SEARCH · Search NASA

Results for “deep generative models”

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 739 records · Page 41

All sources NASA NTRS DOE OSTI

NASA NTRS1993

Origin of high-Ti lunar ultramafic glasses

A series of high pressure melting experiments were performed on a synthetic analog of the Apollo 14 Black Glass. At 15 kb and 1400 C, olivine is the liquidus phase. At 25 kb and 1450 C, orthopyroxene and Cr-spinel crystallize. A multiple saturation point with olivine + orthopyroxene + spinel on the liquidus occurs at 20 kb and 1440 C. It was proposed that the high-Ti ultramafic glass evolved from a low-Ti Green Glass generated deep in the lunar interior that assimilated shallow-level, late stage cumulates of magma ocean differentiation. Heat for assimilation is provided by superheat of the low-Ti magma and melts high-Ca pyroxene and ilmenite cumulates at a depth of 100 km. This model is a variant of that proposed by Hubbard and Minear and rejected by Ringwood and Kesson.

Wagner, T. P.↗

NASA NTRS2006

Variable Stiffness Panel Structural Analyses With Material Nonlinearity and Correlation With Tests

Results from structural analyses of three tow-placed AS4/977-3 composite panels with both geometric and material nonlinearities are presented. Two of the panels have variable stiffness layups where the fiber orientation angle varies as a continuous function of location on the panel planform. One variable stiffness panel has overlapping tow bands of varying thickness, while the other has a theoretically uniform thickness. The third panel has a conventional uniform-thickness [plus or minus 45](sub 5s) layup with straight fibers, providing a baseline for comparing the performance of the variable stiffness panels. Parametric finite element analyses including nonlinear material shear are first compared with material characterization test results for two orthotropic layups. This nonlinear material model is incorporated into structural analysis models of the variable stiffness and baseline panels with applied end shortenings. Measured geometric imperfections and mechanical prestresses, generated by forcing the variable stiffness panels from their cured anticlastic shapes into their flatter test configurations, are also modeled. Results of these structural analyses are then compared to the measured panel structural response. Good correlation is observed between the analysis results and displacement test data throughout deep postbuckling up to global failure, suggesting that nonlinear material behavior is an important component of the actual panel structural response.

Wu, K. Chauncey↗

NASA NTRS1982

A first-order estimate of shock heating and vaporization in oceanic impacts

The vaporization of water in oceanic impacts of asteroids or comets of multikilometer dimensions is estimated by a semianalytical modeling of impact heating and shock isobar geometry that is based on computer code calculations. The mass of water vaporized in an infinitely deep ocean by the impact of a 10 km diameter asteroid at 25 km/sec (these values have been proposed for the Cretaceous/Tertiary extinction bolide) is approximately equal to the total mass of water vapor present in the earth's atmosphere, and 3-4 orders of magnitude larger than the mass of water vapor in the stratosphere. For projectiles of this size, however, the finite depth of the ocean becomes significant and may considerably reduce the amount of water vapor initially generated by the impact. Climatological models and extinction scenarios invoking the effects of impact-generated water vapor may critically depend on the a priori ambiguous details of the hypothesized impact.

Croft, S. K.↗

DOE OSTI2020

Deep neural network improves the estimation of polygenic risk scores for breast cancer

Polygenic risk scores (PRS) estimate the genetic risk of an individual for a complex disease based on many genetic variants across the whole genome. Here, we compared a series of computational models for estimation of breast cancer PRS. A deep neural network (DNN) was found to outperform alternative machine learning techniques and established statistical algorithms, including BLUP, BayesA, and LDpred. In the test cohort with 50% prevalence, the Area Under the receiver operating characteristic Curve (AUC) were 67.4% for DNN, 64.2% for BLUP, 64.5% for BayesA, and 62.4% for LDpred. BLUP, BayesA, and LPpred all generated PRS that followed a normal distribution in the case population. However, the PRS generated by DNN in the case population followed a bimodal distribution composed of two normal distributions with distinctly different means. This suggests that DNN was able to separate the case population into a high-genetic-risk case subpopulation with an average PRS significantly higher than the control population and a normal-genetic-risk case subpopulation with an average PRS similar to the control population. This allowed DNN to achieve 18.8% recall at 90% precision in the test cohort with 50% prevalence, which can be extrapolated to 65.4% recall at 20% precision in a general population with 12% prevalence. Interpretation of the DNN model identified salient variants that were assigned insignificant p values by association studies, but were important for DNN prediction. These variants may be associated with the phenotype through nonlinear relationships.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2015

Simplified predictive models for CO 2 sequestration performance assessment

CO2 sequestration in deep saline formations is increasingly being considered as a viable strategy for the mitigation of greenhouse gas emissions from anthropogenic sources. In this context, detailed numerical simulation based models are routinely used to understand key processes and parameters affecting pressure propagation and buoyant plume migration following CO2 injection into the subsurface. As these models are data and computation intensive, the development of computationally-efficient alternatives to conventional numerical simulators has become an active area of research. Such simplified models can be valuable assets during preliminary CO2 injection project screening, serve as a key element of probabilistic system assessment modeling tools, and assist regulators in quickly evaluating geological storage projects. We present three strategies for the development and validation of simplified modeling approaches for CO2 sequestration in deep saline formations: (1) simplified physics-based modeling, (2) statisticallearning based modeling, and (3) reduced-order method based modeling. In the first category, a set of full-physics compositional simulations is used to develop correlations for dimensionless injectivity as a function of the slope of the CO2 fractional-flow curve, variance of layer permeability values, and the nature of vertical permeability arrangement. The same variables, along with a modified gravity number, can be used to develop a correlation for the total storage efficiency within the CO2 plume footprint. Furthermore, the dimensionless average pressure buildup after the onset of boundary effects can be correlated to dimensionless time, CO2 plume footprint, and storativity contrast between the reservoir and caprock. In the second category, statistical “proxy models” are developed using the simulation domain described previously with two approaches: (a) classical Box-Behnken experimental design with a quadratic response surface, and (b) maximin Latin Hypercube sampling (LHS) based design with a multidimensional kriging metamodel fit. For roughly the same number of simulations, the LHS-based metamodel yields a more robust predictive model, as verified by a k-fold cross-validation approach (with data split into training and test sets) as well by validation with an independent dataset. In the third category, a reduced-order modeling procedure is utilized that combines proper orthogonal decomposition (POD) for reducing problem dimensionality with trajectory-piecewise linearization (TPWL) in order to represent system response at new control settings from a limited number of training runs. Significant savings in computational time are observed with reasonable accuracy from the PODTPWL reduced-order model for both vertical and horizontal well problems – which could be important in the context of history matching, uncertainty quantification and optimization problems. The simplified physics and statistical learning based models are also validated using an uncertainty analysis framework. Reference cumulative distribution functions of key model outcomes (i.e., plume radius and reservoir pressure buildup) generated using a 97-run full-physics simulation are successfully validated against the CDF from 10,000 sample probabilistic simulations using the simplified models. The main contribution of this research project is the development and validation of a portfolio of simplified modeling approaches that will enable rapid feasibility and risk assessment for CO2 sequestration in deep saline formations.

54 ENVIRONMENTAL SCIENCES↗

NASA NTRS2003

Preliminary Analysis and Simulations of July 23rd Extended Anvil Case

A key focus of CRYSTAL-FACE (Cirrus Regional Study of Tropical Anvils and cirrus Layers - Florida Area Cirrus Experiment) was the generation and subsequent evolution of cirrus outflow from deep convective cloud systems. A preliminary integrated look at the observations of an extended cirrus anvil cloud system observed on 23 July 2002 will be presented, including lidar and millimeter radar observations from NASA's ER-2 and in- situ observations from NASA's WB-57 and University of North Dakota Citation. The observations will be compared to preliminary results of simulations using 1-D and 2-D high-resolution (100 meter) cloud resolving models. The CRMs explicitly account for cirrus microphysical development by resolving the evolving ice crystal size distribution (bin model) in time and space. Both homogeneous and heterogeneous nucleation are allowed in the model. The CRM simulations are driven using the output of regional simulations using MM5 that produces deep convection similar to what was observed. The MM5 model employs a 2 km inner grid (32 layers) over a 360 km domain, nested within a 6 km grid over a 600 km domain. Initial and boundary conditions for the 36- hour MM5 simulation are taken from NCEP Eta model analysis at 32 km resolution. Key issues to be explored are the settling of the observed anvil versus the model simulations, and comparisons of dynamical properties, such as vertical motions, occurring in the observations and models. The former provides an integrated measure of the validity of the model microphysics (fallspeed) while the latter is the key factor in forcing continued ice generation.

Lin, R.-F.↗

DOE OSTI2017

A Bottom's-Up Model for Siting and Developing Consent-Based Disposal Solutions for SNF and HLW in the USA - 17332

At the end of 2016, the licensing and continued development of the USA's only candidate HLW-repository since 1987 at the Yucca Mountain (YM) site in Nevada had been on hold since 2010 pending the enactment of enabling legislation for one or more of the following paths forward: 1. Re-start of the YM HLW repository. 2. Consent-based siting and development (S and D) of new HLW-repositories for commingled disposal of commercially- (CHLW) and defense-generated (DHLW) HLW. 3. Consent-based S and D of a repository and/or deep boreholes for DHLW only. Public trust in the implementing and regulating entities and a definitive say in an apolitical S and D process by the directly affected parties (DAPs) are historical imperatives for gaining public acceptance and maintaining sustained support of any or all of them. However, due to the widespread negative legacy currently attributed to past politically-controlled S and D organizations, an even more politically controlled organizational structure and an amorphous, 'consent-based', S and D model like that introduced twice in proposed Senate bills since June 2013 will exacerbate, rather than mitigate, the main root causes to past contentions, lawsuits, delays, and cost-increases. We therefore recommend: 1) Assigning the responsibility for safe disposal of CHLW to an independent, chartered public corporation with requisite subject-matter expertise and adequate financial resources. 2) Enacting a related law that: a. Defines 'consent-based' in measurable terms; b. Assigns irrevocable veto rights to facility host state(s) based upon the majority opinion of its legal residents at pre-determined S and D milestones. c. Requires Congressional approval before the facility can open and close. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

DOE OSTI2011

Collapse and revival dynamics of number-squeezed superfluids of ultracold atoms in optical lattices

Recent experiments have shown a remarkable number of collapse and revival oscillations of the matter-wave coherence of ultracold atoms in optical lattices [Will et al., Nature (London) 465, 197 (2010)]. Using a mean-field approximation to the Bose-Hubbard model, we show that the visibility of collapse and revival interference patterns reveals number squeezing of the initial superfluid state. To describe the dynamics, we use an effective Hamiltonian that incorporates the intrinsic two-body and induced three-body interactions, and we analyze in detail the resulting complex pattern of collapse and revival frequencies generated by virtual transitions to higher bands, as a function of lattice parameters and mean-atom number. Our work shows that a combined analysis of both the multiband, nonstationary dynamics in the final deep lattice and the number squeezing of the initial superfluid state explains important characteristics of optical lattice collapse and revival physics. Finally, by treating the two- and three-body interaction strengths and the coefficients describing the initial superposition of number states as free parameters in a fit to the experimental data, it should be possible to go beyond some of the limitations of our model and obtain insight into the breakdown of the mean-field theory for the initial state or the role of nonperturbative effects in the final-state dynamics.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

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

Learning model combining convolutional deep neural network with a self-attention mechanism for AC optimal power flow

Alternating current optimal power flow (OPF) analysis is critical for efficient and reliable operation of power systems. For large systems or repetitive computations, the traditional methods such as the direct and gradient methods, or non-traditional methods, such as the genetic algorithm and simulating annealing, are time-consuming and unsuitable for real-time computing. The work in this paper proposes a novel framework to obtain the optimal solution of power flow in real-time using a combination of convolutional neural networks and a self-attention mechanism. All parameters of the power networks are rearranged in an image-like shape of a multi-channel image where each channel is a two-dimensional matrix. The proposed approach is adaptive with every input size of power systems as well as frequent variations of network topologies without intervention to the framework core. The encompassment of all power system contexts in which all parameters of internal elements, generation costs, and topology information are included, contributes to the higher accuracy of inference compared to other current machine-learning-based OPF-solving methods. Besides, the proposed framework established on ubiquitous platforms is effortlessly integrated into current infrastructures of power systems, and the great efficiency along with the computation speed may serve as a critical point for practical implications, such as enabling faster decision-making during real-time operations, predicting system contingencies, and remedial actions based on an offline pre-trained model. Furthermore, this supervised learning process is applied to the dataset of four case studies of meshed power systems: the IEEE 5-bus system (IEEE-5), the IEEE 30-bus system (IEEE-30), the IEEE 39-bus system (IEEE-39), and the IEEE 57-bus system (IEEE-57) to prove the efficacy of the proposed method.

42 ENGINEERING↗

DOE OSTI2022

Extracting low energy signals from raw LArTPC waveforms using deep learning techniques — A proof of concept

Here, we investigate the feasibility of using deep learning techniques, in the form of a one-dimensional convolutional neural network (1D-CNN), for the extraction of signals from the raw waveforms produced by the individual channels of liquid argon time projection chamber (LArTPC) detectors. A minimal generic LArTPC detector model is developed to generate realistic noise and signal waveforms used to train and test the 1D-CNN, and evaluate its performance on low-level signals. We demonstrate that our approach overcomes the inherent shortcomings of traditional cut-based methods by extending sensitivity to signals with ADC values below their imposed thresholds. This approach exhibits great promise in enhancing the capabilities of future generation neutrino experiments like DUNE to carry out their low-energy neutrino physics programs.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

DOE OSTI2023

Electron scattering and neutrino physics

Abstract A thorough understanding of neutrino–nucleus scattering physics is crucial for the successful execution of the entire US neutrino physics program. Neutrino–nucleus interaction constitutes one of the biggest systematic uncertainties in neutrino experiments—both at intermediate energies affecting long-baseline deep underground neutrino experiment, as well as at low energies affecting coherent scattering neutrino program—and could well be the difference between achieving or missing discovery level precision. To this end, electron–nucleus scattering experiments provide vital information to test, assess and validate different nuclear models and event generators intended to test, assess and validate different nuclear models and event generators intended to be used in neutrino experiments. Similarly, for the low-energy neutrino program revolving around the coherent elastic neutrino–nucleus scattering (CEvNS) physics at stopped pion sources, such as at ORNL, the main source of uncertainty in the evaluation of the CEvNS cross section is driven by the underlying nuclear structure, embedded in the weak form factor, of the target nucleus. To this end, parity-violating electron scattering (PVES) experiments, utilizing polarized electron beams, provide vital model-independent information in determining weak form factors. This information is vital in achieving a percent level precision needed to disentangle new physics signals from the standard model expected CEvNS rate. In this white paper, we highlight connections between electron- and neutrino–nucleus scattering physics at energies ranging from 10 s of MeV to a few GeV, review the status of ongoing and planned electron scattering experiments, identify gaps, and lay out a path forward that benefits the neutrino community. We also highlight the systemic challenges with respect to the divide between the nuclear and high-energy physics communities and funding that presents additional hurdles in mobilizing these connections to the benefit of neutrino programs.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

DOE OSTI2022

Field Emission Mitigation in CEBAF SRF Cavities Using Deep Learning

The Continuous Electron Beam Accelerator Facility (CEBAF) operates hundreds of superconducting radio frequency (SRF) cavities in its two main linear accelerators. Field emission can occur when the cavities are set to high operating RF gradients and is an ongoing operational challenge. This is especially true in newer, higher gradient SRF cavities. Field emission results in damage to accelerator hardware, generates high levels of neutron and gamma radiation, and has deleterious effects on CEBAF operations. So, field emission reduction is imperative for the reliable, high gradient operation of CEBAF that is required by experimenters. Here we explore the use of deep learning architectures via multilayer perceptron to simultaneously model radiation measurements at multiple detectors in response to arbitrary gradient distributions. These models are trained on collected data and could be used to minimize the radiation production through gradient redistribution. This work builds on previous efforts in developing machine learning (ML) models, and is able to produce similar model performance as our previous ML model without requiring knowledge of the field emission onset for each cavity.

Ahammed, K.↗

DOE OSTI2023

EPiC-GAN: Equivariant point cloud generation for particle jets

With the vast data-collecting capabilities of current and future high-energy collider experiments, there is an increasing demand for computationally efficient simulations. Generative machine learning models enable fast event generation, yet so far these approaches are largely constrained to fixed data structures and rigid detector geometries. In this paper, we introduce EPiC-GAN - equivariant point cloud generative adversarial network - which can produce point clouds of variable multiplicity. This flexible framework is based on deep sets and is well suited for simulating sprays of particles called jets. The generator and discriminator utilize multiple EPiC layers with an interpretable global latent vector. Crucially, the EPiC layers do not rely on pairwise information sharing between particles, which leads to a significant speed-up over graph- and transformer-based approaches with more complex relation diagrams. We demonstrate that EPiC-GAN scales well to large particle multiplicities and achieves high generation fidelity on benchmark jet generation tasks.

Buhmann, Erik↗

DOE OSTI2022

Field Emission Mitigation in CEBAF SRF Cavities Using Deep Learning

Ahammed, K.↗

DOE OSTI2024

Hector-FACTS Technical Report & User Guide: Integrating Hector into the Framework for Assessing Changes To Sea-level

The Framework for Assessing Changes To Sea-level (FACTS) platform was developed by (Kopp et al., 2023) as an open-source modular tool capable of generating probabilistic projections for global mean sea-level change and future relative and extreme sea-level change at New York City (Kopp et al., 2023). FACTS is the first open-source modular tool designed to explore deep uncertainty in SLR while also synthesizing across different modeling groups and integrating the various components of SLR used in key climate reports.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2025

Applications of Nickelate perovskites for neuromorphic computing from electronic structure and Machine Learning

While the limit of Moore's law is presently being reached with current microelectronic technologies, we need to develop new paradigms that overcome this limitation. In that respect, neuromorphic computing is a concept that emulates the neural behavior and response of the human brain, and it has been recognized as a promising alternative approach. In this research project, we will perform multi-fidelity scale bridging to explore the potential use of materials with metal to insulator transition for neuromorphic applications. In particular, rare earth nickelates are promising for such purposes, as the transition in these materials is quite sensitive to a broad set of different external stimuli. Our multi-fidelity approach will bridge the high-fidelity electronic structure calculations with classical potentials. We will bridge dynamical mean field theory with a classical atomistic representation via a deep learning force field. The neural network is trained with energies, charges, and forces obtained by accurate electronic structure theories based on Dynamical Mean Field Theory. The configurational space is generated from known crystal phases, ab initio molecular dynamics with exchange-correlation functionals corrected with the Hubbard model, disordered phases with different concentrations of oxygen vacancies, and nonsymmetrical positions and induced strain by grain interfaces or contact with a substrate. Strategies to train the model with a reduced number of training examples are obtained from active learning methods, and new structures for improving the learning process are generated by using machine learning autoencoders. This classical potential will be validated through a diversity of electronic structure methods and represents an important step to combine the flexibility and accuracy of first-principles with the speed of classical potentials. The generated multi-fidelity surrogate model will be used to understand the role of strain, oxygen vacancies, proton doping, the variation of the crystal phase, substrate effects, vibrational effects as the octahedral rotation, grain boundaries and defect effects on the response of a Metal to Insulator Transition (MIT) in correlated materials. Long time and large-scale simulations will help understand the role of different stimuli to control the hysteresis of the MIT, as it has been experimentally suggested. Selected configurations will be analyzed with higher-level theories to provide an accurate electronic description and to study how the orbitals and charges are rearranged under different conditions.

36 MATERIALS SCIENCE↗

DOE OSTI2024

The Deep Underground Neutrino Experiment (DUNE) Program

The Deep Underground Neutrino Experiment (DUNE) is a next-generation long-baseline neutrino oscillation experiment aimed at determining the neutrino mass hierarchy and the CP-violating phase. The DUNE physics program also includes the detection of astrophysical neutrinos and the search for signatures beyond the Standard Model, such as nucleon decays. DUNE consists of a near detector complex located at Fermilab and four 17 kton Liquid Argon Time Projection Chamber (LArTPC) far detector modules to be built 1.5 km underground at SURF, approximately 1300 km away. The detectors are exposed to a wideband neutrino beam generated by a 1.2 MW proton beam with a planned upgrade to $>$ 2 MW. Two 770 ton LArTPCs (ProtoDUNEs) have been operated at CERN for over 2 years as a testbed for DUNE far detectors and have been optimized to take new cosmic and test-beam data in 2024-2025. The DUNE and ProtoDUNE experiments and physics goals, as well as recent progress and results, are presented.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗