Search NASA⌕ Search

SEARCH · Search NASA

Results for “pruning”

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.

176 records · Page 10

hls4ml: An Open-Source Codesign Workflow to Empower Scientific Low-Power Machine Learning Devices

Accessible machine learning algorithms, software, and diagnostic tools for energy-efficient devices and systems are extremely valuable across a broad range of application domains. In scientific domains, real-time near-sensor processing can drastically improve experimental design and accelerate scientific discoveries. To support domain scientists, we have developed hls4ml, an open-source software-hardware codesign workflow to interpret and translate machine learning algorithms for implementation with both FPGA and ASIC technologies. We expand on previous hls4ml work by extending capabilities and techniques towards low-power implementations and increased usability: new Python APIs, quantization-aware pruning, end-to-end FPGA workflows, long pipeline kernels for low power, and new device backends include an ASIC workflow. Taken together, these and continued efforts in hls4ml will arm a new generation of domain scientists with accessible, efficient, and powerful tools for machine-learning-accelerated discovery.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Automated Shift Detection in Sensor-Based PV Power and Irradiance Time Series

PV power and irradiance sensor-based measurements are prone to error, resulting in issues such as abrupt time series data shifts. These shifts, which are usually unintentional, may be caused by software or hardware configuration changes on a PV system, and do not reflect an actual change in overall system performance. Locating these shifts and segmenting the associated time series aids in more accurate future PV analysis. In this research, an offline changepoint detection (CPD) algorithm that automatically detects these abrupt data shifts in sensor-based time series is introduced. Data shift periods in 101 daily PV power and irradiance time series were labeled manually by two solar experts. These data streams represent sensor-based measurements, and display a variety of data shift behaviors. A changepoint detection algorithm was tuned using the 101 labeled data streams, with each model configuration's ability to detect labeled changepoints benchmarked using metrics such as F1-score, recall, and Rand Index. Best performing models on seasonality-corrected data streams include the Pruned Exact Linear (PELT) method, the Binary Segmentation method, and the Bottom-Up method, all scoring an average F1-score of 0.76 or greater at detecting labeled changepoints within a 30-day window for the labeled data sets. To promote further research in this space, we are releasing the labeled data shift sets on U.S. Department of Energy's (DOE) DuraMAT Data Hub, and the associated algorithm in the Python PVAnalytics package.

changepoint detection↗

Automated Shift Detection in Sensor-Based PV Power and Irradiance Time Series

PV power and irradiance sensor-based measurements are prone to error, resulting in issues such as time series data shifts. In this research, a changepoint detection (CPD) algorithm that automatically detects data shifts in sensor-based time series is introduced. Data shift periods in 101 daily PV power and irradiance time series were labeled manually by two solar experts. These data streams represent sensor-based measurements, and display a variety of data shift behaviors. A changepoint detection algorithm was tuned using the 101 labeled data streams, with each model configuration's ability to detect labeled changepoints benchmarked using metrics such as F1-score, recall, and Rand Index. Best performing models on seasonality-corrected data streams include the Pruned Exact Linear (PELT) method, the Binary Segmentation method, and the Bottom-Up method, all scoring an average F1-score of 0.76 or greater at detecting labeled changepoints within a 30-day window across the labeled data sets. Pending approval, we plan to release the labeled data sets for this research on NREL's DuraMAT Data Hub, and the associated algorithm in the Python PVAnalytics package. By supplying the training sets and algorithm, we hope to encourage further development in this research space.

data shift↗

Union: A Unified HW-SW Co-Design Ecosystem in MLIR for Evaluating Tensor Operationson Spatial Accelerators

To meet the extreme compute demands for deep learning across commercial and scientific applications, dataflow accelerators are becoming increasingly popular. While these“domain-specific” accelerators are not fully programmable like CPUs and GPUs, they retain varying levels of flexibility with respect to data orchestration, i.e., dataflow and tiling optimizations to enhance efficiency. There are several challenges when designing new algorithms and mapping approaches to execute the algorithms for a target problem on new hardware. Previous works have addressed these challenges individually. To address this challenge as a whole, in this work, we present an HW-SW co-design ecosystem for spatial accelerators called Union within the popular MLIR compiler infrastructure. Our framework allows exploring different algorithms and their mappings on several accelerator cost models. Union also includes a plug-and-play library of accelerator cost models and mappers which can easily be extended. The algorithms and accelerator cost models are connected via a novel mapping abstraction that captures the map space of spatial accelerators which can be systematically pruned based on constraints from the hardware, workload, and mapper. We demonstrate the value of Union for the community with several case studies which examine offloading different tensor operations (CONV/GEMM/Tensor Contraction) on diverse accelerator architectures using different mapping schemes.

Jeong, Geonhwa↗

Conditional Tree Reduction in the ADAPT Framework

Probabilistic Risk Assessment (PRA) uses fault tree/event tree analysis to evaluate the environmental impacts of nuclear power plants due to internal or external initiating events. In a fault tree, basic failure events are assembled using primarily And/Or logic to determine the combinations of failures that lead to failure of the system as a whole. An event tree is forward-facing, and begins with a single initiating event. From there the analysis considers what event may occur next, and branches out among the possible configurations of occurrence and non-occurrence of that event. This branching continues until end states are reached. End states may include a safe and stable configuration of the plant, or any number of differing failure states. Both fault and event trees require basic event probabilities as inputs to provide insight on the likelihood of different outcomes. PRA has been applied to nuclear power plants as an analysis tool since the 1975 Reactor Safety Study. In a traditional PRA, the order of events is prescribed by the analyst and each event is typically a binary aleatory uncertainty: occurrence or non-occurrence. Discrete dynamic event tree (DDET) analysis eliminates this subjectivity in ordering of events by using the output of a dynamic system model (simulator) to inform the branching. Branching conditions are triggered by the existence of a relevant plant state in the code, and therefore only occur as physically appropriate. The DDET approach also allows consideration of both epistemic and aleatory uncertainties on a phenomenologically and stochastically consistent platform. A DDET may be orders of magnitude larger than a traditional event tree for the same initiating event, as more of the uncertainty space is likely to be explored and the tree does not require manual assembly. Because of the size, manual inspection of the entire DDET is often infeasible. This problem may be approached from the front by reducing the number of sampled branching values or by online pruning of the tree, thus reducing the total number of branches. Utilizing post-processing techniques to draw actionable insights via clustering has also been proposed. One post-processing tool to reduce the DDET according to user-defined rules has been recently implemented in the Analysis of Dynamic Accident Progression Trees (ADAPT) DDET driver code. The tool creates a 'slice' of the DDET comprised of the set of branches that meet a set of user-input rules, for immediate visualization or for further calculations conditional on the rules being met. This extension is expected to ease analysis of the results of large and complex DDETs, leading to greater applicability of DDETs in PRA. It should be emphasized that this tool is a visualization aide and does not impact the theoretical basis of the DDET approach. (authors)

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Preliminary Cyber-Informed Dynamic Branch Conditions for Analysis with the Dynamic Simplified Cyber MELCOR Model

Instrumentation and control of nuclear power is transforming from analog to modern digital assets. These control systems perform key safety and security functions. This transformation is occurring in new plant designs as well as in the existing fleet of plants as the operation of those plants is extended to 60 years. This transformation introduces new and unknown exploits involving both digital asset induced safety issues and security exploits. Traditional the open literature of nuclear power risk assessment tools and cyber security assessment methods have not been modified or developed to address the unique nature of cyber failure modes and of cyber security threats. Starting in 2015, Sandia National Laboratories (SNL) embarked on a two year dynamic cyber-risk informed project to facilitate the analysis of unique cyber failure modes and the time sequencing of cyber faults, both malicious and non-malicious, and impose those cyber exploits and cyber faults onto a nuclear power plant accident sequence simulator code to assess how cyber exploits and cyber faults could interact with a plant's digital instrumentation and control (DI and C) system in the absence of digital controls. This will be achieved by coupling SNL's existing dynamic scheduler code (ADAPT), nuclear accident simulator code (MELCOR), and cyber Emulytics code (SCEPTRE) to demonstrate real-time simulation of cyber exploits and their impact on automatic DI and C responses. Studying such potential time-sequenced cyber-attacks and their risks (i.e., the associated impact and the associated degree of difficulty to achieve the attack vector) as they pertain to accident management will provide a risk informed basis for developing effective cyber security controls for nuclear power. This paper will discuss the rationale behind the proposed generic cyber exploits for an abstracted Pressurized Water Reactor that has adopted a unified DI and C System. Other papers related to this project submitted to this conference discuss: - The MELCOR model for this abstracted reactor, called the Dynamic Simplified Cyber (DSC) model, - Modifications to ADAPT to: Extend the framework for multiple simulators, Incorporation of importance measures into dynamic event tree analysis, Sort and reduce the dynamic event tree based upon inputs to or outputs from the tree, and - A feasibility study for using density based clustering methods with dynamic time wrapping to cluster and prune dynamic event trees. (authors)

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

8. Open-Loop Toolpath Generation

Open-loop paths, as opposed to closed-loop paths, have start and end points that are at different locations. Open-loop paths, commonly called “skeletons,” can be generated through multiple approaches. Most approaches utilize a shape's medial axis transform and build appropriate pathing around this representation. In this chapter, a specific implementation consisting of three major steps will be discussed. These steps are Voronoi diagram generation, skeleton pruning, and path consolidation. With the major components of the implementation covered, two additional optimizations will also be described. These two optimizations, chamfering and smoothing, will be highlighted as ways to improve overall pathing quality.

Borish, Michael↗

Neural Image Compression: Generalization, Robustness, and Spectral Biases

Recent advances in neural image compression (NIC) have resulted in models which are starting to outperform traditional codecs. While this has led to growing excitement about using these methods in real-world applications, the successful adoption of any machine learning system (including NIC) in the wild requires it to generalize (and be robust) to unseen distribution shifts at deployment time. Unfortunately, current research lacks comprehensive datasets and informative tools to evaluate and understand compression performance in real-world settings. To bridge this crucial gap, first, this paper presents a comprehensive benchmark suite to evaluate the out-of-distribution (OOD) performance of image compression methods. Specifically, we design CLIC-C and Kodak-C by introducing 15 common corruptions to popular CLIC and Kodak benchmarks. Next, we propose spectrally inspired introspection tools to gain a deeper understanding of errors introduced by image compression methods as well as their OOD performance. To this end, we carry out a detailed performance comparison of the classical codec with various variants of NIC (e.g., original, variable rate, pruned), revealing intriguing findings that challenge our current understanding of the strengths and limitations of NIC. Finally, we corroborate our empirical findings with theoretical analysis, providing an in-depth view of the OOD performance of NIC. Our benchmarks, spectral introspection tools, and findings provide a crucial bridge to the real-world adoption of NIC. We hope that our work will propel future efforts in designing more robust and generalizable NIC methods.

neural networks, Variational Autoencoder, robustne↗

autoGEMM: Pushing the Limits of Irregular Matrix Multiplication on Arm Architectures

This paper presents an open-source library that pushes the limits of performance portability for irregular General Matrix Multiplication (GEMM) on the widely-used Arm architectures. Our library, autoGEMM, is designed to support a wide range of Arm processors: from edge devices to HPC-grade CPUs. autoGEMM generates optimized kernels for various hardware configurations by auto-combining fragments of autogenerated micro-kernels that employ hand-written optimizations to maximize computational efficiency. We optimize the kernel pipeline by tuning the register reuse and the data load/store overlapping. In addition, we use a dynamic tiling scheme to generate balanced tile shapes. Finally, we position autoGEMM on top of the TVM framework where our dynamic tiling scheme prunes the search space for TVM to identify the optimal combination of parameters for code optimization. Evaluations on five different classes of Arm chips demonstrate the advantages of autoGEMM. For small matrices, autoGEMM achieves 98% of peak and up to 2.0x speedup over state-of-the-art libraries such as LIBXSMM and LibShalom. For irregular matrices (i.e. tall skinny and long rectangles), autoGEMM is 1.3-2.0x faster than widely-used libraries such as OpenBLAS and Eigen. autoGEMM is publicly available at: https://github.com/wudu98/autoGEMM.

Wu, Du↗

MoE-Inference-Bench: Performance Evaluation of Mixture of Expert Large Language and Vision Models

Mixture of Experts (MoE) models have enabled the scaling of Large Language Models (LLMs) and Vision Language Models (VLMs) by achieving massive parameter counts while maintaining computational efficiency. However, MoEs introduce several inference-time challenges, including load imbalance across experts and the additional routing computational overhead. To address these challenges and fully harness the benefits of MoE, a systematic evaluation of hardware acceleration techniques is essential. We present MoE-Inference-Bench, a comprehensive study to evaluate MoE performance across diverse scenarios. We analyze the impact of batch size, sequence length, and critical MoE hyperparameters such as FFN dimensions and number of experts on throughput. We evaluate several optimization techniques on Nvidia H100 GPUs, including pruning, Fused MoE operations, speculative decoding, quantization, and various parallelization strategies. Our evaluation includes MoEs from the Mixtral, DeepSeek, OLMoE and Qwen families. The results reveal performance differences across configurations and provide insights for the efficient deployment of MoEs.

Chitty-Venkata, Krishna Teja↗

Surrogate Neural Architecture Codesign Package (SNAC-Pack)

Neural architecture search (NAS) is a powerful approach for automating model design, but existing methods often optimize for accuracy alone or rely on proxy metrics such as bit operations (BOPs) that correlate poorly with hardware cost. This gap is particularly large for FPGA deployment, where cost is dominated by a multi-dimensional budget of lookup tables, DSPs, flip-flops, BRAM, and latency. We present the Surrogate Neural Architecture Codesign Package (SNAC-Pack), an open-source AutoML framework for hardware-aware neural architecture codesign and end-to-end FPGA deployment. SNAC-Pack runs a multi-objective global search with Optuna and NSGA-II, loading trials to a shared SQLite store that enables parallel workers across compute nodes. A hardware surrogate model outputs per-trial resource and latency estimates, avoiding the synthesis cost that would otherwise dominate the search loop. A local search stage then applies quantization-aware training (QAT) together with iterative magnitude pruning in a combined compression loop, after which the final model is synthesized to FPGA firmware via the hls4ml Python library. A YAML configuration and an optional agentic frontend let users run the pipeline on new datasets without modifying the framework. We demonstrate SNAC-Pack on jet classification at the Large Hadron Collider and superconducting qubit readout, discovering compact architectures that match or exceed strong baselines on the task metric while reducing FPGA resource utilization and, in the qubit readout case, reducing the design space exploration process from months of manual fine-tuning to hours of automated search.

Weitz, Jason [UC, San Diego]↗

Automated detection and analysis of particle beams in laser-plasma accelerator simulations

Numerical simulations of laser-plasma wakefield (particle) accelerators model the acceleration of electrons trapped in plasma oscillations (wakes) left behind when an intense laser pulse propagates through the plasma. The goal of these simulations is to better understand the process involved in plasma wake generation and how electrons are trapped and accelerated by the wake. Understanding of such accelerators, and their development, offer high accelerating gradients, potentially reducing size and cost of new accelerators. One operating regime of interest is where a trapped subset of electrons loads the wake and forms an isolated group of accelerated particles with low spread in momentum and position, desirable characteristics for many applications. The electrons trapped in the wake may be accelerated to high energies, the plasma gradient in the wake reaching up to a gigaelectronvolt per centimeter. High-energy electron accelerators power intense X-ray radiation to terahertz sources, and are used in many applications including medical radiotherapy and imaging. To extract information from the simulation about the quality of the beam, a typical approach is to examine plots of the entire dataset, visually determining the adequate parameters necessary to select a subset of particles, which is then further analyzed. This procedure requires laborious examination of massive data sets over many time steps using several plots, a routine that is unfeasible for large data collections. Demand for automated analysis is growing along with the volume and size of simulations. Current 2D LWFA simulation datasets are typically between 1GB and 100GB in size, but simulations in 3D are of the order of TBs. The increase in the number of datasets and dataset sizes leads to a need for automatic routines to recognize particle patterns as particle bunches (beam of electrons) for subsequent analysis. Because of the growth in dataset size, the application of machine learning techniques for scientific data mining is increasingly considered. In plasma simulations, Bagherjeiran et al. presented a comprehensive report on applying graph-based techniques for orbit classification. They used the KAM classifier to label points and components in single and multiple orbits. Love et al. conducted an image space analysis of coherent structures in plasma simulations. They used a number of segmentation and region-growing techniques to isolate regions of interest in orbit plots. Both approaches analyzed particle accelerator data, targeting the system dynamics in terms of particle orbits. However, they did not address particle dynamics as a function of time or inspected the behavior of bunches of particles. Ruebel et al. addressed the visual analysis of massive laser wakefield acceleration (LWFA) simulation data using interactive procedures to query the data. Sophisticated visualization tools were provided to inspect the data manually. Ruebel et al. have integrated these tools to the visualization and analysis system VisIt, in addition to utilizing efficient data management based on HDF5, H5Part, and the index/query tool FastBit. In Ruebel et al. proposed automatic beam path analysis using a suite of methods to classify particles in simulation data and to analyze their temporal evolution. To enable researchers to accurately define particle beams, the method computes a set of measures based on the path of particles relative to the distance of the particles to a beam. To achieve good performance, this framework uses an analysis pipeline designed to quickly reduce the amount of data that needs to be considered in the actual path distance computation. As part of this process, region-growing methods are utilized to detect particle bunches at single time steps. Efficient data reduction is essential to enable automated analysis of large data sets as described in the next section, where data reduction methods are steered to the particular requirements of our clustering analysis. Previously, we have described the application of a set of algorithms to automate the data analysis and classification of particle beams in the LWFA simulation data, identifying locations with high density of high energy particles. These algorithms detected high density locations (nodes) in each time step, i.e. maximum points on the particle distribution for only one spatial variable. Each node was correlated to a node in previous or later time steps by linking these nodes according to a pruned minimum spanning tree (PMST). We call the PMST representation 'a lifetime diagram', which is a graphical tool to show temporal information of high dense groups of particles in the longitudinal direction for the time series. Electron bunch compactness was described by another step of the processing, designed to partition each time step, using fuzzy clustering, into a fixed number of clusters.

43 PARTICLE ACCELERATORS↗

Directed aging, memory, and nature’s greed

Disordered materials are often out of equilibrium and evolve very slowly in a rugged and tortuous energy landscape. This slow evolution, referred to as aging, is deemed undesirable as it often leads to material degradation. However, we show that aging also encodes a memory of the stresses imposed during preparation. Because of inhomogeneous local stresses, the material itself decides how to evolve by modifying stressed regions differently from those under less stress. Because material evolution occurs in response to stresses, aging can be “directed” to produce sought-after responses and unusual functionalities that do not inherently exist. Aging obeys a natural “greedy algorithm” as, at each instant, the material simply follows the path of most rapid and accessible relaxation. Our experiments and simulations illustrate directed aging in examples in which the material’s elasticity transforms as desired because of an imposed deformation.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Cacao Intensification in Sulawesi: A Green Prosperity Model Project

NREL conducted eight model projects for Millennium Challenge Corporation's (MCC) Compact with Indonesia. Green Prosperity, the largest project of the Compact, seeks to address critical constraints to economic growth while supporting the Government of Indonesia's commitment to a more sustainable, less carbon-intensive future. This study evaluates techniques to improve cacao farming in Sulawesi Indonesia with an emphasis on Farmer Field Schools and Cocoa Development Centers to educate farmers and for train the trainer programs. The study estimates the economic viability of cacao farming if smallholder implement techniques to increase yield as well as social and environmental impacts of the project.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗