SEARCH · Engineering Papers

Results for “data representations”

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 73 records · Page 4

All sources NASA NTRS DOE OSTI

DOE OSTI2025

Data and Code for Understanding Generative AI Content with Embedding Models

This repository contains code for the experiments in the paper "Understanding Generative AI Content with Embedding Models". Constructing high-quality features is critical to any quantitative data analysis. While feature engineering was historically addressed by carefully hand-crafting data representations based on domain expertise, deep neural networks (DNNs) now offer a radically different approach. DNNs implicitly engineer features by transforming their input data into hidden feature vectors called embeddings. For embedding vectors produced by foundation models -- which are trained to be useful across many contexts -- we demonstrate that simple and well-studied dimensionality-reduction techniques such as Principal Component Analysis uncover inherent heterogeneity in input data concordant with human-understandable explanations. Of the many applications for this framework, we find empirical evidence that there is intrinsic separability between real samples and those generated by artificial intelligence (AI).

Vargas, Max [Pacific Northwest National Laboratory↗

DOE OSTI2026

Optimal binning of correlated measurements

Experimental measurements are commonly represented on a discrete grid, requiring a balance between granularity and statistical noise. Two strategies have traditionally been used to improve such representations: selecting an appropriate bin width to control discretization error and applying kernel-based smoothing to suppress fluctuations. Despite their shared goal, these approaches have largely developed independently, without a unified statistical description of how discretization and correlation jointly determine measurement precision. Here, we extend the discussion of optimal interval averaging to a correlation-aware setting by Gaussian process regression, which explicitly accounts for correlations among neighboring bins. Starting from first principles, we derive the mean-squared error of discretized measurements and obtain closed-form asymptotic expressions for the optimal bin width and correlation length. When recast in reduced variables, the theory reveals distinct universal scaling laws governing the error in the correlation-free and correlation-controlled regimes. Characterized by intrinsically smooth intensity profiles and counting-based statistics, neutron scattering measurements are well suited for demonstrating the enhanced error contraction enabled by inter-bin correlations. We show that such improvement is achievable over the experimentally accessible Q-range and across multiple instruments and material systems. These results show that explicitly accounting for correlations systematically reshapes the limits of precision in discretized, noise-limited measurements. More broadly, the framework provides a transferable statistical foundation for optimizing data representation, inference, and experimental design across the physical and data sciences.

Tung, Chi-Huan [ORNL] (ORCID:0000000221972074)↗

DOE OSTI2020

Efficient and Flexible Hierarchical Data Layouts for a Unified Encoding of Scalar Field Precision and Resolution

To address the problem of ever-growing scientific data sizes making data movement a major hindrance to analysis, we introduce a novel encoding for scalar fields: a unified tree of resolution and precision, specifically constructed so that valid cuts correspond to sensible approximations of the original field in the precision-resolution space. Furthermore, we introduce a highly flexible encoding of such trees that forms a parameterized family of data hierarchies. We discuss how different parameter choices lead to different trade-offs in practice, and show how specific choices result in known data representation schemes such as zfp[52], idx[58], and jpeg2000 [76]. Lastly, we provide system-level details and empirical evidence on how such hierarchies facilitate common approximate queries with minimal data movement and time, using real-world data sets ranging from a few gigabytes to nearly a terabyte in size. Experiments suggest that our new strategy of combining reductions in resolution and precision is competitive with state-of-the-art compression techniques with respect to data quality, while being significantly more flexible and orders of magnitude faster, and requiring significantly reduced resources.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2024

Understanding Generative AI Content with Embedding Models

The construction of high-quality numerical features is critical to any quantitative data analysis. Feature engineering has been historically addressed by carefully hand-crafting data representations based on domain expertise. This work views the internal representations of modern deep neural networks (DNNs), called embeddings, as an implicit form of traditional feature engineering. For trained DNNs, we show that these embeddings can reveal interpretable, high-level concepts in unstructured sample data. We use these embeddings in natural language and computer vision tasks to uncover both inherent heterogeneity in the underlying data and human-understandable explanations for it. In particular, we find empirical evidence that there is inherent separability between real data and those generated from AI models.

Vargas, Max↗

DOE OSTI2022

Case Study: NREL Campus Chilled Water Storage Potential: Benchmark Datasets Development and Applications, Task 4 - Use Case Demonstration

The Benchmark Datasets Development and Applications project is a three-year collaboration between the National Renewable Energy Laboratory (NREL), Oak Ridge National Laboratory, Pacific Northwest National Laboratory, and Lawrence Berkeley National Laboratory. The project seeks to collect and curate high-resolution, well-calibrated time series of building operational and indoor/outdoor environmental data, which are crucial to understanding and optimizing building energy efficiency performance and demand flexibility capabilities as well as benchmarking energy algorithms. Project outcomes include approximately twelve high-fidelity building datasets, enhanced data representation tools, and four case studies to illustrate example applications. The goal of these case studies is to define and execute analyses that demonstrate how one or more datasets collected through this project can address a data gap or challenge historically faced by building stakeholders. This technical paper summarizes the findings of one of these case studies, in which we studied the operational efficiencies of the central cooling system at NREL. We looked at three years of data from the three chillers in the Field Test Laboratory Building (FTLB), from 2019 to 2021, to compare equipment operation and demand throughout the time period. Our analysis indicates that all three chillers are operating at or below the optimal loading conditions for most of the operation time, and thus there was no efficiency drop due to loading of the chillers at full capacity. Our recommendation is that no chiller capacity increase is needed; instead, the central plant could benefit from adopting advanced control logics for optimal sequencing of chillers during part load operations. Analysis of adding chilled water thermal storage to the central plant indicated 34% savings in demand cost and 24.5% savings in total cost (energy consumption and demand charge cost). The payback period is estimated to be 11-22 years with an assumed TES cost of $\$$100-$200 per ton. This case study shows how a selected dataset is used to solve a practical building problem - learning the operational status of its components, analyzing the effectiveness of a proposed new technique, and aiding decision-making for the building operations and maintenance team.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

DOE OSTI2022

On the integration of molecular dynamics, data science, and experiments for studying solvent effects on catalysis

Computational workflows that combine molecular dynamics (MD) simulations and emerging data-centric (DC) methods can accelerate the screening and analysis of solvent systems experimentally and computationally. Here, MD simulations provide atomic positions and velocities of reactant, solvent, and catalyst materials that can be manipulated into data representations that in turn can be used by DC techniques to conduct predictive modeling, feature extraction, and experimental design. For liquid-phase catalytic applications, emerging DC techniques such as Convolutional and Graph Neural Networks (CNN/GNN), Topological Data Analysis (TDA), and Active Learning (AL) can leverage MD and experimental data to quickly predict solvent effects on reaction outcomes. For instance, in recent studies, 3D solvent environments obtained with MD have been exploited by CNNs to predict experimental reaction rates for homogeneous acid-catalyzed lignocellulosic processes. In this perspective, we discuss basic principles of DC methods and how these can be combined with MD to enable high-throughput screening of solvent selection for diverse catalysis applications.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2021

Phoenix: A Scalable Streaming Hypergraph Analysis Framework

We present Phoenix, a scalable hypergraph analytics framework for data analytics and knowledge discovery that was implemented on the leadership class computing platforms at Oak Ridge National Laboratory (ORNL). Our software framework comprises a distributed implementation of a streaming server architecture which acts as a gateway for various hypergraph generators/external sources to connect. Phoenix has the capability to utilize diverse hypergraph generators, including HyGen, a very large-scale hypergraph generator developed by ORNL. Phoenix incorporates specific algorithms for efficient data representation by exploiting hidden structures of the hypergraphs. Our experimental results demonstrate Phoenix’s scalable and stable performance on massively parallel computing platforms. Phoenix’s superior performance is due to the merging of high-performance computing with data analytic.

Kurte, Kuldeep↗

DOE OSTI2020

Machine Learning Approaches for Nuclear Material Accounting Data from Irradiation and Reprocessing

We are currently exploring data analysis methods for their ability to strengthen the synthesis and evaluation of information generated within domestic and international safeguard regimes. Safeguard data typically includes rich heterogenous datasets amenable to advanced data analytics methods, such as machine learning. We have converted transactional data containing both numerical and categorical attributes from a domestic nuclear material control and accountability (NMC&A) system into a low-dimensional numerical structure via linear principal component analysis. This data representation allows for global structure discovery via cluster analysis, which can characterize the typical behavior of each of the primary types of transaction events. Furthermore, the structure of the top principal components captures the “typical behavior” of the data and is thus amenable to anomaly detection through statistical hypothesis testing. We explored this capability by generating erroneous permutations of the data and computing the Q-residual quantity increase associated with the information loss when these data are projected into the low-dimensional principal component analysis space representative of typical transactions.Future work will focus on identifying data transformations (e.g., graph networks) that more closely align with the inherent structure of these transactions to explain more salient information and disambiguate the underlying nuclear process—in this case, irradiation and reprocessing—from artifacts of NMC&A system transactional record keeping data.

Drescher, Adam↗

DOE OSTI2020

Generation and representation of synthetic smart meter data

Advanced energy algorithms running at big-data scale will be necessary to identify, realize, and verify energy savings to meet government and utility goals of building energy efficiency. Any algorithm must be well characterized and validated before it is trusted to run at these scales. Smart meter data from real buildings will ultimately be required for the development, testing, and validation of these energy algorithms and processes. However, for initial development and testing, smart meter data are difficult to work with due to privacy restrictions, noise from unknown sources, data accessibility, and other concerns which can complicate algorithm development and validation. This paper describes a new methodology to generate synthetic smart meter data of electricity use in buildings using detailed building energy modeling, which aims to capture the variability and stochastics of real energy use in buildings. The methodology can create datasets tailored to represent specific scenarios with known truth and controllable amounts of synthetic noise. Knowledge of ground truth also allows the development and validation of enhanced processes which leverage building metadata, such as building type or size (floor area), in addition to smart meter data. The methodology described in this paper includes the key influencing factors of real-world building energy use including weather data, occupant-driven loads, building operation and maintenance practices, and special events. Data formats to support workflows leveraging both synthetic meter data and associated metadata are proposed and discussed. Finally, example use cases of the synthetic meter data are described to illustrate potential applications.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

DOE OSTI2024

Evaluating the Use of Foundational Chemical Language Models in Multimodal Graph Fusion

Rapid and accurate prediction of the physicochemical properties of molecules given their structures remains a key challenge in cheminformatics. Machine learning approaches offer high-throughput options, but the optimality of inductive biases and data representations are up for debate. For example, BERT-based masked language models (MLMs) can be trained in a self-supervised way on hundreds of millions to billions of readily available SMILES strings. Another option is graph neural networks (GNNs), which can operate directly on molecular structures. Yet, generating accurate molecular geometry is computationally expensive, leading to a relative scarcity in data compared to SMILES strings. It is attractive to combine these two paradigms by pre-training an LM on a large corpus of SMILES strings and embedding these representation into a geometric graph neural network. Despite the promise of such an approach, and contrary to previous studies, we find mixed results with the combination of the LMs and GNNs on several molecule datasets. In particular, we found evidence for improvement on the FreeSolv and QM7 benchmarks, but degraded performance on the ESOL, LIPO and QM9 datasets compared to a GNN baseline.

Francel, Collin [University of Alabama]↗

DOE OSTI2016

SOL: An end-to-end solution for real-world remote monitoring systems

This paper introduces SOL, which is both an efficient data representation for sensor measurements and network statistics, and a complete low-power wireless sensor management system that builds around it. A SOL system consists of multiple low-power wireless mesh networks in which motes connected to sensors and actuators send data to a single server. It offers multi-tier data replication and the associated data recovery. SOL is used in 3 pilot deployments for micro-climate monitoring, building automation and agriculture applications. In conjunction with Metronome Systems' NeoMote and Manager devices, SOL is a turn-key ready-to-deploy end-to-end system.

47 OTHER INSTRUMENTATION↗

DOE OSTI2016

SchemaOnRead: A Package for Schema-on-Read in R

Schema-on-read is an agile approach to data storage and retrieval that defers investments in data organization until production queries need to be run by working with data directly in native form. Schema-on-read functions have been implemented in a wide range of analytical systems, most notably Hadoop. SchemaOnRead is a CRAN package that uses R’s flexible data representations to provide transparent and convenient support for the schema-on-read paradigm in R. The schema-on- read tools within the package include a single function call that recursively reads folders with text, comma separated value, raster image, R data, HDF5, NetCDF, spreadsheet, Weka, Epi Info, Pajek network, R network, HTML, SPSS, Systat, and Stata files. The provided tools can be used as-is or easily adapted to implement customized schema-on-read tool chains in R. This paper’s contribution is that it introduces and describes SchemaOnRead, the first R package specifically focused on providing explicit schema-on-read support in R.

Data Science↗

DOE OSTI2016

SchemaOnRead: A Package for Schema-on-Read in R

Data Science↗

DOE OSTI2023

Toward extracting $$\gamma $$ from $$B\rightarrow DK$$ without binning

Abstract $$B^\pm \rightarrow DK^\pm $$ B ± → D K ± transitions are known to provide theoretically clean information about the CKM angle $$\gamma $$ γ , with the most precise available methods exploiting the cascade decay of the neutral D into CP self-conjugate states. Such analyses currently require binning in the D decay Dalitz plot, while a recently proposed method replaces this binning with the truncation of a Fourier series expansion. In this paper, we present a proof of principle of a novel alternative to these two methods, in which no approximations at the level of the data representation are required. In particular, our new strategy makes no assumptions about the amplitude and strong phase variation over the Dalitz plot. This comes at the cost of a degree of ambiguity in the choice of test statistic quantifying the compatibility of the data with a given value of $$\gamma $$ γ , with improved choices of test statistic yielding higher sensitivity. While our current proof-of-principle implementation does not demonstrate optimal sensitivity to $$\gamma $$ γ , its conceptually novel approach opens the door to new strategies for $$\gamma $$ γ extraction. More studies are required to see if these can be competitive with the existing methods.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

DOE OSTI2018

Human attitude analysis based on fuzzy soft differential equations with Bonferroni mean

In this paper, we model a system of fuzzy soft differential equations to analyze the behavior over the time of an individual depending on their companion’s actions under any particular situation against some decision. The Bonferroni mean (BM) is a very useful tool for group decision-making problems when arguments are interrelated to each other as Bonferroni mean can capture the interrelationship of the individual arguments. Using this ability of BM, we define Bonferroni fuzzy soft matrix (BFSM) and weighted Bonferroni fuzzy soft matrix (WBFSM) for data representation. WBFSM is a decision matrix and provide the optimum fuzzy soft constant (OFSC) which is the key element of fuzzy soft differential equation. By utilizing the OFSC, we develop a system of fuzzy soft differential equations to study a dynamical process with nonlinear uncertain and vague data. Second, we present a novel efficient technique for analyzing the future attitude of people due to their present decisions. To illustrate the practicality and feasibility of proposed technique an illustrative example is also discussed with the help of phase portrait and line graphs.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2026

GPU-Accelerated Analytic Simulation of Sparse Ionization Signal Formation in Pixelated Projection Detector

This paper presents a GPU-accelerated simulation package, TRED, for next-generation neutrino detectors with pixelated charge readout, leveraging community-driven software ecosystems to ensure adaptability and extensibility. We introduce two generic contributions: (i) an effective-charge representation based on Gaussian quadrature rules, in which the linear- interpolation factors for the field response inside each voxel are absorbed into the effective charge, and (ii) a sparse, block- binned tensor representation that enables efficient FFT-based computation of induced signals on readout electrodes for sparsely activated detector volumes. The former captures structure inside a voxel without dense sampling, while the latter achieves low memory usage and scalable runtime, as demonstrated in bench- mark studies. The underlying data representation is applicable to large-scale detectors and to other computational problems involving sparse activity.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

DOE OSTI2016

Treatments for Neutron Resonance Elastic Scattering Using the Multipole Formalism in Monte Carlo Codes

The elastic scattering cross sections of resonant nuclides can vary drastically within small energy intervals. In reactor physics simulations, accurate treatment of neutron resonance elastic scattering is vital, as the phenomenon can have a strong effect on scattering kinematics which, in turn, affect macroscopic values, such as the fuel temperature reactivity coefficient. To reduce memory requirements and facilitate coupling with heat transfer, a new representation for cross sections was recently adopted in OpenMC based on the multipole formalism. However, this new approach invalidates the previous implementations of correct resonance treatments, including the Doppler Broadening Rejection Correction (DBRC) method. This work thus developed a modified DBRC algorithm, called DBRC-MP, compatible with the new data representation. This new method is also verified against the previous DBRC method. While slightly more computationally costly than the traditional DBRC approach using ACE format cross sections, the use of the multipole representation in treating resonance scattering enables the reduction of memory requirements by a hundred-fold and facilitates the representation of temperature dependent cross sections. (authors)

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

DOE OSTI2019

Multiresolution 3D-DenseNet for Chemical Shift Prediction in NMR Crystallography

We have developed a deep learning algorithm for chemical shift prediction for atoms in molecular crystals that utilizes an atom-centered Gaussian density model for the 3D data representation of a molecule. We define multiple channels that describe different spatial resolutions for each atom type that utilizes cropping, pooling, and concatenation to create a multiresolution 3D-DenseNet architecture (MR-3D-DenseNet). Because the training and testing time scale linearly with the number of samples, the MR-3D-DenseNet can exploit data augmentation that takes into account the property of rotational invariance of the chemical shifts, thereby also increasing the size of the training data set by an order of magnitude without additional cost. We obtain very good agreement for 13 C, 15 N, and 17 O chemical shifts when compared to ab initio quantum chemistry methods, with the highest accuracy found for 1 H chemical shifts that is comparable to the error between the ab initio results and experimental measurements. Principal component analysis (PCA) is used to both understand these greatly improved predictions for 1 H , as well as indicating that chemical shift prediction for 13 C, 15 N, and 17 O, which have far fewer training environments than the 1 H atom type, will improve once more unique training samples are made available to exploit the deep network architecture.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗