the GridPACK strategy of developing a software development kit (SDK) for power systems application development to be sound and well aligned with Office of Electricity long term objectives. The SDK aims to provide power systems software vendors an easy access to high performance computing (HPC) libraries.

24 POWER TRANSMISSION AND DISTRIBUTION↗

EV Everywhere Sandia Appendix

Sandia National Laboratories has a large cybersecurity research program supporting the Department of Energy as well as other federal departments and agencies. Fundamental research helps us stay ahead of threats. Sandia’s research efforts in cybersecurity are focused in three broad areas: 1. Trusted hardware, software, and systems 2. Networks and systems architectures and analysis 3. Effective cyber defense systems Research is conducted in state-of-the-art facilities with extensive computing and information science capabilities, which range from architectures and algorithms to advanced modeling and simulation. Three areas of active research most applicable to connected vehicle security are: Red Team Assessments, Emulytics™, and Supply Chain Risk Management.

97 MATHEMATICS AND COMPUTING↗

Trade Studies of Space Launch Architectures using Modular Probabilistic Risk Analysis

A top-down risk assessment in the early phases of space exploration architecture development can provide understanding and intuition of the potential risks associated with new designs and technologies. In this approach, risk analysts draw from their past experience and the heritage of similar existing systems as a source for reliability data. This top-down approach captures the complex interactions of the risk driving parts of the integrated system without requiring detailed knowledge of the parts themselves, which is often unavailable in the early design stages. Traditional probabilistic risk analysis (PRA) technologies, however, suffer several drawbacks that limit their timely application to complex technology development programs. The most restrictive of these is a dependence on static planning scenarios, expressed through fault and event trees. Fault trees incorporating comprehensive mission scenarios are routinely constructed for complex space systems, and several commercial software products are available for evaluating fault statistics. These static representations cannot capture the dynamic behavior of system failures without substantial modification of the initial tree. Consequently, the development of dynamic models using fault tree analysis has been an active area of research in recent years. This paper discusses the implementation and demonstration of dynamic, modular scenario modeling for integration of subsystem fault evaluation modules using the Space Architecture Failure Evaluation (SAFE) tool. SAFE is a C++ code that was originally developed to support NASA s Space Launch Initiative. It provides a flexible framework for system architecture definition and trade studies. SAFE supports extensible modeling of dynamic, time-dependent risk drivers of the system and functions at the level of fidelity for which design and failure data exists. The approach is scalable, allowing inclusion of additional information as detailed data becomes available. The tool performs a Monte Carlo analysis to provide statistical estimates. Example results of an architecture system reliability study are summarized for an exploration system concept using heritage data from liquid-fueled expendable Saturn V/Apollo launch vehicles.

Mathias, Donovan L.↗

Human N -acetylglucosaminyltransferase II substrate recognition uses a modular architecture that includes a convergent exosite

Asn-linked oligosaccharides are extensively modified during transit through the secretory pathway, first by trimming of the nascent glycan chains and subsequently by initiating and extending multiple oligosaccharide branches from the trimannosyl glycan core. Trimming and branching pathway steps are highly ordered and hierarchal based on the precise substrate specificities of the individual biosynthetic enzymes. A key committed step in the synthesis of complex-type glycans is catalyzed by N-acetylglucosaminyltransferase II (MGAT2), an enzyme that generates the second GlcNAcβ1,2- branch from the trimannosyl glycan core using UDP-GlcNAc as the sugar donor. We determined the structure of human MGAT2 as a Mn 2+ -UDP donor analog complex and as a GlcNAcMan 3 GlcNAc 2 -Asn acceptor complex to reveal the structural basis for substrate recognition and catalysis. The enzyme exhibits a GT-A Rossmann-like fold that employs conserved divalent cation-dependent substrate interactions with the UDP-GlcNAc donor. MGAT2 interactions with the extended glycan acceptor are distinct from other related glycosyltransferases. These interactions are composed of a catalytic subsite that binds the Man-α1,6- monosaccharide acceptor and a distal exosite pocket that binds the GlcNAc-β1,2Man-α1,3Manβ- substrate “recognition arm.” Recognition arm interactions are similar to the enzyme–substrate interactions for Golgi α-mannosidase II, a glycoside hydrolase that acts just before MGAT2 in the Asn-linked glycan biosynthetic pathway. Here, these data suggest that substrate binding by MGAT2 employs both conserved and convergent catalytic subsite modules to provide substrate selectivity and catalysis. More broadly, the MGAT2 active-site architecture demonstrates how glycosyltransferases create complementary modular templates for regiospecific extension of glycan structures in mammalian cells.

59 BASIC BIOLOGICAL SCIENCES↗

HARMONY: Large-Scale Architecture Search for Efficient Hybrid Language Models

As large language models scale to trillions of parameters, their computational and memory requirements present critical challenges for efficient training and deployment. While Mixture of Experts (MoE) architectures enable efficient scaling through sparse parameter activation, and state-space models like Mamba offer linear-time complexity, principled methods for combining these paradigms remain undeveloped. We introduce HARMONY (Hybrid Architecture Research for Mamba, Optimized with Neural efficiencY), a multi-objective evolutionary neural architecture search framework for discovering efficient hybrid language models that integrate Transformer attention mechanisms, Mixture-of-Experts routing, and Mamba state-space components. Through large-scale distributed search using 16,384 MI250X GPUs on the Frontier supercomputer, HARMONY explores a comprehensive design space encompassing six attention variants (MHA, MQA, GQA, MLA, SWA, and Mamba-2), variable MoE configurations with both routed and shared experts, and extensive Mamba hyperparameters. Our framework discovers heterogeneous architectures that balance training performance with computational efficiency through multi-objective optimization incorporating latency penalties and fitness-based selection. Analysis of discovered architectures reveals that optimal hybrid designs favor heterogeneous component mixing rather than homogeneous patterns, with Mamba-2 and Multi-Head Latent Attention (MLA) emerging as preferred mechanisms. Discovered architectures demonstrate superior training efficiency: our best configuration achieves a final perplexity of 1.0874 with 2.38B parameters while processing 4,320 tokens/second, outperforming significantly larger manually designed models. Full-scale evaluation shows HARMONY's top architectures achieve better loss trajectories than equivalently-sized models using state-of-the-art configurations including Mixtral, Jamba, and Samba. Additionally, we demonstrate 91% weak scaling efficiency when training discovered 36B-parameter models across 1,024 GPUs. HARMONY is released as an open framework with comprehensive tools for building and training hybrid models using expert-data-pipeline parallelism, democratizing access to automated architecture design for next-generation language models.

Herron, Emily [ORNL] (ORCID:0000000273008172)↗