Search NASA⌕ Search

SEARCH · Search NASA

Results for “Factorization machine”

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 271 records · Page 15
All sourcesNASA NTRSDOE OSTI
DOE OSTI2021

Explicit physics-informed neural networks for nonlinear closure: The case of transport in tissues

In upscaling methods, closures for nonlinear problems present a well-known challenge. While a number of theoretical methods have been proposed for handling such closures, nonlinearities still remain a significant obstacle for many problems. In this work, we use a combination of formal upscaling and data-driven machine learning for explicitly closing a nonlinear transport and reaction process in multiscale tissues. The classical effectiveness factor model is used to formulate the macroscale reaction kinetics. We train a multilayer perceptron network using training data generated by direct numerical simulations over microscale examples. Once trained, the network is used in an algorithm for numerically solving the upscaled (coarse-grained) differential equation describing mass transport and reaction in two example tissues. The network is described as being explicit in the sense that the network is trained using macroscale concentrations and gradients of concentration as components of the feature space rather than incorporating them as part of a constraint in the optimization process. Network training and solutions to the macroscale transport equations were computed for two different tissues. The two tissue types (brain and liver) exhibit markedly different geometrical complexity and spatial scale (cell size and sample size). The upscaled solutions for the average concentration are compared with numerical solutions derived from the microscale concentration fields by a posteriori averaging. There are three outcomes of this work of particular note. 1) Our overall approach results in an upscaled nonlinear PDE. The PDE is closed using a neural network, and our approach results in the definition of the classical effectiveness factor for effecting closure. 2) We identify particular source terms for the closure problem that are important for representing the structure of the closure. These source terms involve macroscale concentrations and their gradients. We adopt these source terms to use as explicit features in the learning algorithm. We find the trained networks that include the macroscale source terms generate models that are able to predict the correction factor with increased fidelity over those that do not. 3) We find that the trained network exhibits good generalizability, and it is able to predict the effectiveness factor with high fidelity for realistically-structured tissues despite the significantly different scale and geometrical complexity of the two example tissue types. This latter result emphasizes our purposeful connection between conventional averaging methods with the use of machine learning for closure; this contrasts with some machine learning methods for upscaling where the exact form of the macroscale equation remains unknown.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

NISQ+: Boosting quantum computing power by approximating quantum error correction

Quantum computers are growing in size, and design decisions are being made now that attempt to squeeze more computation out of these machines. In this spirit, we design a method to boost the computational power of near-term quantum computers by adapting protocols used in quantum error correction to implement "Approximate Quantum Error Correction (AQEC)." By approximating fully-fledged error correction mechanisms, we can increase the compute volume (qubits × gates, or "Simple Quantum Volume (SQV)") of near-term machines. The crux of our design is a fast hardware decoder that can approximately decode detected error syndromes rapidly. Specifically, we demonstrate a proof-of-concept that approximate error decoding can be accomplished online in near-term quantum systems by designing and implementing a novel algorithm in Single-Flux Quantum (SFQ) superconducting logic technology. This avoids a critical decoding backlog, hidden in all offline decoding schemes, that leads to idle time exponential in the number of T gates in a program. Our design utilizes one SFQ processing module per physical qubit. Employing state-of-the-art SFQ synthesis tools, we show that the circuit area, power, and latency are within the constraints of contemporary quantum system designs. Under pure dephasing error models, the proposed accelerator and AQEC solution is able to expand SQV by factors between 3,402 and 11,163 on expected near-term machines. The decoder achieves a 5% accuracy-threshold and pseudo-thresholds of ~ 5%,4.75%,4.5%, and 3.5% physical error-rates for code distances 3,5,7, and 9. Decoding solutions are achieved in a maximum of ~20 nanoseconds on the largest code distances studied. By avoiding the exponential idle time in offline decoders, we achieve a 10x reduction in required code distances to achieve the same logical performance as alternative designs.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Fertilizer management for global ammonia emission reduction

Crop production is a large source of atmospheric ammonia (NH 3 ), which poses risks to air quality, human health and ecosystems. However, estimating global NH 3 emissions from croplands is subject to uncertainties because of data limitations, thereby limiting the accurate identification of mitigation options and efficacy. In this report we develop a machine learning model for generating crop-specific and spatially explicit NH 3 emission factors globally (5-arcmin resolution) based on a compiled dataset of field observations. We show that global NH 3 emissions from rice, wheat and maize fields in 2018 were 4.3 ± 1.0 Tg N yr -1 , lower than previous estimates that did not fully consider fertilizer management practices. Furthermore, spatially optimizing fertilizer management, as guided by the machine learning model, has the potential to reduce the NH3 emissions by about 38% (1.6 ± 0.4 Tg N yr -1 ) without altering total fertilizer nitrogen inputs. Specifically, we estimate potential NH3 emissions reductions of 47% (44–56%) for rice, 27% (24–28%) for maize and 26% (20–28%) for wheat cultivation, respectively. Under future climate change scenarios, we estimate that NH3 emissions could increase by 4.0 ± 2.7% under SSP1–2.6 and 5.5 ± 5.7% under SSP5–8.5 by 2030–2060. However, targeted fertilizer management has the potential to mitigate these increases.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2015

An investigation of used electronics return flows: A data-driven approach to capture and predict consumers storage and utilization behavior

Highlights: • We analyzed a data set of HDDs returned back to an e-waste collection site. • We studied factors that affect the storage behavior. • Consumer type, brand and size are among factors which affect the storage behavior. • Commercial consumers have stored computers more than household consumers. • Machine learning models were used to predict the storage behavior. - Abstract: Consumers often have a tendency to store their used, old or un-functional electronics for a period of time before they discard them and return them back to the waste stream. This behavior increases the obsolescence rate of used still-functional products leading to lower profitability that could be resulted out of End-of-Use (EOU) treatments such as reuse, upgrade, and refurbishment. These types of behaviors are influenced by several product and consumer-related factors such as consumers’ traits and lifestyles, technology evolution, product design features, product market value, and pro-environmental stimuli. Better understanding of different groups of consumers, their utilization and storage behavior and the connection of these behaviors with product design features helps Original Equipment Manufacturers (OEMs) and recycling and recovery industry to better overcome the challenges resulting from the undesirable storage of used products. This paper aims at providing insightful statistical analysis of Electronic Waste (e-waste) dynamic nature by studying the effects of design characteristics, brand and consumer type on the electronics usage time and end of use time-in-storage. A database consisting of 10,063 Hard Disk Drives (HDD) of used personal computers returned back to a remanufacturing facility located in Chicago, IL, USA during 2011–2013 has been selected as the base for this study. The results show that commercial consumers have stored computers more than household consumers regardless of brand and capacity factors. Moreover, a heterogeneous storage behavior is observed for different brands of HDDs regardless of capacity and consumer type factors. Finally, the storage behavior trends are projected for short-time forecasting and the storage times are precisely predicted by applying machine learning methods.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗
DOE OSTI2016

A Deep Generative Deconvolutional Image Model

A deep generative model is developed for representation and analysis of images, based on a hierarchical convolutional dictionary-learning framework. Stochastic unpooling is employed to link consecutive layers in the model, yielding top-down image generation. A Bayesian support vector machine is linked to the top-layer features, yielding max-margin discrimination. Deep deconvolutional inference is employed when testing, to infer the latent features, and the top-layer features are connected with the max-margin classifier for discrimination tasks. The model is efficiently trained using a Monte Carlo expectation-maximization (MCEM) algorithm; the algorithm is implemented on graphical processor units (GPU) to enable large-scale learning, and fast testing. Excellent results are obtained on several benchmark datasets, including ImageNet, demonstrating that the proposed model achieves results that are highly competitive with similarly sized convolutional neural networks.

convolutional neural network↗
DOE OSTI2015

Convective Heat Transfer Coefficients of Automatic Transmission Fluid Jets with Implications for Electric Machine Thermal Management: Preprint

Thermal management for electric machines (motors/ generators) is important as the automotive industry continues to transition to more electrically dominant vehicle propulsion systems. Cooling of the electric machine(s) in some electric vehicle traction drive applications is accomplished by impinging automatic transmission fluid (ATF) jets onto the machine's copper windings. In this study, we provide the results of experiments characterizing the thermal performance of ATF jets on surfaces representative of windings, using Ford's Mercon LV ATF. Experiments were carried out at various ATF temperatures and jet velocities to quantify the influence of these parameters on heat transfer coefficients. Fluid temperatures were varied from 50 degrees C to 90 degrees C to encompass potential operating temperatures within an automotive transaxle environment. The jet nozzle velocities were varied from 0.5 to 10 m/s. The experimental ATF heat transfer coefficient results provided in this report are a useful resource for understanding factors that influence the performance of ATF-based cooling systems for electric machines.

33 ADVANCED PROPULSION SYSTEMS↗
DOE OSTI2026

Heavy-Duty Nonroad Material Handler Electrification Part 1: Real-World Drive Cycle Development

Knowing a detailed operating cycle is critical for developing and testing equipment. Operating cycles can be separated by two clear distinctions: (1) regulatory or non-regulatory and (2) application at the engine-only or full machine level. The Environmental Protection Agency’s (EPA) Nonroad Transient Cycle (NRTC) may be a good representation of engine use in many types of equipment, but there is a gap in standardized and validated drive cycles specifically for nonroad material handlers. Lacking a standardized drive cycle makes it difficult to accurately benchmark machine performance and validate new powertrain technologies. The objective of this investigation is to illustrate the development of a custom drive cycle augmented with real-world customer use data that serves multiple purposes: (1) understand the range of operation and utilization that formulated inputs for electrified architecture analysis and (2) develop a repetitive and consistent maneuver to establish baseline energy consumption enabling equivalent comparison to future electrified prototype builds. This article presents a solution specifically for a 23-ton nonroad material handler in which material handling, machine transport, and extended idle were homologated to form representative short cycles defined by machine velocity and hydraulic cylinder position. The most intensive material handling short cycles had a load factor of 40% and an average fuel rate of 16 L/h. Combined with a visual aid, the short cycles exhibited low variability, having less than 5% root mean square (RMS) error in lift and reach position with respect to the average. The machine’s performance on these short cycles at the Advanced Power Systems Research Center (APSRC) was compared to results from two real-world customer locations operating the instrumented test machine in a cyclical manner, and for similar ground conditions were found to be comparable in fuel consumption.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2025

Improving tropical cyclone rapid intensification forecasts with satellite measurements of sea surface salinity and calibrated machine learning

Forecasting rapid intensification (RI) of tropical cyclones (TC) is a mission known for large errors. One under-researched factor that affects TC intensification is salinity, which is important for density stratification in certain ocean regions and can affect the surface enthalpy flux under a strengthening hurricane. To investigate the impact and efficacy of using salinity information in state-of-the-art forecasting, we use a statistical model consisting of a variety of machine learning (ML) methods. For salinity data, we use satellite measurements of pre-storm sea surface salinity (SSS) as a proxy for the salinity stratification. We train and test the model on various ocean basins, including the Atlantic, eastern North Pacific and western North Pacific. A calibrator is trained on top of the ML models to correct and enhance probability forecasts. The calibrator significantly improves probability forecasts relative to recent works. The ML model performance is improved with the addition of SSS in the Eastern North Pacific, western North Pacific, and the Caribbean subregion of the North Atlantic, and the overall model performance is better than previous studies. SSS decreases model skill for a model trained on the full Atlantic basin. In the Indian Ocean, SSS is also notably correlated with RI occurrence, but the TC samples are not sufficient to train ML models.

hurricane↗
DOE OSTI2025

Discovery of unconventional and nonintuitive self-assembling peptide materials using experiment-driven machine learning

Prediction of peptide secondary structure is challenging because of complex molecular interactions, sequence-specific behavior, and environmental factors. Traditional design strategies, based on hydrophobicity and structural propensity, can be biased and could indeed prevent discovery of interesting, diverse, and unconventional peptides with desired nanostructure assembly. Using β sheet formation in pentapeptides as a case study, we used an integrated high-throughput experimental workflow and an artificial intelligence–driven active learning framework to improve prediction accuracy of self-assembly. By focusing on sequences where machine learning (ML) predictions deviate from conventional design strategies, we synthesized and tested 268 pentapeptides, successfully finding 96 forming β sheet assemblies, including unconventional sequences (e.g., ILFSM, LMISI, MITIY, MISIW, and WKIYI) not predicted by traditional methods. Our ML models outperformed conventional β sheet propensity tables, revealing useful chemical design rules. A web interface is provided to facilitate community access to these models. This work highlights the value of ML-driven approaches in overcoming the limitations of current peptide design strategies.

Talluri, Y. Nissi [Indian Inst. of Technology (IIT↗
DOE OSTI2015

SU-E-T-273: Do Task Group External Beam QA Recommendations Guarantee Accurate Treatment Plan Dose Delivery?

Purpose: AAPM task groups 40/142 have provided an invaluable set of goals for physicists designing QA programs, attempting to standardize what would otherwise likely be a highly variable phenomenon across institutions. However, with the complexity of modalities such as VMAT, we hypothesize that following these guidelines to the letter might still allow unacceptable dose discrepancies. To explore this hypothesis we simulated machines bordering on QA acceptability, and calculated the effect on patient plans. Methods: Two errant machines were simulated in Aria/Eclipse, each just within task group criteria for output, percent depth dose, beam profile, gantry and collimator rotations, and jaw and MLC positions. One machine minimized dose to the PTV (machine A) and the other maximized dose to the OARs (machine B). Clinical treatment plans (3-phase prostate, n=3; hypofractionated lung, n=1) were calculated on these machines and the dose distributions compared. A prostate case was examined for contribution of error sources and evaluated using delivery QA data. Results: The prostate plans showed mean decreases in target D95 of 9.9% of prescription dose on machine A. On machine B, The rectal and bladder V70Gy each increased by 7.1 percentage points, while their V45Gy increased by 16.2% and 15.0% respectively. In the lung plan, the target D95 decreased by 12.8% and the bronchial tree Dmax increased by 21% of prescription dose, on machines A and B. One prostate plan showed target dose errors of 3.8% from MLC changes, 2% from output, ∼3% from energy and ∼0.5% from other factors. This plan achieved an 88.4% gamma passing rate using 3%/3mm using ArcCHECK. Conclusion: In the unlikely event that a machine exhibits all maximum errors allowed by TG 40/142, unacceptably large changes in dose delivered are possible especially in highly modulated VMAT plans, despite the machine passing routine QA.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2025

Machine Learning-Guided Identification of PET Hydrolases from Natural Diversity

The enzymatic depolymerization of poly(ethylene terephthalate) (PET) is emerging as a leading chemical recycling technology for waste polyester. As part of this endeavor, new candidate enzymes identified from natural diversity can serve as useful starting points for enzyme evolution and engineering. In this study, we improved upon HMM searches by applying an iterative machine learning strategy to identify 400 putative PET-degrading enzymes (PET hydrolases) from naturally occurring homologs. Using high-throughput (HTP) experimental techniques, we successfully expressed and purified >200 enzyme candidates and assayed them for PET hydrolysis activity as a function of pH, temperature, and substrate crystallinity. From this library, we discovered 91 previously unknown PET hydrolases, 35 of which retain activity at pH 4.5 on crystalline material, which are conditions relevant to developing more efficient commercial processes. Notably, four enzymes showed equal to or higher activity than LCC-ICCG, a benchmark PET hydrolase, at this challenging condition in our screening assay, and 11 of which have pH optima <7. Using these data, we identified regions of PETases statistically correlated to activity at lower pH. We additionally investigated the effect of condition-specific activity data on trained machine learning predictors and found a precision (putative hit rate) improvement of up to 30% compared to a Hidden Markov Model alone. Our findings show that by pointing enzyme discovery toward conditions of interest with multiple rounds of experimental and machine learning, we can discover large sets of active enzymes and explore factors associated with activity at those conditions.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2020

Preshot Report for the NIF Discovery Science DDCyl Campaign: D_Hyd_DDCyl_DRT_DDD/EEE/FFF

The first Discovery Science shot day with the NIF DDCyl campaign demonstrated that Rayleigh-Taylor instability growth during the deceleration phase is scale-invariant for targets that varied in radial dimension and implosion time-scale by a factor of three. Smaller cylindrical implosions at the OMEGA laser facility were compared to the larger cylinders fielded in our first NIF outing. Both sets of targets had an initial m = 20 sinusoidal perturbation machined on the inner surface of an embedded aluminum marker layer, with the amplitude larger by the same factor of three for the NIF-scale targets. We observed growth factors of 14 at late time (t=λ = 7:7 ns, where λ = 1 for OMEGA and λ = 3 for NIF). These used a high density (300 mg/cm³) CH foam core, providing a very long period of deceleration before the rebounding shock hit the incoming spike tips, but the final convergence was only CR≈ 2:25. For our second shot day with this campaign, we are attempting to increase the convergence ratio to CR≥ 5 before reshock, while still measuring instability growth during the convergence/deceleration phase. As discussed in previous reports, the best way to increase convergence before reshock is to lower the density of the central foam. The preshot designs called for the use of a 30 mg/cm³ CH foam, though the as-built foams are closer to 40-42 mg/cm³, which is predicted to result in a final CR≈ 5:1 before reshock. Other than the lower foam density, the target design is identical to those used previously.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2020

Design study of a combined interferometer and polarimeter for a high-field, compact tokamak

This article is the first design study of a combined interferometer and polarimeter on a compact, high-field, high-density, net-energy tokamak. Recent advances in superconducting technology have made possible designs for compact, high magnetic field fusion power plants, such as ARC [Sorbom et al., Fusion Eng. Des. 100, 378 (2015)], and experiments, such as SPARC [Greenwald et al., PSFC Report No. RR-18-2 (2018)]. These new designs create both challenges and opportunities for plasma diagnostics. The diagnostic proposed in this work, called InterPol, takes advantage of unique opportunities provided by high magnetic field and density to measure both line-averaged density and poloidal magnetic field with a single set of CO 2 and quantum cascade lasers. These measurements will be used for fast density feedback control, constraint of density and safety factor profiles, and density fluctuation measurements. Synthetic diagnostic testing using a model machine geometry, called MQ1 (Mission Q ≥ 1), and profiles simulated with Tokamak Simulation Code indicate that InterPol will be able to measure steady state density and poloidal magnetic field, as well as fluctuations caused by toroidal Alfven eigenmodes and other phenomena on a high-field compact tokamak.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2023

Combining variational autoencoders and physical bias for improved microscopy data analysis *

Electron and scanning probe microscopy produce vast amounts of data in the form of images or hyperspectral data, such as electron energy loss spectroscopy or 4D scanning transmission electron microscope, that contain information on a wide range of structural, physical, and chemical properties of materials. To extract valuable insights from these data, it is crucial to identify physically separate regions in the data, such as phases, ferroic variants, and boundaries between them. In order to derive an easily interpretable feature analysis, combining with well-defined boundaries in a principled and unsupervised manner, here we present a physics augmented machine learning method which combines the capability of variational autoencoders to disentangle factors of variability within the data and the physics driven loss function that seeks to minimize the total length of the discontinuities in images corresponding to latent representations. Our method is applied to various materials, including NiO-LSMO, BiFeO 3 , and graphene. The results demonstrate the effectiveness of our approach in extracting meaningful information from large volumes of imaging data. The customized codes of the required functions and classes to develop phyVAE is available at https://github.com/arpanbiswas52/phy-VAE.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2015

MO-F-CAMPUS-T-03: Data Driven Approaches for Determination of Treatment Table Tolerance Values for Record and Verification Systems

Purpose: To determine appropriate couch tolerance values for modern radiotherapy linac R&V systems with indexed patient setup. Methods: Treatment table tolerance values have been the most difficult to lower, due to many factors including variations in patient positioning and differences in table tops between machines. We recently installed nine linacs with similar tables and started indexing every patient in our clinic. In this study we queried our R&V database and analyzed the deviation of couch position values from the acquired values at verification simulation for all patients treated with indexed positioning. Mean and standard deviations of daily setup deviations were computed in the longitudinal, lateral and vertical direction for 343 patient plans. The mean, median and standard error of the standard deviations across the whole patient population and for some disease sites were computed to determine tolerance values. Results: The plot of our couch deviation values showed a gaussian distribution, with some small deviations, corresponding to setup uncertainties on non-imaging days, and SRS/SRT/SBRT patients, as well as some large deviations which were spot checked and found to be corresponding to indexing errors that were overriden. Setting our tolerance values based on the median + 1 standard error resulted in tolerance values of 1cm lateral and longitudinal, and 0.5 cm vertical for all non- SRS/SRT/SBRT cases. Re-analizing the data, we found that about 92% of the treated fractions would be within these tolerance values (ignoring the mis-indexed patients). We also analyzed data for disease site based subpopulations and found no difference in the tolerance values that needed to be used. Conclusion: With the use of automation, auto-setup and other workflow efficiency tools being introduced into radiotherapy workflow, it is very essential to set table tolerances that allow safe treatments, but flag setup errors that need to be reassessed before treatments.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2017

High-Luminosity Large Hadron Collider (HL-LHC). Technical Design Report V. 0.1

The Large Hadron Collider (LHC) is one of the largest scientific instruments ever built. Since opening up a new energy frontier for exploration in 2010, it has gathered a global user community of about 7,000 scientists working in fundamental particle physics and the physics of hadronic matter at extreme temperature and density. To sustain and extend its discovery potential, the LHC will need a major upgrade in the 2020s. This will increase its instantaneous luminosity (rate of collisions) by a factor of five beyond the original design value and the integrated luminosity (total collisions created) by a factor ten. The LHC is already a highly complex and exquisitely optimised machine so this upgrade must be carefully conceived and will require about ten years to implement. The new configuration, known as High Luminosity LHC (HL-LHC), relies on a number of key innovations that push accelerator technology beyond its present limits. Among these are cutting-edge 11-12 tesla superconducting mag-nets, compact superconducting cavities for beam rotation with ultra-precise phase control, new technology and physical processes for beam collimation and 100 metre-long high-power superconducting links with negligible energy dissipation. The present document describes the technologies and components that will be used to realise the project and is intended to serve as the basis for the detailed engineering design of HL-LHC.

43 PARTICLE ACCELERATORS↗
DOE OSTI2015

High-Luminosity Large Hadron Collider (HL-LHC) : Preliminary Design Report

The Large Hadron Collider (LHC) is one of the largest scientific instruments ever built. Since opening up a new energy frontier for exploration in 2010, it has gathered a global user community of about 7,000 scientists working in fundamental particle physics and the physics of hadronic matter at extreme temperature and density. To sustain and extend its discovery potential, the LHC will need a major upgrade in the 2020s. This will increase its luminosity (rate of collisions) by a factor of five beyond the original design value and the integrated luminosity (total collisions created) by a factor ten. The LHC is already a highly complex and exquisitely optimised machine so this upgrade must be carefully conceived and will require about ten years to implement. The new configuration, known as High Luminosity LHC (HL-LHC), will rely on a number of key innovations that push accelerator technology beyond its present limits. Among these are cutting-edge 11-12 tesla superconducting magnets, compact superconducting cavities for beam rotation with ultra-precise phase control, new technology and physical processes for beam collimation and 300 metre-long high-power superconducting links with negligible energy dissipation. The present document describes the technologies and components that will be used to realise the project and is intended to serve as the basis for the detailed engineering design of HL-LHC.

43 PARTICLE ACCELERATORS↗