Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “DATA BASE MANAGEMENT”

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 775 records · Page 43
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Models, data, and scripts associated with “Prediction of Distributed River Sediment Respiration Rates using Community-Generated Data and Machine Learning”

This data package is associated with the publication “Prediction of Distributed River Sediment Respiration Rates using Community-Generated Data and Machine Learning’’ submitted to the Journal of Geophysical Research: Machine Learning and Computation (Scheibe et al. 2024). River sediment respiration observations are expensive and labor intensive to obtain and there is no physical model for predicting this quantity. The Worldwide Hydrobiogeochemisty Observation Network for Dynamic River Systems (WHONDRS) observational data set (Goldman et al.; 2020) is used to train machine learning (ML) models to predict respiration rates at unsampled sites. This repository archives training data, ML models, predictions, and model evaluation results for the purposes of reproducibility of the results in the associated manuscript and community reuse of the ML models trained in this project. One of the key challenges in this work was to find an optimum configuration for machine learning models to work with this feature-rich (i.e. 100+ possible input variables) data set. Here, we used a two-tiered approach to managing the analysis of this complex data set: 1) a stacked ensemble of ML models that can automatically optimize hyperparameters to accelerate the process of model selection and tuning and 2) feature permutation importance to iteratively select the most important features (i.e. inputs) to the ML models. The major elements of this ML workflow are modular, portable, open, and cloud-based, thus making this implementation a potential template for other applications. This data package is associated with the GitHub repository found at Please see the file level metadata (flmd; “sl-archive-whondrs_flmd.csv”) for a list of all files contained in this data package and descriptions for each. Please see the data dictionary (dd; “sl-archive-whondrs_dd.csv”) for a list of all column headers contained within comma separated value (csv) files in this data package and descriptions for each. The GitHub repository is organized into five top-level directories: (1) “input_data” holds the training data for the ML models; (2) “ml_models” holds machine learning models trained on the data in “input_data”; (3) “scripts” contains data preprocessing and postprocessing scripts and intermediate results specific to this data set that bookend the ML workflow; (4) “examples” contains the visualization of the results in this repository including plotting scripts for the manuscript (e.g., model evaluation, FPI results) and scripts for running predictions with the ML models (i.e., reusing the trained ML models); (5) “output_data” holds the overall results of the ML model on that branch. Each trained ML model resides on its own branch in the repository; this means that inputs and outputs can be different branch-to-branch. Furthermore, depending on the number of features used to train the ML models, the preprocessing and postprocessing scripts, and their intermediate results, can also be different branch-to-branch. The “main-*” branches are meant to be starting points (i.e. trunks) for each model branch (i.e. sprouts). Please see the Branch Navigation section in the top-level README.md in the GitHub repository for more details. There is also one hidden directory “.github/workflows”. This hidden directory contains information for how to run the ML workflow as an end-to-end automated GitHub Action but it is not needed for reusing the ML models archived here. Please the top-level README.md in the GitHub repository for more details on the automation.

13C↗
DOE OSTI2021

Cardinality-Regularized Hawkes-Granger Model

We propose a new sparse Granger-causal learning framework for temporal event data. We focus on a specific class of point processes called the Hawkes process. We begin by pointing out that most of the existing sparse causal learning algorithms for the Hawkes process suffer from a singularity in maximum likelihood estimation. As a result, their sparse solutions can appear only as numerical artifacts. In this paper, we propose a mathematically well-defined sparse causal learning framework based on a cardinality-regularized Hawkes process, which remedies the pathological issues of existing approaches. We leverage the proposed algorithm for the task of instance-wise causal event analysis, where sparsity plays a critical role. We validate the proposed framework with two real use-cases, one from the power grid and the other from the cloud data center management domain.

Ide, Tsuyoshi↗
DOE OSTI2021

Subsurface Characterization and Machine Learning Predictions at Brady Hot Springs: Preprint

Subsurface data analysis, reservoir modeling, and machine learning (ML) techniques have been applied to the Brady Hot Springs (BHS) geothermal field in Nevada, USA to further characterize the subsurface and assist with optimizing reservoir management. Hundreds of reservoir simulations have been conducted in TETRAD-G and CMG STARS to explore different injection and production fluid flow rates and allocations and to develop a training data set for ML. This process included simulating the historical injection and production since 1979 and prediction of future performance through 2040. ML networks were created and trained using TensorFlow based on multilayer perceptron (MLP), long short-term memory (LSTM), and convolutional neural network (CNN) architectures. These networks took as input selected flow rates, injection temperatures, and historical field operation data and produced estimates of future production temperatures. This approach was first successfully tested on a simplified single fracture doublet system, followed by the application to the BHS reservoir. Using an initial BHS dataset with 37 simulated scenarios, the trained and validated network predicted the production temperature for 6 production wells with the mean absolute percentage error of less than 8%. In a complementary analysis effort, the principal component analysis applied to 13 BHS geological parameters revealed that vertical fracture permeability shows the strongest correlation with fault density and fault intersection density. A new BHS reservoir model was developed considering the fault intersection density as proxy for permeability. This new reservoir model helps to explore under-exploited zones in the reservoir. A data gathering plan to obtain additional subsurface data was developed; it includes temperature surveying for three idle injection wells, at which the reservoir simulations indicate high bottom-hole temperatures. The collected data assist with calibrating the reservoir model and may lead to converting these wells to producers to access under-exploited zones in the reservoir. Data gathering activities are planned for the first quarter of 2021.

40 EE - Geothermal Technologies Office (EE-4G)↗
DOE OSTI2020

Energy Intensity Baselining and Tracking Guidance

Each company joining the U.S. Department of Energy’s (DOE’s) Better Buildings, Better Plants Program (Better Plants) commits to establishing an energy consumption and energy intensity (EI) baseline and to tracking its energy performance over a 10-year period against that baseline. The baseline must reflect a company’s energy consumption over a 12-month period, covering all its U.S.-based operations. Energy consumption is calculated by fuel type in terms of primary energy (also known as source energy). EI is broadly defined as the amount of energy consumed per unit of output produced. For this guidance document and for the program, the term energy performance represents an evaluation of a facility’s capacity to use energy efficiently. Metrics used to assess a facility’s energy performance can include EI, energy consumption, improvements in EI, etc. Establishing an energy baseline and tracking system is a critical first step in effectively managing energy use. Developing a baseline can help a company understand energy use within the corporation and give it a point of comparison to evaluate future efforts to improve energy performance. It can also support efforts to validate a company’s energy management activities, improve comparative analyses when using benchmarks, and help in predicting future energy needs. In addition, a company that normalizes its performance data can determine highly defensible measures of energy savings generated through implemented energy efficiency projects. Establishing a baseline and tracking energy performance is also a requirement for ISO 50001 certification. Although basic energy data can be collected through utility bills, most manufacturers will have to perform additional analyses to develop accurate and robust energy baselines and tracking systems. Energy is consumed in many ways within the manufacturing sector and can come from multiple sources. Energy is sometimes generated and sold to other parties or captured and reused on-site. External events can exert a significant impact on a facility or company’s energy use independent of any purposeful efforts to improve energy efficiency. Operational changes, such as production shifts—which may be inevitable for some companies over the 10-year period covered by the program—can also make a big difference in energy use. Since Better Plants asks companies to account for all their U.S.-based operations, mergers, acquisitions, and divestitures can also have significant implications for a company’s energy metrics. This document aims to demystify the sometimes complex baselining process. It devotes special attention to the task of normalizing and adjusting energy consumption to account for external factors, such as weather and production changes. A key recommendation is that companies use regression analysis to normalize their energy consumption data whenever possible. Regression analysis is a statistical technique that estimates the dependence of a variable (i.e., energy use in the context of Better Plants) on one or more independent variables such as ambient temperature, while controlling for the influence of other variables at the same time. A properly developed regression analysis can provide a reliable estimate of energy savings resulting from energy improvement strategies and projects by accounting for the effects of variables such as annual production levels and weather. DOE has developed a companion Energy Performance Indicator software tool (EnPI) to simplify the baselining process. This tool can run regression models, calculate changes in EI at the facility level, and automatically compile facility-level data into a corporate-wide metric. Note that although the relevant equations used to calculate EI are provided in this document, the EnPI tool will automatically perform most calculations for the user. Additionally, Better Plants Partners (Partners) can call on their Technical Account Manager (TAM) to help them establish a baseline and assist with the necessary calculations to track progress.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2024

Integrating multi-modal remote sensing, deep learning, and attention mechanisms for yield prediction in plant breeding experiments

In both plant breeding and crop management, interpretability plays a crucial role in instilling trust in AI-driven approaches and enabling the provision of actionable insights. The primary objective of this research is to explore and evaluate the potential contributions of deep learning network architectures that employ stacked LSTM for end-of-season maize grain yield prediction. A secondary aim is to expand the capabilities of these networks by adapting them to better accommodate and leverage the multi-modality properties of remote sensing data. In this study, a multi-modal deep learning architecture that assimilates inputs from heterogeneous data streams, including high-resolution hyperspectral imagery, LiDAR point clouds, and environmental data, is proposed to forecast maize crop yields. The architecture includes attention mechanisms that assign varying levels of importance to different modalities and temporal features that, reflect the dynamics of plant growth and environmental interactions. The interpretability of the attention weights is investigated in multi-modal networks that seek to both improve predictions and attribute crop yield outcomes to genetic and environmental variables. This approach also contributes to increased interpretability of the model's predictions. The temporal attention weight distributions highlighted relevant factors and critical growth stages that contribute to the predictions. The results of this study affirm that the attention weights are consistent with recognized biological growth stages, thereby substantiating the network's capability to learn biologically interpretable features. Accuracies of the model's predictions of yield ranged from 0.82-0.93 R 2 ref in this genetics-focused study, further highlighting the potential of attention-based models. Further, this research facilitates understanding of how multi-modality remote sensing aligns with the physiological stages of maize. The proposed architecture shows promise in improving predictions and offering interpretable insights into the factors affecting maize crop yields, while demonstrating the impact of data collection by different modalities through the growing season. By identifying relevant factors and critical growth stages, the model's attention weights provide valuable information that can be used in both plant breeding and crop management. The consistency of attention weights with biological growth stages reinforces the potential of deep learning networks in agricultural applications, particularly in leveraging remote sensing data for yield prediction. To the best of our knowledge, this is the first study that investigates the use of hyperspectral and LiDAR UAV time series data for explaining/interpreting plant growth stages within deep learning networks and forecasting plot-level maize grain yield using late fusion modalities with attention mechanisms.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2024

Water Resource Assessment In The New Mexico Permian Basin: BLM 2023 Assessment Report

The Permian Basin is the highest producing oil field in the United States and is comprised of three component basins including; the Midland Basin, Delaware Basin and the Marfa Basin. This report describes the work conducted by Sandia National Laboratories (SNL) for the Bureau of Land Management (BLM) to investigate the occurrence of usable water (quality, and depth to water) in the Delaware Sub-basin of the Permian Basin. High Production Areas (HPAs) were identified for the region based on Reasonable Foreseeable Development Scenario (RFD) published by New Mexico Tech University. HPAs were established based on potential for future development of oil reserves. The study was initiated by the BLM-Carlsbad Field Office (CFO) based on concerns that special protections for groundwater in these HPAs may be warranted. A combination of analysis of existing data and field work to collect new data for analysis were used to complete the investigation objectives. This study summarizes the most recent analyses in an ongoing project and builds on previously completed work as listed in Table 1. Advancements in directional drilling and well completion technologies have resulted in an exponential growth in the use of hydraulic fracturing for oil and gas extraction in the Permian Basin. Within the New Mexico portion of the Delaware Sub-basin, water demand to complete each hydraulically fractured well is estimated to average 7.3 acre-feet (2.4 million gallons), resulting in 30 billion barrels of water over the life of the plan or 1.5 billion barrels per year. This rising demand is creating concern for the regions ability to provide the necessary water in a manner that fulfills BLM’s role of protecting human health and the environment while sustainably meeting the needs of various water users in the region. This report documents water-level and water chemistry baselines to aid the BLM in understanding the regional water supply dynamics under various management, policy, and growth scenarios and to pre-emptively identify risks to water sustainability.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2020

Southeast Regional CO 2 Utilization and Storage Acceleration Partnership (SECARB-USA): Preliminary Regional Assessment of Potential Storage Complexes, Including Capacities and Costs, Based on Application of NRAP and SCO2T (Deliverable 3.3) Storage Complexes

The Recipient will assist with the field validation of one or more risk assessment toolsets through: 1) collaborative research; 2) sharing of relevant datasets, information, and technical insights from field efforts; and 3) other modeling and simulation toolsets as they are developed. Collaborative research could include, but is not limited to: 1) demonstrating use of field data for improved leakage pathway characterization, uncertainty reduction, and risk management and uncertainty workflows; 2) testing and validation of a specific National Risk Assessment Partnership (NRAP) tool using site specific data; and 3) improved modeling or reduced-order modeling of geophysical monitoring.

42 ENGINEERING↗
DOE OSTI2023

Adaptive cache management based on programming model information

A processing system selectively allocates space to store a group of one or more cache lines at a cache level of a cache hierarchy having a plurality of cache levels based on memory access patterns of a software application executing at the processing system. The processing system generates bit vectors indicating which cache levels are to allocate space to store groups of one or more cache lines based on the memory access patterns, which are derived from data granularity and movement information. Based on the bit vectors, the processing system provides hints to the cache hierarchy indicating the lowest cache level that can exploit the reuse potential for a particular data.

Na, Weon Taek↗
DOE OSTI2020

Prototype energy models for data centers

Data centers in the United States consume about two percent of the nation’s electricity. Because heat gains from IT equipment drive cooling demand, data centers offer unique opportunities for energy savings. However, no prototype energy model for data centers is available in the suite of existing U.S. Department of Energy’s Commercial Prototype Building Models. Here we present the development of two new data center prototype models and their implementation in OpenStudio and EnergyPlus. The small-size data center model represents a computer room in a building served by computer room air conditioners (CRACs); while the large-sized model represents stand-alone data centers served by computer room air handlers (CRAHs) with a central chiller plant. For each data center model, two levels of IT equipment (ITE) load density were considered, to cover the wide range of IT power density of data centers: 40 and 100 W/ft 2 (430 and 1076 W/m 2 ) for the computer room, and 100 and 500 W/ft 2 (1076 and 5382 W/m 2 ) for the stand-alone data center. All other assumptions, such as building envelope, lighting, HVAC efficiencies and schedules, were based on the minimal requirements of ASHRAE Standard 90.1 at various vintages. We introduced a novel concept of supply and return air approach temperatures to capture the essential effects of non-uniform airflow and temperature distribution in data centers. The approach temperatures were pre-computed by computational fluid dynamics (CFD) simulations for various configurations of ITE loads and airflow containment management in data centers. A new feature was developed in EnergyPlus to implement the approach temperature method. A case study was conducted to demonstrate the use of the data center models. The two data center models cover all U.S. climate zones and can be used to evaluate energy saving measures for data centers, as well as to support development of data center energy efficiency codes and standards.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2023

Caribbean deepwater snappers: Application of the bomb radiocarbon age estimation validation in understanding aspects of ecology and life history

Reef fishes have been utilized as food fish throughout the U.S. Caribbean and Gulf of Mexico waters for centuries, with increasing fishing effort in recent decades. As a result, many species have experienced declines in landings, including deepwater snappers such as queen snapper Etelis oculatus and blackfin snapper Lutjanus buccanella . However, little to no peer-reviewed published research exists on basic life history parameters for either species. Confirming the accuracy of an age estimation method for a fish species is essential for ensuring sustainable fisheries management. This is because in the assessment of fisheries species population age-based parameters, including longevity, age at sexual maturity, growth rate, mortality, age-specific reproductive output, and lifetime reproductive output, are important in understanding overall life history strategies of managed stocks. The past stock assessment on U.S. Caribbean queen snapper utilized an estimated longevity of 8 y, derived from length frequencies for fish from St. Lucia. Blackfin snapper has an estimated longevity of 27 y based on a relatively small study from offshore waters of the southeastern U.S. The focus of our investigation was to estimate maximum longevity of two data-poor species in the U.S. Caribbean. The accuracy of ageing methods was tested via bomb radiocarbon age estimation validation and effects of depth on Δ 14 C in otolith cores and eye lens core values were examined. Results from our work indicate a maximum validated age of 45 y for queen snapper, and 43 y for blackfin snapper. Our findings indicate queen snapper and blackfin snapper are long-lived (> 40 y). The resulting Δ 14 C comparison between eye lens cores and otolith cores has important implications for the study of age validation, specifically when deepwater species are involved.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2022

EJFAT: Towards Intelligent Compute Destination Load Balancing

To handle increased data flow, Jefferson Lab (JLab) is partnering with ESnet for development of an AI/ML directed compute work Load Balancer (LB) of UDP streamed data. The LB is FPGA based featuring dynamically configurable, low latency and high throughput destination address switching. The LB provides integration of edge and core computing to support JLab experimental programs, the Electron-Ion Collider, as well as data centers of the future. In the ESnet/JLab FPGA Accelerated Transport (EJFAT) initiative, the function of the LB Data Plane (DP) is to redirect data streams to selectable (but unknown to sender) destination hosts based on current worload and within that host to destination ports as a function of sub- stream id. This effects hierarchical scaling, first across compute machines for processing over a series of events and second, across ports so different data source sub-streams may be assigned to different processors for further parallelization. The LB Control Plane (CP) programs the DP using compute farm telemetry to direct and balance workloads across a compute cluster as the operating conditions require. While Proportional/Integrative/Derivative (PID) controllers are often seen in similar applications, here we investigate the feasibility of a Reinforcement Learning (RL) based schedule manager running in the CP to provide dynamic updates to the DP scheduling policy.

Lawrence, David↗
DOE OSTI2024

Radio Frequency Spectrum Audit to Inventory Private Cellular Base Station Infrastructure

The ever-changing cellular communication landscape makes it difficult to identify, map, and localize cellular base stations. Localizing cellular base stations provides various advantages, including information security, cybersecurity, spectrum management, and interference detection. For example, the MITRE ATT&CK® (Adversarial Tactics, Techniques, and Common Knowledge architecture) [1] and Common Attack Pattern Enumeration and Classification [2] emphasize the importance of being able to minimize the cyber security threat presented by unregulated private cellular base stations (PCBS). The majority of published research looks at the malicious use of PCBSs and focuses on using data retrieved from user equipment (UE), data obtained from an application on the UE, or data shared between the UE and a mobile network to locate it. This innovative strategy, however, focuses on the passively discovered uniqueness of radio frequency (RF) transmissions from commercial cellular infrastructure received in a designated monitoring position (DMP).

42 ENGINEERING↗
DOE OSTI2022

Energy Emergency and Preparedness Data: Frequently Asked Questions (FAQs) and Quick Guidance on Crisis Communications

State Energy Offices and Public Utility Commissions rely on timely, accurate, and actionable information to perform their energy emergency response duties and execute their roles as state energy security planners. In support of this need, the National Association of State Energy Officials (NASEO) and the National Association of Regulatory Utility Commissions (NARUC) hosted an Energy Security and Data Analysis Workshop in Washington, DC to identify energy security response and planning data sources; and to share successful methods of data use and integration in state, federal, and private sector tools. Following the workshop, NASEO and NARUC hosted two topical data-centric webinars based on state priorities to identify best practices in Geographic Information Systems (GIS) and Crisis Communications programs leveraged in energy assurance planning and response. The Crisis Communications webinar covered best practice tactics for how states can respond during energy emergencies and other crises. Based on the workshop and webinars, this document summarizes commonly used data sources and includes frequently asked questions which may be used by state energy officials (i.e., consisting of staff from Public Utility Commissions and Governor-designated State Energy Offices) to help guide them in developing or improving their Crisis Communications capabilities. A strong public information program is a key crisis management tool. Timely and accurate information helps prevent confusion and uncertainty and encourages public support and cooperation. As energy subject matter experts, state energy officials are vital in the distilling, clarifying, and conveying energy sector information and implications to decision-makers and the public. Other participants in an effective public information program include the Governor’s Office, other state agencies, local governments, energy providers, local businesses, state legislature, and the federal government. It is essential to provide stakeholders and the public with information about the nature, severity, and duration of an emergency because inadequate understanding and awareness can lead to undesirable actions that could further exacerbate the situation. Before a state government can provide information to the public, it must gather information, describe the emergency accurately, and develop recommendations to manage the situation.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗
DOE OSTI2026

AmeriFlux FLUXNET-1F US-RC2 Cook Agronomy Farm - Conventional Till

This is the AmeriFlux Management Project (AMP) created FLUXNET-1F version of the carbon flux data for the site US-RC2 Cook Agronomy Farm - Conventional Till. This is the FLUXNET version of the carbon flux data for the site US-RC2 Cook Agronomy Farm - Conventional Till produced by applying the standard ONEFlux (1F) software. Site Description - RC2 operated from 2013-2016 at the R.J. Cook Agronomy Farm, as part of a cluster of 5 towers (RC1 to RC5) operated for the Regional Approaches to Climate Change (REACCH) USDA-supported research project. The tower predates the Longterm Agroecosystem Research (LTAR) site common experiment, which was established in nearby fields at the Cook Agronomy Farm in 2017. Cook Agronomy Farm is in the high-precipitation agroecological zone of the Columbia Plateau’s dryland cropping region. Wheat-based crop rotations are grown annually. RC2 had conventional tillage management (reduced-till) since at least 1998. It was contrasted with RC1, which had no-till management over the same time period. RC2 captured the same tillage practices as the US-CF2 site established in 2017 as part of LTAR common experiment. However, the towers have distinct footprints, aspects, and soil series composition.

Chi, Jinshu [The Hong Kong University of Science a↗
DOE OSTI2022

NPD Classification Tools – User Guide NPD Explorer and NPDamCAT Apps

The existing infrastructure at non-powered dams (NPDs) presents a variety of opportunities from generating electricity and economic value to the myriad services they provide. However, it also presents a significant challenge because aging structures must be maintained, and changes to the ecosystems and river systems by NPDs must be managed. The various stakeholders interested in these opportunities and challenges require varying information about NPDs; in many cases, interest in NPDs can extend across the entire population of dams. Even when interest is more narrowly focused on an individual dam or a small subset of dams, understanding how these dams relate to the broader context of NPD infrastructure can be important. As noted in TM 2021/2155, “Each NPD has unique characteristics describing its design, operation, environmental impacts, social impacts, and economic potential. The large number of dams, the diversity of interests related to dams, the variety of dam characteristics, and the types of data required to describe dams all pose major challenges to an analysis of the entire dam population." This user guide describes two web-based tools that facilitate exploration of dams from a variety of perspectives: the NPD Explorer and the NPD Custom Analysis and Taxonomy (NPDamCAT). These tools facilitate access to information about dams and help users interact with the information, making small- to large-scale analyses more convenient for a broad set of stakeholders.

13 HYDRO ENERGY↗
DOE OSTI2020

Projecting spatiotemporally explicit effects of climate change on stream temperature: A model comparison and implications for coldwater fishes

Conservation planners and resource managers seek information about how the availability and locations of cold-water habitats will change in the future and how these predictions vary among models. In this work, we used a physical process-based model to demonstrate the implications of climate change for streamflow and water temperature in two watersheds with distinctive flow regimes: the Snoqualmie watershed (WA) and Siletz watershed (OR), USA. Our model incorporated a downscaled ensemble of global climate model outputs and was calibrated with in situ and remotely sensed water temperatures. Furthermore, we compared predictions from our processed-based model to those from a publicly available and widely used statistical model. The process-based model projected greater changes in summer maximum water temperatures for the mixed-rain-snow Snoqualmie watershed than for the rain-dominated Siletz watershed as a result of the near-complete loss of winter snowpack and significant reduction in summer flow in the Snoqualmie watershed expected by the 2080s. Both models projected generally similar future spatial patterns of maximum water temperature in the two rivers, with cool reaches distributed farther upstream and fewer in number. However, the process-based model projected higher spatial heterogeneity in water temperature due to our spatially explicit simulation of streamflow and because we calibrated the model with spatially continuous remotely sensed water temperature data. We used stream temperature projections to assess the vulnerability of Pacific salmon and trout to changes in the spatial distribution of cold-water habitats during August by the 2080s. Results suggest that salmonids may have fewer summertime cold-water habitats in both watersheds. Projected stream warming may further limit particular species and life stages, especially in the Snoqualmie watershed. Our comparison of models highlights the importance of considering what might be gained by using a process-based model for evaluating and prioritizing management actions that mitigate climate impacts on cold-water habitats for stream fishes.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2024

Efficient distributed continual learning for steering experiments in real-time

Deep learning has emerged as a powerful method for extracting valuable information from large volumes of data. However, when new training data arrives continuously (i.e., is not fully available from the beginning), incremental training suffers from catastrophic forgetting (i.e., new patterns are reinforced at the expense of previously acquired knowledge). Training from scratch each time new training data becomes available would result in extremely long training times and massive data accumulation. Rehearsal-based continual learning has shown promise for addressing the catastrophic forgetting challenge, but research to date has not addressed performance and scalability. To fill this gap, we propose an approach based on a distributed rehearsal buffer that efficiently complements data-parallel training on multiple GPUs to achieve high accuracy, short runtime, and scalability. It leverages a set of buffers (local to each GPU) and uses several asynchronous techniques for updating these local buffers in an embarrassingly parallel fashion, all while handling the communication overheads necessary to augment input minibatches using unbiased, global sampling. We further propose a generalization of rehearsal buffers to support both classification and generative learning tasks, as well as more advanced rehearsal strategies (notably Dark Experience Replay, leveraging knowledge distillation). We illustrate this approach with a real-life HPC streaming application from the domain of ptychographic image reconstruction. Furthermore, we run extensive experiments on up to 128 GPUs of the ThetaGPU supercomputer to compare our approach with baselines representative of training-from-scratch (the upper bound in terms of accuracy) and incremental training (the lower bound). Results show that rehearsal-based continual learning achieves a top-5 validation accuracy close to the upper bound, while simultaneously exhibiting a runtime close to the lower bound.

Asynchronous data management↗
DOE OSTI2022

A comprehensive and fair comparison of two neural operators (with practical extensions) based on $\mathrm{FAIR}$ data

Neural operators can learn nonlinear mappings between function spaces and offer a new simulation paradigm for real-time prediction of complex dynamics for realistic diverse applications as well as for system identification in science and engineering. Herein, we investigate the performance of two neural operators, which have shown promising results so far, and we develop new practical extensions that will make them more accurate and robust and importantly more suitable for industrial-complexity applications. The first neural operator, DeepONet, was published in 2019 (Lu et al., 2019), and its original architecture was based on the universal approximation theorem of Chen & Chen (1995). The second one, named Fourier Neural Operator or FNO, was published in 2020, and it is based on parameterizing the integral kernel in the Fourier space. DeepONet is represented by a summation of products of neural networks (NNs), corresponding to the branch NN for the input function and the trunk NN for the output function; both NNs are general architectures, e.g., the branch NN can be replaced with a CNN or a ResNet. According to Kovachki et al. (2021), FNO in its continuous form can be viewed conceptually as a DeepONet with a specific architecture of the branch NN and a trunk NN represented by a trigonometric basis. In order to compare FNO with DeepONet computationally for realistic setups, we develop several extensions of FNO that can deal with complex geometric domains as well as mappings where the input and output function spaces are of different dimensions. We also develop an extended DeepONet with special features that provide inductive bias and accelerate training, and we present a faster implementation of DeepONet with cost comparable to the computational cost of FNO, which is based on the Fast Fourier Transform. Here we consider 16 different benchmarks to demonstrate the relative performance of the two neural operators, including instability wave analysis in hypersonic boundary layers, prediction of the vorticity field of a flapping airfoil, porous media simulations in complex-geometry domains, etc. We follow the guiding principles of FAIR (Findability, Accessibility, Interoperability, and Reusability) for scientific data management and stewardship. The performance of DeepONet and FNO is comparable for relatively simple settings, but for complex geometries the performance of FNO deteriorates greatly. We also compare theoretically the two neural operators and obtain similar error estimates for DeepONet and FNO under the same regularity assumptions.

42 ENGINEERING↗