Search NASA⌕ Search

SEARCH · Search NASA

Results for “map task”

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 451 records · Page 25
All sourcesNASA NTRSDOE OSTI
DOE OSTI2023

Comparison of quantum advantage experiments using random circuit sampling

Random circuit sampling, the task of sampling bit strings from a random unitary operator, has been implemented to demonstrate quantum advantage on the Sycamore quantum processor with 53 qubits and on the Zuchongzhi quantum processor with 56 and 61 qubits. Recently, it was claimed that classical computers using tensor network simulation could catch on to current noisy quantum processors for random circuit sampling. While the linear cross-entropy benchmark fidelity was used to certify all these claims, it may not capture statistical properties of outputs in detail. Here, we compare the bit strings sampled from classical computers using tensor network simulation by Pan et al. [F. Pan, K. Chen, and P. Zhang, Phys. Rev. Lett. 129, 090502 (2022)] and by Kalachev et al. [G. Kalachev, P. Panteleev, P. Zhou, and M.-H. Yung, arXiv:2112.15083] with the bit strings from the Sycamore quantum processor. It is shown that all of Kalachev et al.'s samples passed the NIST random number tests. The heat maps of bit strings show that Pan et al.'s and Kalachev et al.'s samples are quite different from the Sycamore or Zuchongzhi samples. The analysis with the Marchenko-Pastur distribution and the Wasssertein distances demonstrates that Kalachev et al.'s samples are statistically closer to the Sycamore samples than Pan et al.'s while the three datasets have similar values for the linear cross-entropy fidelity. In conclusion, our finding implies that further study is needed to certify or beat the claims of quantum advantage using random circuit sampling.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2020

Hawai‘i National Marine Renewable Energy Center (HINMREC)

The Hawaii National Marine Renewable Energy Center (HINMREC) was established to facilitate the commercialization of wave energy converter (WEC) devices, and to accelerate development and testing of Ocean Thermal Energy Conversion (OTEC) technologies. Housed at the Hawaii Natural Energy Institute (HNEI) of the University of Hawaii (UH), this program started with execution of the contract between the US Department of Energy (DOE) and UH in March 2009, and activities ran through the end of the project in September 2019. The DOE tasked HINMREC with facilitating and accelerating the build-out of the Navy’s Wave Energy Test Site (WETS), the nation’s first grid-connected open water wave energy conversion test facility, located off Marine Corps Base Hawaii (MCBH) in Kaneohe, on the island of Oahu. WETS was expanded from an existing test berth at 30m water depth to include test berths at 60m and 80m water depths. In recognition of the high costs associated with at-sea testing and evaluation of WEC devices, it was deemed imperative that HINMREC seek ways to reduce costs to developers by providing key research support to these early-stage technologies. WETS and HINMREC’s support allowed development of wave energy technology by providing a testing infrastructure that allows technology developers to test, demonstrate and evaluate their WEC devices, and generate data in order to advance WEC designs toward commercial readiness. HINMREC’s roles have included support in the establishment of the full site, including of wave resource characterization and site survey, followed by the essential roles of independent WEC device power performance assessment and environmental monitoring in support of deployed pre-commercial WECs. This DOE-funded work transitioned effectively to continue Navy-funded tasks for OTEC development, and activities at WETS, now under Naval Facilities Engineering Command (NAVFAC) funding. A secondary objective in the establishment of HINMREC was to assist the private sector in moving OTEC systems beyond proof-of-concept to pre-commercialization, primarily focused on system and component engineering, and local and global investigations into the potential environmental impacts of OTEC systems. HINMREC was tasked with maintaining high-resolution models of ocean thermal resources and the potential sustainable power output of OTEC systems. Ongoing tests begun previously under Navy funding at the OTEC Heat Exchangers (HXs) Test Facility at the Natural Energy Laboratory of Hawaii Authority (NELHA), in Kona on Hawaii Island, have been continued to identify cost-effective aluminum alloys for use in OTEC systems operating in the corrosive marine environment.

16 TIDAL AND WAVE POWER↗
DOE OSTI2017

Novel mechanisms for solid-state processing and grain growth with microstructure alignment in alnico-8 based permanent magnets

An estimated 750,000 new hybrid electric and plug-in battery vehicles, most with permanent magnet synchronous alternating current (PMAC) drive motors, took to the road in 2016 alone. Accompanied by 40% year over year growth in the EV market significant challenges exist in producing large quantities of permanent magnets (on the order of tens of millions) for reliable, low-cost traction motors [IE Agency, Energy Technology Perspectives (2017)]. Since the rare earth permanent magnet (REPM) market is essentially 100% net import reliant in the United States and has proven to have an unstable cost and supply structure in recent years, a replacement RE-free PM material must be designed or selected, fully developed, and implemented. Alnico, with its high saturation magnetization and excellent thermal stability, appears to be uniquely suited for this task. Further, while alnico typically has been considered a relatively low coercivity hard magnet, strides have been made to increase the coercivity to levels suitable for traction drive motors [W Tang, IEEE Trans. Magn., 51 (2015)]. If a simple non-cast approach for achieving near [001] easy axis grain aligned permanent magnets can be found, this would allow massproduced final-shape anisotropic high energy product magnets suitable for usage in compact high RPM rotor designs. Therefore, a powder metallurgical approach is being explored that uses classic compression molding with “de-bind and sinter” methods, where a novel applied uniaxial loading, and an applied magnetic field may create final-shape magnets with highly textured resulting microstructures by two different mechanisms. Results indicate a positive correlation between applied uniaxial load and resulting texture (Fig. 1), along with benefits from using an applied magnetic field for improved texture, as well. Lastly, the apparent mechanisms and resulting properties will be described using closed loop hysteresisgraph measurements, EBSD orientation mapping, and high-resolution SEM.

42 ENGINEERING↗
DOE OSTI2024

A Method for Producing Hierarchical and Statistically Calibrated Predictions of Nuclear Material Properties from Existing Models

Computer vision-based analysis of micrographs of nuclear materials is an emerging technique for property prediction, synthetic route identification, and other material analysis tasks. These analysis tasks play a pivotal role in many material characterization applications such as signature development for treaty verification, process optimization, etc. The backbone in many of the recent computer vision-based techniques is a deep learning model, which takes a fixed-size set of pixels and provides a class prediction for that set of pixels. For example, previous work developed a deep convolutional neural network (CNN) to predict the synthetic route from a 256 px x 256 px patch taken from a larger image of uranium ore concentrates. In this work, we present several methods for first calibrating these models in a manner that they can provide accurate probabilities of their predictions’ veracity, and several methods of combining these probabilities. Overall, the combination of these two steps into a pipeline allows for full-image and even full-sample (where a sample has many images) predictions with associated confidence values. Finally, we show that one can also use the patch predictions and confidence to produce a visualization to map predicted constituents through the image. Results and examples for predicting and mapping uranium ore concentrates’ synthetic process from imagery will be presented.

artificial intelligence↗
DOE OSTI2017

Accurate Holdup Calculations with Predictive Modeling & Data Integration

In facilities that process special nuclear material (SNM) it is important to account accurately for the fissile material that enters and leaves the plant. Although there are many stages and processes through which materials must be traced and measured, the focus of this project is material that is “held-up” in equipment, pipes, and ducts during normal operation and that can accumulate over time into significant quantities. Accurately estimating the holdup is essential for proper SNM accounting (vis-à-vis nuclear non-proliferation), criticality and radiation safety, waste management, and efficient plant operation. Usually it is not possible to directly measure the holdup quantity and location, so these must be inferred from measured radiation fields, primarily gamma and less frequently neutrons. Current methods to quantify holdup, i.e. Generalized Geometry Holdup (GGH), primarily rely on simple source configurations and crude radiation transport models aided by ad hoc correction factors. This project seeks an alternate method of performing measurement-based holdup calculations using a predictive model that employs state-of-the-art radiation transport codes capable of accurately simulating such situations. Inverse and data assimilation methods use the forward transport model to search for a source configuration that best matches the measured data and simultaneously provide an estimate of the level of confidence in the correctness of such configuration. In this work the holdup problem is re-interpreted as an inverse problem that is under-determined, hence may permit multiple solutions. A probabilistic approach is applied to solving the resulting inverse problem. This approach rates possible solutions according to their plausibility given the measurements and initial information. This is accomplished through the use of Bayes’ Theorem that resolves the issue of multiple solutions by giving an estimate of the probability of observing each possible solution. To use Bayes’ Theorem, one must have a model y(x) that maps the state variables x (the solution in this case) to the measurements y. In this case, the unknown state variables are the configuration and composition of the heldup SNM. The measurements are the detector readings. Thus, the natural model is neutral-particle radiation transport where a wealth of computational tools exists for performing these simulations accurately and efficiently. The combination of predictive model and Bayesian inference forms the Data Integration with Modeled Predictions (DIMP) method that serves as foundation for this project. The cost functional describing the model-to-data misfit is computed via a norm created by the inverse of the covariance matrix of the model parameters and responses. Since the model y(x) for the holdup problem is nonlinear, a nonlinear optimization on Q is conducted via Newton-type iterative methods to find the optimal values of the model parameters x. This project comprised a collaboration between NC State University (NCSU), the University of South Carolina (USC), and Oak Ridge National Laboratory (ORNL). The project was originally proposed in seven main tasks with an eighth contingency task to be performed if time and funding permitted; in fact time did not permit commencement of the contingency task and it was not performed. The remaining tasks involved holdup analysis with gamma detection strategies and separately with neutrons based on coincidence counting. Early in the project, and upon consultation with experts in coincidence counting it became evident that this approach is not viable for holdup applications and this task was replaced with an alternative, but valuable investigation that was carried out by the USC partner. Nevertheless, the experimental 4 measurements at ORNL of both gamma and neutron sources for the purpose of constructing Detector Response Functions (DRFs) with the associated uncertainties were indeed completed.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2015

Integrated Energy-Water Planning in the Western and Texas Interconnections

While long-term regional electricity transmission planning has traditionally focused on cost, infrastructure utilization, environmental impact, and reliability, the availability of water is an emerging issue. Toward this growing need, thermoelectric expansion should consider competing demands from other water use sectors balanced with fresh and non-traditional water supplies subject to climate variability. To address this need the Department of Energy's Office of Electricity Delivery and Energy Reliability supported an integrated planning project with funding through the American Reinvestment and Recovery Act (2009). Specifically, an integrated energy-water analysis was performed to support transmission system planners in the Western and Texas Interconnections to explore the potential implications of water availability and cost for long-term transmission planning. The project brought together electric transmission planners (e.g., Western Electricity Coordinating Council (WECC) and the Electric Reliability Council of Texas (ERCOT)) with western water planners (e.g., Western Governors' Association and the Western States Water Council). Efforts were organized into ten specific tasks: (1) project coordination and outreach; (2) thermoelectric water use; (3) non-thermoelectric water use; (4) water availability; (5) water cost; (6) environmental risk; (7) climate variability; (8) energy for water; (9) decision support system interface; and, (10) transmission planning support. Major accomplishments associated with this effort include: For the first time water availability was used to inform generation expansion planning by WECC and ERCOT. For the first time, projections of intensifying drought and its effect on reservoir levels, and thermal effluent discharge permitting were used to inform operational and expansion planning by ERCOT. Water withdrawal and consumption were characterized for each power plant in the WECC and ERCOT service areas/regions. Water use factors were also developed for a range of unit processes that allowed projection of future water demands related to electric generation expansion planning. Working with state water managers current and future water use (withdrawal and consumption) were projected throughout the Western United States at an 8-digit Hydraulic Unit Code (HUC-8) level (over 1200 watersheds). In a similar fashion water availability and cost were mapped across the Western United States. Considered were five different sources of water: unappropriated surface water, unappropriated groundwater, appropriated water, municipal wastewater and brackish groundwater. Water basins (at the HUC-8 level) were mapped across the Western United States with regard to their potential for conflicts between aquatic and riparian species and habitats listed under the Endangered Species Act and water availability for future energy development. Water planners were engaged through the Western States Water Council and thus reflects their membership of the 17 contiguous western states (i.e., Texas up through the Dakotas and West). Power plants at greatest risk to the impacts of drought were identified. The analysis considered the hazards of low flows, insufficient reservoir storage, and elevated water temperatures under intensifying drought conditions projected for the future. The electricity used to provide water-related services was mapped at a county level throughout the Western U.S. Considered was the electricity required for interbasin conveyance, agricultural pumping, drinking water and wastewater services. To communicate our results the project has produced 6 journal articles, 1 book chapter, 11 reports, and 47 presentations at related conferences.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗
DOE OSTI2020

Dynamic Equivalence of Large-Scale Power Systems Based on Boundary Measurements

Parallel computing helped speed up many tasks that can be done independently. The advance in gaming industry did not produce a technology that helps the power industry to reduce the computation time for dynamic simulation in individual cases. To perform faster than real-time simulation for the purpose of predicting power system dynamic trajectory, power industry continues to struggle to reduce the system size and simulation time of large-scale power systems while keeping its dynamic behavior under various disturbances. There is also an acute need for real-time dynamic model reduction as more renewables enter the generation mix with dramatic changes in the generation outputs. All existing model reduction solutions are based on having access to a detailed dynamic model of the system. The system changes by the minutes, dynamic models are only updated annually. To solve this problem, wide-area measurements obtained by the phasor measurement unit (PMU) at the boundaries between the reduced system and the study system is used to represent the external area. An artificial neuro-fuzzy inference system (ANFIS) is established to perform the mapping of the measurement to the external equivalent model. Here the external area is regarded as a black-box. Model reduction studies are conducted on the Northeast Power Coordinating Council (NPCC) under various types of contingencies, by using a co-simulation approach between PSS/E and MATLAB. The results look very promising and will be discussed in this paper.

Tong, Ning↗
DOE OSTI2023

Multi-fidelity Uncertainty Quantification for Homogenization Problems in Structure-Property Relationships from Crystal Plasticity Finite Elements

Crystal plasticity finite element method (CPFEM) has been an integrated computational materials engineering (ICME) workhorse to study materials behaviors and structure-property relationships for the last few decades. These relations are mappings from the microstructure space to the materials properties space. Due to the stochastic and random nature of microstructures, there is always some uncertainty associated with materials properties, for example, in homogenized stress-strain curves. For critical applications with strong reliability needs, it is often desirable to quantify the microstructure-induced uncertainty in the context of structure-property relationships. However, this uncertainty quantification (UQ) problem often incurs a large computational cost because many statistically equivalent representative volume elements (SERVEs) are needed. In this article, we apply a multi-level Monte Carlo (MLMC) method to CPFEM to study the uncertainty in stress-strain curves, given an ensemble of SERVEs at multiple mesh resolutions. By using the information at coarse meshes, we show that it is possible to approximate the response at fine meshes with a much reduced computational cost. We focus on problems where the model output is multi-dimensional, which requires us to track multiple quantities of interest (QoIs) at the same time. In conclusion, our numerical results show that MLMC can accelerate UQ tasks around 2.23x, compared to the classical Monte Carlo (MC) method, which is widely known as ensemble average in the CPFEM literature.

36 MATERIALS SCIENCE↗
DOE OSTI2015

Spectroscopic Needs for Imaging Dark Energy Experiments

Ongoing and near-future imaging-based dark energy experiments are critically dependent upon photometric redshifts (a.k.a. photo-z’s): i.e., estimates of the redshifts of objects based only on flux information obtained through broad filters. Higher-quality, lower-scatter photo-z’s will result in smaller random errors on cosmological parameters; while systematic errors in photometric redshift estimates, if not constrained, may dominate all other uncertainties from these experiments. The desired optimization and calibration is dependent upon spectroscopic measurements for secure redshift information; this is the key application of galaxy spectroscopy for imaging-based dark energy experiments. Hence, to achieve their full potential, imaging-based experiments will require large sets of objects with spectroscopically-determined redshifts, for two purposes: Training: Objects with known redshift are needed to map out the relationship between object color and z (or, equivalently, to determine empirically-calibrated templates describing the rest-frame spectra of the full range of galaxies, which may be used to predict the color-z relation). The ultimate goal of training is to minimize each moment of the distribution of differences between photometric redshift estimates and the true redshifts of objects, making the relationship between them as tight as possible. The larger and more complete our “training set” of spectroscopic redshifts is, the smaller the RMS photo-z errors should be, increasing the constraining power of imaging experiments; Requirements: Spectroscopic redshift measurements for ~30,000 objects over >~15 widely-separated regions, each at least ~20 arcmin in diameter, and reaching the faintest objects used in a given experiment, will likely be necessary if photometric redshifts are to be trained and calibrated with conventional techniques. Larger, more complete samples (i.e., with longer exposure times) can improve photo-z algorithms and reduce scatter further, enhancing the science return from planned experiments greatly (increasing the Dark Energy Task Force figure of merit by up to ~50%); Options: This spectroscopy will most efficiently be done by covering as much of the optical and near-infrared spectrum as possible at modestly high spectral resolution (λ/Δλ > ~3000), while maximizing the telescope collecting area, field of view on the sky, and multiplexing of simultaneous spectra. The most efficient instrument for this would likely be either the proposed GMACS/MANIFEST spectrograph for the Giant Magellan Telescope or the OPTIMOS spectrograph for the European Extremely Large Telescope, depending on actual properties when built. The PFS spectrograph at Subaru would be next best and available considerably earlier, c. 2018; the proposed ngCFHT and SSST telescopes would have similar capabilities but start later. Other key options, in order of increasing total time required, are the WFOS spectrograph at TMT, MOONS at the VLT, and DESI at the Mayall 4 m telescope (or the similar 4MOST and WEAVE projects); of these, only DESI, MOONS, and PFS are expected to be available before 2020. Table 2-3 of this white paper summarizes the observation time required at each facility for strawman training samples. To attain secure redshift measurements for a high fraction of targeted objects and cover the full redshift span of future experiments, additional near-infrared spectroscopy will also be required; this is best done from space, particularly with WFIRST-2.4 and JWST; Calibration: The first several moments of redshift distributions (the mean, RMS redshift dispersion, etc.), must be known to high accuracy for cosmological constraints not to be systematics-dominated (equivalently, the moments of the distribution of differences between photometric and true redshifts could be determined instead). The ultimate goal of calibration is to characterize these moments for every subsample used in analyses - i.e., to minimize the uncertainty in their mean redshift, RMS dispersion, etc. – rather than to make the moments themselves small. Calibration may be done with the same spectroscopic dataset used for training if that dataset is extremely high in redshift completeness (i.e., no populations of galaxies to be used in analyses are systematically missed). Accurate photo-z calibration is necessary for all imaging experiments; Requirements: If extremely low levels of systematic incompleteness (<~0.1%) are attained in training samples, the same datasets described above should be sufficient for calibration. However, existing deep spectroscopic surveys have failed to yield secure redshifts for 30–60% of targets, so that would require very large improvements over past experience. This incompleteness would be a limiting factor for training, but catastrophic for calibration. If <~0.1% incompleteness is not attainable, the best known option for calibration of photometric redshifts is to utilize cross-correlation statistics in some form. The most direct method for this uses cross-correlations between positions on the sky of bright objects of known spectroscopic redshift with the sample of objects that we wish to calibrate the redshift distribution for, measured as a function of spectroscopic z. For such a calibration, redshifts of ~100,000 objects over at least several hundred square degrees, spanning the full redshift range of the samples used for dark energy, would be necessary; and Options: The proposed BAO experiment eBOSS would provide sufficient spectroscopy for basic calibrations, particularly for ongoing and near-future imaging experiments. The planned DESI experiment would provide excellent calibration with redundant cross-checks, but will start after the conclusion of some imaging projects. An extension of DESI to the Southern hemisphere would provide the best possible calibration from cross-correlation methods for DES and LSST. We thus anticipate that our two primary needs for spectroscopy – training and calibration of photometric redshifts – will require two separate solutions. For ongoing and future projects to reach their full potential, new spectroscopic samples of faint objects will be needed for training; those new samples may be suitable for calibration, but the latter possibility is uncertain. In contrast, wide-area samples of bright objects are poorly suited for training, but can provide high-precision calibrations via cross-correlation techniques. Additional training/calibration redshifts and/or host galaxy spectroscopy would enhance the use of supernovae and galaxy clusters for cosmology. We also summarize additional work on photometric redshift techniques that will be needed to prepare for data from ongoing and future dark energy experiments.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2024

HumoNet: A Framework for Realistic Modeling and Simulation of Human Mobility Network

Understanding, analyzing, and predicting human mobility and dynamics are valuable to solving pressing problems, developing effective plans, and prescribing timely remedies. As a computational approach, realistic human mobility simulations allow us to understand, analyze, and predict complex systems, including human societies. Accurate simulations rely on (1) the model that captures interactions and behaviors of myriad entities in our society and (2) the mapping of model instances to real-world entities. Taking this into account, this paper introduces the Human Mobility Network simulation framework (HumoNet), an integrated patterns of life (POL) simulation framework that leverages real-world data layers including transportation networks, points of interest, populations, popularity, and human trajectories. HumoNet is a data informed model in which agents are equipped with activities, locomotion, and planning capabilities. To simulate realistic kinematic maneuvers of individuals in transportation networks, HumoNet harnesses a microscopic traffic simulator that provides interaction among vehicles and traffic objects. In this paper, we describe the framework, outline our methodologies, and discuss the data processing and challenges of each data layer. Through experiments, we demonstrate that our simulations capture key features of human mobility by comparing them to the literature and real data using standard measures of human mobility (i.e., the radius of gyration, number of locations visited, level of exploration) and metrics scoring (i.e., Jensen-Shannon divergence). We envision that the synthetic data produced by HumoNet will serve as a benchmark for analyzing epidemics, deploying EV charging networks, and validating AI/ML tasks such as location prediction.

Kim, Joon-Seok↗
DOE OSTI2017

Mapping and controlling ultrafast dynamics of highly excited H 2 molecules by VUV-IR pump-probe schemes

Here, we used ultrashort femtosecond vacuum ultraviolet (VUV) and infrared (IR) pulses in a pump-probe scheme to map the dynamics and nonequilibrium dissociation channels of excited neutral H 2 molecules. A nuclear wave packet is created in the B 1 Σ$+\atop{u}$ state of the neutral H 2 molecule by absorption of the ninth harmonic of the driving infrared laser field. Due to the large stretching amplitude of the molecule excited in the B 1 Σ$+\atop{u}$ electronic state, the effective H 2 + ionization potential changes significantly as the nuclear wave packet vibrates in the bound, highly electronically and vibrationally excited B potential-energy curve. We probed such dynamics by ionizing the excited neutral molecule using time-delayed VUV-or-IR radiation. We identified the nonequilibrium dissociation channels by utilizing three-dimensional momentum imaging of the ion fragments. We also found that different dissociation channels can be controlled, to some extent, by changing the IR laser intensity and by choosing the wavelength of the probe laser light. Furthermore, we concluded that even in a benchmark molecular system such as H 2 *, the interpretation of the nonequilibrium multiphoton and multicolor ionization processes is still a challenging task, requiring intricate theoretical analysis.

74 ATOMIC AND MOLECULAR PHYSICS↗
DOE OSTI2015

Correction of MRI-induced geometric distortions in whole-body small animal PET-MRI

Purpose: The fusion of positron emission tomography (PET) and magnetic resonance imaging (MRI) data can be a challenging task in whole-body PET-MRI. The quality of the registration between these two modalities in large field-of-views (FOV) is often degraded by geometric distortions of the MRI data. The distortions at the edges of large FOVs mainly originate from MRI gradient nonlinearities. This work describes a method to measure and correct for these kind of geometric distortions in small animal MRI scanners to improve the registration accuracy of PET and MRI data. Methods: The authors have developed a geometric phantom which allows the measurement of geometric distortions in all spatial axes via control points. These control points are detected semiautomatically in both PET and MRI data with a subpixel accuracy. The spatial transformation between PET and MRI data is determined with these control points via 3D thin-plate splines (3D TPS). The transformation derived from the 3D TPS is finally applied to real MRI mouse data, which were acquired with the same scan parameters used in the phantom data acquisitions. Additionally, the influence of the phantom material on the homogeneity of the magnetic field is determined via field mapping. Results: The spatial shift according to the magnetic field homogeneity caused by the phantom material was determined to a mean of 0.1 mm. The results of the correction show that distortion with a maximum error of 4 mm could be reduced to less than 1 mm with the proposed correction method. Furthermore, the control point-based registration of PET and MRI data showed improved congruence after correction. Conclusions: The developed phantom has been shown to have no considerable negative effect on the homogeneity of the magnetic field. The proposed method yields an appropriate correction of the measured MRI distortion and is able to improve the PET and MRI registration. Furthermore, the method is applicable to whole-body small animal imaging routines including different standard MRI sequences.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2021

Train Like a (Var)Pro: Efficient Training of Neural Networks with Variable Projection

Deep neural networks (DNNs) have achieved state-of-the-art performance across a variety of traditional machine learning tasks, e.g., speech recognition, image classification, and segmentation. The ability of DNNs to efficiently approximate high-dimensional functions has also motivated their use in scientific applications, e.g., to solve partial differential equations and to generate surrogate models. In this paper, we consider the supervised training of DNNs, which arises in many of the above applications. We focus on the central problem of optimizing the weights of the given DNN such that it accurately approximates the relation between observed input and target data. Devising effective solvers for this optimization problem is notoriously challenging due to the large number of weights, nonconvexity, data sparsity, and nontrivial choice of hyperparameters. To solve the optimization problem more efficiently, we propose the use of variable projection (VarPro), a method originally designed for separable nonlinear least-squares problems. Our main contribution is the Gauss--Newton VarPro method (GNvpro) that extends the reach of the VarPro idea to nonquadratic objective functions, most notably cross-entropy loss functions arising in classification. These extensions make GNvpro applicable to all training problems that involve a DNN whose last layer is an affine mapping, which is common in many state-of-the-art architectures. In our four numerical experiments from surrogate modeling, segmentation, and classification, GNvpro solves the optimization problem more efficiently than commonly used stochastic gradient descent (SGD) schemes. Finally, GNvpro finds solutions that generalize well, and in all but one example better than well-tuned SGD methods, to unseen data points.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Extreme sparsification of physics-augmented neural networks for interpretable model discovery in mechanics

Data-driven constitutive modeling with neural networks has received increased interest in recent years due to its ability to easily incorporate physical and mechanistic constraints and to overcome the challenging and time-consuming task of formulating phenomenological constitutive laws that can accurately capture the observed material response. However, even though neural network-based constitutive laws have been shown to generalize proficiently, the generated representations are not easily interpretable due to their high number of trainable parameters. Sparse regression approaches exist that allow for obtaining interpretable expressions, but the user is tasked with creating a library of model forms which by construction limits their expressiveness to the functional forms provided in the libraries. Here, in this work, we propose to train regularized physics-augmented neural network-based constitutive models utilizing a smoothed version of $L^0$-regularization. This aims to maintain the trustworthiness inherited by the physical constraints, but also enables interpretability which has not been possible thus far on any type of machine learning-based constitutive model where model forms were not assumed a priori but were actually discovered. During the training process, the network simultaneously fits the training data and penalizes the number of active parameters, while also ensuring constitutive constraints such as thermodynamic consistency. We show that the method can reliably obtain interpretable and trustworthy constitutive models for compressible and incompressible hyperelasticity, yield functions, and hardening models for elastoplasticity, using synthetic and experimental data. This work aims to set a new paradigm for interpretable machine learning models in the broad area of solid mechanics where low and limited data is available along with prior knowledge of physical constraints that the learned maps need to obey. This paradigm can potentially be extended to a broader spectrum of scientific exploration.

Data-driven constitutive models↗
DOE OSTI2021

A generalized machine learning workflow to visualize mechanical discontinuity

Accurate detection and mapping of mechanical discontinuity in materials has widespread industrial and research applications. Herein, we developed a generalized machine-learning framework for visualizing single mechanical discontinuity embedded in material of any composition, velocity, density, porosity, and size with limited data. The proposed visualization of discontinuity requires accurate estimations of the length, location, and orientation of the embedded discontinuity by processing multipoint wave-transmission measurements. k-Wave simulator is used to create a large dataset of elastic waveforms recorded during multi-point wave-transmission measurements through materials containing single mechanical discontinuity. k-Wave simulator considers the wave attenuation, dispersion, and mode conversion in wave motion. Discrete wavelet transform (DWT) and statistical feature extraction are essential for data preprocessing prior to the data-driven model development. DWT also minimizes the effect of noise. Using hyper-parameter tuning and cross validation, gradient boosting regression can visualize the mechanical discontinuity with an accuracy of 0.85, in terms of coefficient of determination. A double-layered neural network-based regression has better performance with an accuracy of 0.95. Use of convolutional neural network converts the predictive task from a waveform processing to an image processing problem. Convolutional neural network achieved a generalization performance of 0.91. The proposed generalized workflow requires robust simulation of wave propagation, signal processing, feature engineering, and model evaluation. Sensors closest to the source and those located opposite the source are the most significant for the desired visualization. Notably, the sensors closest to the source capture the non-linear associations, whereas the sensor on the border opposite to the source capture the linear associations between the measured waveforms and the properties of the mechanical discontinuity.

42 ENGINEERING↗
DOE OSTI2025

DeepAndes: A Self-Supervised Vision Foundation Model for Multispectral Remote Sensing Imagery of the Andes

By mapping sites at large scales usingremotely sensed data, archaeologists can generate unique insights into long-term demographic trends, interregional social networks, and human adaptations in the past. Remote sensing surveys complement field-based approaches, and their reach can be especially great when combined with deep learning and computer vision techniques. However, conventional supervised deep learning methods face challenges in annotating fine-grained archaeological features at scale. In addition, while recent vision foundation models have shown remarkable success in learning large-scale remote sensing data with minimal annotations, most off-the-shelf solutions are designed for RGB images rather than multispectral satellite imagery, such as the eight-band data used in our study. In this article, we introduce DeepAndes, a transformer-based vision foundation model trained on three million multispectral satellite images, specifically tailored for Andean archaeology. DeepAndes incorporates a customized DINOv2 self-supervised learning algorithm optimized for eight-band multispectral imagery, marking the first foundation model designed explicitly for the Andes region. We evaluate its image understanding performance through imbalanced image classification, image instance retrieval, and pixel-level semantic segmentation tasks. Our experiments show that DeepAndes achieves superior F1 scores, mean average precision, and Dice scores in few-shot learning scenarios, significantly outperforming models trained from scratch or pretrained on smaller datasets. This underscores the effectiveness of large-scale self-supervised pretraining in archaeological remote sensing.

Guo, Junlin [Vanderbilt Univ., Nashville, TN (Unit↗
DOE OSTI2017

Predictive coarse-graining

We propose a data-driven, coarse-graining formulation in the context of equilibrium statistical mechanics. In contrast to existing techniques which are based on a fine-to-coarse map, we adopt the opposite strategy by prescribing a probabilistic coarse-to-fine map. This corresponds to a directed probabilistic model where the coarse variables play the role of latent generators of the fine scale (all-atom) data. From an information-theoretic perspective, the framework proposed provides an improvement upon the relative entropy method and is capable of quantifying the uncertainty due to the information loss that unavoidably takes place during the coarse-graining process. Furthermore, it can be readily extended to a fully Bayesian model where various sources of uncertainties are reflected in the posterior of the model parameters. The latter can be used to produce not only point estimates of fine-scale reconstructions or macroscopic observables, but more importantly, predictive posterior distributions on these quantities. Predictive posterior distributions reflect the confidence of the model as a function of the amount of data and the level of coarse-graining. The issues of model complexity and model selection are seamlessly addressed by employing a hierarchical prior that favors the discovery of sparse solutions, revealing the most prominent features in the coarse-grained model. A flexible and parallelizable Monte Carlo – Expectation–Maximization (MC-EM) scheme is proposed for carrying out inference and learning tasks. A comparative assessment of the proposed methodology is presented for a lattice spin system and the SPC/E water model.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗