SEARCH · Engineering Papers

Results for “mathematics of arrays”

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.

SourceResearch connectionFrom yearThrough year

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

At least 325 records · Page 18

All sources NASA NTRS DOE OSTI

DOE OSTI2024

The VIBES Are Shifting: Assessing Emergent Capabilities in Multi-Modal Models

Researchers assessing open-source domains such as the internet, and particularly those studying information conflict, often have no single prescribed workflow. In the course of their research, they may need to perform a diverse array of tasks far beyond simply identifying an ever-changing set of objects. These analytical tasks can include ascertaining the provenance of images, understand an image in the context of accompanying text-based data, or identifying indicators of digital image manipulation. They must further be able to do this at the scale of tens of thousands of images or more. Traditional machine vision models have typically lacked the flexibility and breadth of performance sufficient for these needs. The research team from Pacific Northwest National Laboratory assessed the performance of a single baseline CLIP ViT-L model against a series of analytical tasks relevant for the study of online information conflict.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2023

Machine Learning Models for Network Traffic Classification in Programmable Logic

Network traffic classification via machine learning on network packet payloads has emerged as an active area of research for network security due to the high accuracy machine learning models have achieved in classifying payloads. For effective deployment as part of network security, these machine learning models must not only classify malicious packet payloads accurately, they must also identify anomalous payloads and perform inference at speeds generally faster than 10,000 packets per second to be effective. This work explores the in- ference speeds and accuracy of several neural network models implemented in programmable logic on various field programmable gate arrays (FPGA) including the Xilinx VC1902 and Xilinx Zynq Ultrascale+. This work also presents the design and performance of both an autoencoder and variational autoencoder programmed on the FPGA for identifying anomalous packet payloads. The performance benefits of the FPGA implementation for this type of packet payload inspection driven by machine learning are compared against graphics processing unit (GPU) inference implementations run on two state-of-the-art datacenter GPU devices, the NVIDIA V100 and A100. The model accuracy difference between the FPGA and GPU implementations was found to be 4% or less while the Xilinx VC1902 outperformed both the NVIDIA V100 and A100 for inference speeds on all the models explored except the variational autoencoder.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2020

Final Reports of the 2020 Los Alamos National Laboratory Computational Physics Student Summer Workshop

For the past ten years, the workshop has been bringing a highly talented and diverse group of student every summer. Students work in teams of two, alongside typically two mentors, on research projects reflecting a broad range of topics within computational physics. In addition, students attend a series of lectures on topics within computational physics, facility tours, and networking events. The program lasts ten weeks, with this year’s workshop running from June 8 to August 14. At the end of the summer, students give a final presentation, along with a written report. Those reports are what make up the remaining sections of this document. Admission to the workshop is by a competitive process, with the mentors forming the selection committee. One of the important accomplishments of the workshop has been to create a student pipeline from diverse schools that sometimes are not normally tapped by LANL recruiting. Many workshop students maintain a continuing relationship with LANL, returning as students interns, post-doctoral researchers, and staff members. Additionally, workshop alumni act as ambassadors for LANL. The result is a wider awareness both of LANL as a potential employer, and of the technical work that happens at LANL. This year, the workshop format was changed in several ways, in order to accommodate the off-site, virtual format. Students worked on LANL virtual desktop systems remotely, also accessing LANL HPC resources. In order to facilitate communication, student were given accounts on both Webex, a video teleconferencing platform, and Mattermost, an online team collaboration and chat platform, similar to Slack. Daily communication between students and mentors was primarily on Mattermost, with Webex conferencing as needed. The lectures were all done on Webex. Given the difficulty of the virtual format, and a concern that students might have video teleconferencing burn-out after an academic semester largely moved to that format, all lectures were optional this year. In spite of this, the attendance was generally high. Lecturers were asked to try to move to a more high-level, ”What is it?,” format. Once again, the students did a tremendous job. Over the course of ten weeks, they did important research across a staggering array of disciplines. The following pages contain the final report for each team’s research efforts. We hope you will find reading them as exciting as it was for us to produce them.

36 MATERIALS SCIENCE↗

DOE OSTI2021

Final Reports of the 2021 Los Alamos National Laboratory Computational Physics Student Summer Workshop

Since 2011, the Los Alamos National Laboratory Computational Physics Student Summer Workshop has been bringing together a highly talented and diverse group of students every summer. Students work in teams of two, alongside typically two mentors, on research projects reflecting a broad range of topics within computational physics. In addition, students attend a series of lectures on topics within computational physics, facility tours, and networking events. The program lasts ten weeks, with this year’s workshop running from June 7 to August 13. At the end of the summer, students give a final presentation, along with a written report. Those reports are what make up the remaining sections of this document. Admission to the workshop is by a competitive process, with the mentors forming the selection committee. One of the important accomplishments of the workshop has been to create a student pipeline from diverse schools that sometimes are not normally tapped by LANL recruiting. Many workshop students maintain a continuing relationship with LANL, returning as student interns, post-doctoral researchers, and staff members. Additionally, workshop alumni act as ambassadors for LANL. The result is a wider awareness both of LANL as a potential employer, and of the technical work that happens at LANL. This year, the workshop was once again in an off-site, virtual format. Students worked on LANL virtual desktop systems remotely, also accessing LANL HPC resources. In order to facilitate communication, student were given accounts on both Webex, a video teleconferencing platform, and Mattermost, an online team collaboration and chat platform, similar to Slack. Daily communication between students and mentors was primarily on Mattermost, with Webex conferencing as needed. The lectures were all done on Webex. Given the difficulty of the virtual format, and a concern that students might have video teleconferencing burn-out, all lectures were optional this year. In spite of this, the attendance was generally high. Lecturers were asked to try to move to a more high-level, ”What is it?,” format. Once again, the students did a tremendous job. Over the course of ten weeks, they did important research across a staggering array of disciplines. The following pages contain the final report for each team’s research efforts. Enjoy!

45 MILITARY TECHNOLOGY, WEAPONRY, AND NATIONAL DEF↗

DOE OSTI2023

Scaling Wind Power Innovation Assessment for Rapid Energy Transition with Artificial Intelligence

Planning for energy system decarbonization requires new insights into the potential of renewable technologies, deployed at unprecedented scale, to meet urgent sustainability goals. However, limited scalability of current wind energy research tools restricts characterization of innovation impacts to isolated reference sites, challenging investment and decision making under rapid growth. We demonstrate the transformative potential of artificial intelligence (AI) to inform future technology advancement and energy systems design by leveraging a state-of-the-art surrogate model to conduct a series of fleet-wide wind plant layout optimizations for greater than 6,800 projected U.S. onshore buildout locations. We show how innovative wake steering technology can address an array of barriers to large-scale deployment and integration of wind power. Specifically, wake steering reduces required plant area by an average of 18% and could preserve upwards of 13,000 km2 for future greenfield deployment, potentially easing siting challenges associated with wind energy infrastructure. Further, by enabling reduced turbine spacing and increased energy production, flexible operations of wake steering improve levelized cost of energy, particularly for large plants and in land-constrained settings. Finally, optimizations that consider dynamic energy prices can deliver increased power production and revenue capture during high-value (often low-wind) periods, further bolstering plant economics. Our computationally efficient approach offers a pathway to accelerate nationwide geographic evaluation of innovative technologies.

graph neural networks↗

NASA NTRS2010

Kepler Mission's Focal Plane Characterization Models Implementation

The Kepler Mission photometer is an unusually complex array of CCDs. A large number of time-varying instrumental and systemic effects must be modeled and removed from the Kepler pixel data to produce light curves of sufficiently high quality for the mission to be successful in its planet-finding objective. After the launch of the spacecraft, many of these effects are difficult to remeasure frequently, and various interpolations over a small number of sample measurements must be used to determine the correct value of a given effect at different points in time. A library of software modules, called Focal Plane Characterization (FC) Models, is the element of the Kepler Science Data Pipeline (hereafter "pipeline") that handles this. FC, or products generated by FC, are used by nearly every element of the SOC processing chain. FC includes Java components: database persistence classes, operations classes, model classes, and data importers; and MATLAB code: model classes, interpolation methods, and wrapper functions. These classes, their interactions, and the database tables they represent, are discussed. This paper describes how these data and the FC software work together to provide the pipeline with the correct values to remove non-photometric effects caused by the photometer and its electronics from the Kepler light curves. The interpolation mathematics is reviewed, as well as the special case of the sky-to-pixel,pixel-to-sky coordinate transformation code, which incorporates a compound model that is unique in the SOC software.

mission↗

NASA NTRS2006

Jet Noise Diagnostics Supporting Statistical Noise Prediction Methods

The primary focus of my presentation is the development of the jet noise prediction code JeNo with most examples coming from the experimental work that drove the theoretical development and validation. JeNo is a statistical jet noise prediction code, based upon the Lilley acoustic analogy. Our approach uses time-average 2-D or 3-D mean and turbulent statistics of the flow as input. The output is source distributions and spectral directivity. NASA has been investing in development of statistical jet noise prediction tools because these seem to fit the middle ground that allows enough flexibility and fidelity for jet noise source diagnostics while having reasonable computational requirements. These tools rely on Reynolds-averaged Navier-Stokes (RANS) computational fluid dynamics (CFD) solutions as input for computing far-field spectral directivity using an acoustic analogy. There are many ways acoustic analogies can be created, each with a series of assumptions and models, many often taken unknowingly. And the resulting prediction can be easily reverse-engineered by altering the models contained within. However, only an approach which is mathematically sound, with assumptions validated and modeled quantities checked against direct measurement will give consistently correct answers. Many quantities are modeled in acoustic analogies precisely because they have been impossible to measure or calculate, making this requirement a difficult task. The NASA team has spent considerable effort identifying all the assumptions and models used to take the Navier-Stokes equations to the point of a statistical calculation via an acoustic analogy very similar to that proposed by Lilley. Assumptions have been identified and experiments have been developed to test these assumptions. In some cases this has resulted in assumptions being changed. Beginning with the CFD used as input to the acoustic analogy, models for turbulence closure used in RANS CFD codes have been explored and compared against measurements of mean and rms velocity statistics over a range of jet speeds and temperatures. Models for flow parameters used in the acoustic analogy, most notably the space-time correlations of velocity, have been compared against direct measurements, and modified to better fit the observed data. These measurements have been extremely challenging for hot, high speed jets, and represent a sizeable investment in instrumentation development. As an intermediate check that the analysis is predicting the physics intended, phased arrays have been employed to measure source distributions for a wide range of jet cases. And finally, careful far-field spectral directivity measurements have been taken for final validation of the prediction code. Examples of each of these experimental efforts will be presented. The main result of these efforts is a noise prediction code, named JeNo, which is in middevelopment. JeNo is able to consistently predict spectral directivity, including aft angle directivity, for subsonic cold jets of most geometries. Current development on JeNo is focused on extending its capability to hot jets, requiring inclusion of a previously neglected second source associated with thermal fluctuations. A secondary result of the intensive experimentation is the archiving of various flow statistics applicable to other acoustic analogies and to development of time-resolved prediction methods. These will be of lasting value as we look ahead at future challenges to the aeroacoustic experimentalist.

Bridges, James E.↗

DOE OSTI2024

Field Programmable Gate Array-Based Reactor Protection Systems and Potential for Inclusion of Secure Elements to Improve Cybersecurity

For acceptable implementations of technologies like wireless communications, remote monitoring, etc., strong mitigations must be developed and evaluated to ensure that new attack pathways do not increase risk for Advanced Reactors. Secure Elements can be adopted and adapted for this purpose based on tamper resistance and cryptographic abilities, but research must be done to properly integrate into critical components such as FPGA-based Important to Safety systems in conjunction with current and future regulations on cyber security features in Advanced Reactors. Typically, the integration of a Secure Element happens during the POST and UEFI boot of a computing platform, performed by the Operating System, which is not possible with FPGAs because they do not include these firmware components. Work must be done to identify a reliable and secure method for integration in FPGA-based systems which lack Operating Systems and therefore complex boot procedures, system calls, etc.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2023

Ultra-low latency recurrent neural network inference on FPGAs for physics applications with hls4ml

Abstract Recurrent neural networks have been shown to be effective architectures for many tasks in high energy physics, and thus have been widely adopted. Their use in low-latency environments has, however, been limited as a result of the difficulties of implementing recurrent architectures on field-programmable gate arrays (FPGAs). In this paper we present an implementation of two types of recurrent neural network layers—long short-term memory and gated recurrent unit—within the hls4ml framework. We demonstrate that our implementation is capable of producing effective designs for both small and large models, and can be customized to meet specific design requirements for inference latencies and FPGA resources. We show the performance and synthesized designs for multiple neural networks, many of which are trained specifically for jet identification tasks at the CERN Large Hadron Collider.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2021

Probabilistic Risk Assessment Research Needs and Activities

Probabilistic risk assessments (PRAs) have benefitted the safe operation of the U.S. reactor fleet over many decades. Risk insights from these PRAs have provided information from many different perspectives, from what is most important to maintain at a facility to a better understanding of how to address new information regarding safety issues. The methods and tools that have supported the creation and enhancement of PRA models were established through multiple decades of research, from the 1970s starting with the WASH-1400 Reactor Safety Study through comprehensive plant-specific models in use today. Because PRA models are being used to support such a diverse array of plant decisions, they are now being applied to analyze an increasing number and diverse set of aspects of the plant, which are well beyond the initial efforts envisioned in the 1970s. These demands on PRA methods and tools have led to challenges in terms of further expansion of the use of PRA and raised the potential for future research into how to address these challenges to ensure continued nuclear safety. In addition to the PRA needs, the domain of computer science has led to advances in computational approaches that may serve to help nuclear power plant PRA applications. These newer technologies have great potential to improve the effectiveness and economics of PRA and are being considered for exploration of potential benefits to the PRA and nuclear-power communities.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2024

Entanglement Generation in Weakly Driven Arrays of Multilevel Atoms via Dipolar Interactions

We investigate the driven-dissipative dynamics of multilevel atomic arrays interacting via dipolar interactions at subwavelength spacings. Unlike two-level atoms in the weakly excited regime, multilevel atoms can become strongly entangled. Here, the entanglement manifests as the growth of spin waves in the ground-state manifold and survives after turning off the drive. We propose the 2.9 μ⁢m transition between 3 P 2 ↔ 3 D 3 in 88 Sr with 389 nm trapping light as a platform to test our predictions and explore many-body physics with light-matter interactions.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

DOE OSTI2021

Evaluating Spatial Accelerator Architectures with Tiled Matrix-Matrix Multiplication

There is a growing interest in custom spatial accelerators for machine learning applications. These accelerators employ a spatial array of processing elements (PEs) interacting via custom buffer hierarchies and networks-on-chip. The efficiency of these accelerators comes from employing optimized dataflow (i.e., spatial/temporal partitioning of data across the PEs and fine-grained scheduling) strategies to optimize data reuse. The focus of this work is to evaluate these accelerator architectures using a tiled general matrix-matrix multiplication (GEMM) kernel. To do so, we develop a framework that finds optimized mappings (dataflow and tile sizes) for a tiled GEMM for a given spatial accelerator and workload combination, leveraging an analytical cost model for runtime and energy. Our evaluations over five spatial accelerators demonstrate that the tiled GEMM mappings systematically generated by our framework achieve high performance on various GEMM workloads and accelerators.

43 PARTICLE ACCELERATORS↗

DOE OSTI2021

Revealing the Chemical Bonding in Adatom Arrays via Machine Learning of Hyperspectral Scanning Tunneling Spectroscopy Data

The adatom arrays on surfaces offer an ideal playground to explore the mechanisms of chemical bonding via changes in the local electronic tunneling spectra. While this information is readily available in hyperspectral scanning tunneling spectroscopy data, its analysis has been considerably impeded by a lack of suitable analytical tools. Here we develop a machine learning based workflow combining supervised feature identification in the spatial domain and unsupervised clustering in the energy domain to reveal the details of structure-dependent changes of the electronic structure in adatom arrays on the Co 3 Sn 2 S 2 cleaved surface. This approach, in combination with first-principles calculations, provides insight for using artificial neural networks to detect adatoms and classifies each based on their local neighborhood comprised of other adatoms. These structurally classified adatoms are further spectrally deconvolved. The unexpected inhomogeneity of electronic structures among adatoms in similar configurations is unveiled using this method, suggesting there is not a single atomic species of adatoms, but rather multiple types of adatoms on the Co 3 Sn 2 S 2 surface. This is further supported by a slight contrast difference in the images (or slight size variation) of the topography of the adatoms.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2026

Waveform Simulation Framework: User Manual with Tutorials

This manuscript describes the Waveform Simulation Framework (WSF), a Python-based framework that provides a unified, programmable interface for generating synthetic seismograms for applications such as seismic array design, method development, and special event analysis. WSF standardizes how users define sources, receivers, and velocity models while abstracting simulator-specific configuration details, enabling workflows that are largely independent of the underlying numerical engine. The document provides installation guidance and tutorial-driven examples for three WSF simulator wrappers—WSF PyFK, WSF SW4, and WSF SPECFEM2D—illustrating end-to-end workflows from forward waveform simulation to common post-processing tasks (e.g., visualization and backprojection) using consistent data products (e.g., ObsPy Stream objects and SAC files).

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2024

Field Programmable Gate Array-Based Reactor Protection Systems and Potential for Inclusion of Secure Elements to Improve Cybersecurity

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

DOE OSTI2020

High performance context switching for virtualized FPGA accelerators

A hardware context manager in a field-programmable gate array (FPGA) device includes configuration logic configured to program one or more programming regions in the FPGA device based on configuration data for implementing a target configuration of the one or more programming regions. Context management logic in the hardware context manager is coupled with the configuration logic and saves a first context corresponding to the target configuration by retrieving first state information from the set of one or more programming regions, where the first state information is generated based on the target configuration, and storing the retrieved first state information in a context memory. The context management logic restores the first context by transferring the first state information from the context memory to the one or more programming regions, and causing the configuration logic to program the one or more programming regions based on the configuration data.

42 ENGINEERING↗

DOE OSTI2024

Refactoring the elastic–viscous–plastic solver from the sea ice model CICE v6.5.1 for improved performance

This study focuses on the performance of the elastic–viscous–plastic (EVP) dynamical solver within the sea ice model, CICE v6.5.1. The study has been conducted in two steps. First, the standard EVP solver was extracted from CICE for experiments with refactored versions, which are used for performance testing. Second, one refactored version was integrated and tested in the full CICE model to demonstrate that the new algorithms do not significantly impact the physical results. The study reveals two dominant bottlenecks, namely (1) the number of Message Parsing Interface (MPI) and Open Multi-Processing (OpenMP) synchronization points required for halo exchanges during each time step combined with the irregular domain of active sea ice points and (2) the lack of single-instruction, multiple-data (SIMD) code generation. The standard EVP solver has been refactored based on two generic patterns. The first pattern exposes how general finite differences on masked multi-dimensional arrays can be expressed in order to produce significantly better code generation by changing the memory access pattern from random access to direct access. The second pattern takes an alternative approach to handle static grid properties. The measured single-core performance improvement is more than a factor of 5 compared to the standard implementation. The refactored implementation of strong scales on the Intel® Xeon® Scalable Processors series node until the available bandwidth of the node is used. For the Intel® Xeon® CPU Max series, there is sufficient bandwidth to allow the strong scaling to continue for all the cores on the node, resulting in a single-node improvement factor of 35 over the standard implementation. This study also demonstrates improved performance on GPU processors.

58 GEOSCIENCES↗

DOE OSTI2023

Cell invasion during competitive growth of polycrystalline solidification patterns

Abstract Spatially extended cellular and dendritic array structures forming during solidification processes such as casting, welding, or additive manufacturing are generally polycrystalline. Both the array structure within each grain and the larger scale grain structure determine the performance of many structural alloys. How those two structures coevolve during solidification remains poorly understood. By in situ observations of microgravity alloy solidification experiments onboard the International Space Station, we have discovered that individual cells from one grain can unexpectedly invade a nearby grain of different misorientation, either as a solitary cell or as rows of cells. This invasion process causes grains to interpenetrate each other and hence grain boundaries to adopt highly convoluted shapes. Those observations are reproduced by phase-field simulations further demonstrating that invasion occurs for a wide range of misorientations. Those results fundamentally change the traditional conceptualization of grains as distinct regions embedded in three-dimensional space.

36 MATERIALS SCIENCE↗