Search NASA⌕ Search

SEARCH · Search NASA

Results for “operator inference”

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 811 records · Page 45

Large-Scale Optimization for Bayesian Inference in Complex Systems

The SAGUARO (Scalable Algorithms for Groundwater Uncertainty Analysis and Robust Optimization) Project focused on the development of scalable numerical algorithms for large-scale Bayesian inversion in complex systems that capitalize on advances in large-scale simulation-based optimization and inversion methods. The project was a collaborative effort among MIT, the University of Texas at Austin, Georgia Institute of Technology, and Sandia National Laboratories. The research was directed in three complementary areas: efficient approximations of the Hessian operator, reductions in complexity of forward simulations via stochastic spectral approximations and model reduction, and employing large-scale optimization concepts to accelerate sampling. The MIT₋Sandia component of the SAGUARO Project addressed the intractability of conventional sampling methods for large-scale statistical inverse problems by devising reduced-order models that are faithful to the full-order model over a wide range of parameter values; sampling then employs the reduced model rather than the full model, resulting in very large computational savings. Results indicate little effect on the computed posterior distribution. On the other hand, in the Texas₋Georgia Tech component of the project, we retain the full-order model, but exploit inverse problem structure (adjoint-based gradients and partial Hessian information of the parameter-to-observation map) to implicitly extract lower dimensional information on the posterior distribution; this greatly speeds up sampling methods, so that fewer sampling points are needed. We can think of these two approaches as "reduce then sample" and "sample then reduce." In fact, these two approaches are complementary, and can be used in conjunction with each other. Moreover, they both exploit deterministic inverse problem structure, in the form of adjoint-based gradient and Hessian information of the underlying parameter-to-observation map, to achieve their speedups.

97 MATHEMATICS AND COMPUTING↗

Toward an Autonomous Workflow for Single Crystal Neutron Diffraction

The operation of the neutron facility relies heavily on beamline scientists. Some experiments can take one or two days with experts making decisions along the way. Leveraging the computing power of HPC platforms and AI advances in image analyses, here we demonstrate an autonomous workflow for the single-crystal neutron diffraction experiments. The workflow consists of three components: an inference service that provides real-time AI segmentation on the image stream from the experiments conducted at the neutron facility, a continuous integration service that launches distributed training jobs on Summit to update the AI model on newly collected images, and a frontend web service to display the AI tagged images to the expert. Ultimately, the feedback can be directly fed to the equipment at the edge in deciding the next-step experiment without requiring an expert in the loop. With the analyses of the requirements and benchmarks of the performance for each component, this effort serves as the first step toward an autonomous workflow for real-time experiment steering at ORNL neutron facilities.

Yin, Junqi↗

Design of a new charge exchange recombination spectroscopy diagnostic for impurity transport experiments at Wendelstein 7-X

In this study, we present the design of a new charge exchange recombination spectroscopy (CXRS) system for Wendelstein 7-X (W7-X), which aims at measuring line radiation from highly ionized iron impurities after laser blow off injections over transport timescales. New fiber bundles with a core diameter of 600 µm have been added to one of the existing optical systems at W7-X used for CXRS. The fibers direct collected light to five newly developed single channel high frame rate F/2.8 spectrometers. The new custom-made spectrometers are built using entirely off-the-shelf and 3D printed components, and employ a round-to-linear fiber bundle, instead of an entrance slit, to maximize light throughput. The detector is an ultra-low readout noise EM-CCD camera capable of frame rates of up to 10 000 fps for about 500 consecutive frames when operated using a single readout channel. Such high frame rates will enable the study of inward convection of injected impurities. An initial sensitivity study is performed using a newly developed Markov chain Monte Carlo approach based on the pySTRAHL impurity transport code. This study indicates that, with the addition of the new CX diagnostic, impurity convection velocities can be inferred.

Instruments & Instrumentation↗

On the synchrotron emission in kinetic simulations of runaway electrons in magnetic confinement fusion plasmas

Developing avoidance or mitigation strategies of runaway electrons (REs) in magnetic confinement fusion (MCF) plasmas is of crucial importance for the safe operation of ITER. In order to develop these strategies, an accurate diagnostic capability that allows good estimates of the RE distribution function in these plasmas is needed. Synchrotron radiation (SR) of RE in MCF, besides of being one of the main damping mechanisms for RE in the high energy relativistic regime, is routinely used in current MCF experiments to infer the parameters of RE energy and pitch angle distribution functions. In the present paper we address the long standing question about what are the relationships between different REs distribution functions and their corresponding synchrotron emission simultaneously including: full-orbit effects, information of the spectral and angular distribution of SR of each electron, and basic geometric optics of a camera. We study the spatial distribution of the SR on the poloidal plane, and the statistical properties of the expected value of the synchrotron spectra of REs. Here we observe a strong dependence of the synchrotron emission measured by the camera on the pitch angle distribution of runaways, namely we find that crescent shapes of the spatial distribution of the SR as measured by the camera relate to RE distributions with small pitch angles, while ellipse shapes relate to distributions of runaways with larger the pitch angles. A weak dependence of the synchrotron emission measured by the camera with the RE energy, value of the q-profile at the edge, and the chosen range of wavelengths is observed. Furthermore, we find that oversimplifying the angular dependence of the SR changes the shape of the synchrotron spectra, and overestimates its amplitude by approximately 20 times for avalanching runaways and by approximately 60 times for mono-energetic distributions of runaways.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Facet effects on generation-recombination currents in semiconductor laser diodes

The contribution of facet defect currents to the overall generation-recombination current of laser diodes operating near 800 nm is quantified experimentally, using the dependence of current on cavity length to isolate facet effects. Here the results show that facet currents exhibit an ideality factor much greater than 2, while currents associated with the interior of the laser diode stripes exhibit an ideality factor of 2. These differences in behavior provide an approach to infer additional details of defect evolution in aging studies of semiconductor laser diodes.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Measurements of emittance at low energy

Ideally, experimental measurements of all beam parameters should be made at the exit of the injector. This has not been possible on DARHT Axis-II for various practical reasons. In particular, the emittance has not been measured there. This note reviews the methods we have used for emittance measurements on the linear induction accelerator (LIA) at low-energy (< 10 MeV). It includes a discussion of some of the issues associated with making those measurements on a beam from the injector. A successful plan for emittance measurements at any energy should incorporate measures to deal with these issues. To begin with, it should be recognized that emittance measurements at the injector are not critical for beam physics in the accelerator. The emittance at the injector has very little effect on accelerator beam dynamics except for an increase in size near the end, where the beam becomes emittance dominated. Therefore, imperfect knowledge has little effect on the design and implementation of a tune. This point is illustrated in Fig. 1, which shows beam envelope calculations for our present tune over a wide range of emittances. Compare this with using the wrong diode voltage in the tune design. Figure 2 shows the effect on the beam envelope of a 2% error in the measurement of diode voltage. Therefore, a few years ago, when we made plans for beam measurements at the diode while we were upgrading the injector cells, our highest priority was on magnetic spectrometer measurements of energy. These measurements were to be used to cross calibrate the diode-voltage monitor, so that accurate diode voltage measurements could be made on every shot after the spectrometer was removed. If emittance measurements at the injector are required, then three widely used methods should be considered. These are focusing magnet scans, the use of an emittance mask (pepper pot), and optical transition radiation (OTR). We have used all three of these on our accelerators. These are invasive methods. There are noninvasive methods that we have not yet used. For example, under some conditions the beam size obtained with diamagnetic loops can be used in conjunction with the envelope equation to infer emittance. To date our diamagnetic loop data has not been of sufficient quality to do this. Another technique is to measure the beam-distribution quadrupole moment with BPMs dispersed in a quadrupole magnet lattice. Our BPMs have this capability, but the only quad lattice is in the downstream transport (DST), so this method is deferred until we operate the DST. All three methods reviewed here are invasive, and require the insertion of a target into the beam. Therefore, there is always the accompanying risk of inadvertent destruction of the target, and scattering of target material throughout the accelerator or injector. The target damage mechanism is target heating to a temperature hot enough to evolve gas, which is impact ionized by the beam, which is then ion focused to a small enough spot for the energy density to damage the target material, often destructively. Target heating is directly proportional to beam current density and pulse length. Therefore, the key to minimizing risk of target damage is to design for a large beam at the target, low current from the diode, and the shortest possible current pulse. The large-beam requirement compromises emittance measurements with OTR or focal scans, but not with the pepper pot technique. Finally, to ensure a short pulse even if the crowbar fails, the Marx generator should be reconfigured to produce the shortest possible pulse.

43 PARTICLE ACCELERATORS↗

Systematic Method for Evaluating Extraction and Injection Flow Rates for 100-KR-4 and 100-HR-3 Groundwater Operable Unit Pump-and-Treat Interim Actions for Hydraulic Containment

This document describes a systematic method to develop flow rate recommendations for Pump-and-Treat (P&T) extraction and injection wells in 100-KR-4 and 100-HR-3 Groundwater Operable Units (OU) of the Hanford Site. Flow rate recommendations are developed as part of ongoing performance monitoring and remedy optimization of the P&T interim actions to develop hydraulic containment of the dissolved chromium plume in groundwater and protect the Columbia River from further discharges of groundwater from inland. This document details the methodology and data required to infer the influence of individual wells near the shoreline on hydraulic containment and river protection and develop flow rate recommendations to improve system performance and mitigate potential shortcomings of the system configuration in place.

54 ENVIRONMENTAL SCIENCES↗

Program for Experimentation With Expert Systems

CERBERUS is forward-chaining, knowledge-based system program useful for experimentation with expert systems. Inference-engine mechanism performs deductions according to user-supplied rule set. Information stored in intermediate area, and user interrogated only when no applicable data found in storage. Each assertion posed by CERBERUS answered with certainty ranging from 0 to 100 percent. Rule processor stops investigating applicable rules when goal reaches certainty of 95 percent or higher. Capable of operating for wide variety of domains. Sample rule files included for animal identification, pixel classification in image processing, and rudimentary car repair for novice mechanic. User supplies set of end goals or actions. System complexity decided by user's rule file. CERBERUS written in FORTRAN 77.

Engle, S. W.↗

Large Eddy Simulation Study for Fluid Disintegration and Mixing

A new modeling approach is based on the concept of large eddy simulation (LES) within which the large scales are computed and the small scales are modeled. The new approach is expected to retain the fidelity of the physics while also being computationally efficient. Typically, only models for the small-scale fluxes of momentum, species, and enthalpy are used to reintroduce in the simulation the physics lost because the computation only resolves the large scales. These models are called subgrid (SGS) models because they operate at a scale smaller than the LES grid. In a previous study of thermodynamically supercritical fluid disintegration and mixing, additional small-scale terms, one in the momentum and one in the energy conservation equations, were identified as requiring modeling. These additional terms were due to the tight coupling between dynamics and real-gas thermodynamics. It was inferred that if these terms would not be modeled, the high density-gradient magnitude regions, experimentally identified as a characteristic feature of these flows, would not be accurately predicted without the additional term in the momentum equation; these high density-gradient magnitude regions were experimentally shown to redistribute turbulence in the flow. And it was also inferred that without the additional term in the energy equation, the heat flux magnitude could not be accurately predicted; the heat flux to the wall of combustion devices is a crucial quantity that determined necessary wall material properties. The present work involves situations where only the term in the momentum equation is important. Without this additional term in the momentum equation, neither the SGS-flux constant-coefficient Smagorinsky model nor the SGS-flux constant-coefficient Gradient model could reproduce in LES the pressure field or the high density-gradient magnitude regions; the SGS-flux constant- coefficient Scale-Similarity model was the most successful in this endeavor although not totally satisfactory. With a model for the additional term in the momentum equation, the predictions of the constant-coefficient Smagorinsky and constant-coefficient Scale-Similarity models were improved to a certain extent; however, most of the improvement was obtained for the Gradient model. The previously derived model and a newly developed model for the additional term in the momentum equation were both tested, with the new model proving even more successful than the previous model at reproducing the high density-gradient magnitude regions. Several dynamic SGS-flux models, in which the SGS-flux model coefficient is computed as part of the simulation, were tested in conjunction with the new model for this additional term in the momentum equation. The most successful dynamic model was a "mixed" model combining the Smagorinsky and Gradient models. This work is directly applicable to simulations of gas turbine engines (aeronautics) and rocket engines (astronautics).

Bellan, Josette↗

Quantifying the Hypervolatile Abundances in Jupiter-family Comet 46P/Wirtanen

We present analysis of IR and optical spectroscopy of Jupiter-family comet (JFC) 46P/Wirtanen obtained in 2019 January, when the comet had sufficient geocentric velocity to enable studies of the hypervolatiles CO and CH4, as well as [OI] emission. These species could not be studied near closest approach in mid-December because there was insufficient Doppler shift to separate the cometary emission from their corresponding telluric absorption lines. We employed the [OI] observations as a proxy for the CO2/H2O ratio, as CO2 cannot be observed directly from the ground, and space-based assets sensitive to CO2 were not able to observe 46P during this apparition. We focused our analysis on H2O, CO, CH4, C2H6, CH3OH, and CO2 (via [OI] emission). We detected strong emissions from H2O, C2H6, and CH3OH. Over the 3 nights, we found evidence for changing mixing ratios, mostly due to a variable H2O production rate. In 46P, C2H6 and CH3OH are enriched compared to cometary averages, with mixing ratios relative to H2O of ∼1% and ∼3%, respectively. Measurements of CH4 and CO have been especially rare in JFCs. We report significant 3σ upper limits on CH4/H2O<0.97% and CO/H2O<0.54%. They place CH4 being near-average or depleted, and CO being strongly depleted in 46P compared with Oort cloud comets. 46P has comparable CO/H2O to the few other measurements in JFCs, but enriched in C2H6 and CH3OH. Our inferred CO2/H2O mixing ratio is∼15%, though accounting for systematic uncertainties from the lack of knowledge of [OI] photochemistry means a value between 10% and 20% is likely. The compositional profile of46P is similar to another small, hyperactive comet: 103P/Hartley 2. The mechanism of CO2-driven water-rich ice grain production proposed for 103P/Hartley 2 may be operating on 46P as well.

Comets↗

3D Cloud Tomography and Droplet Size Retrieval from Multi-Angle Polarimetric Imaging of Scattered Sunlight from Above

Tomography aims to recover a three-dimensional (3D) density map of a medium or an object. In medical imaging,it is extensively used for diagnostics via X-ray computed tomography (CT). We define and derive a tomographyof cloud droplet distributions via passive remote sensing. We use multi-view polarimetric images to fit a 3Dpolarized radiative transfer (RT) forward model. Our motivation is 3D volumetric probing of vertically-developedconvectively-driven clouds that are ill-served by current methods in operational passive remote sensing. Currenttechniques are indeed based on strictly 1D RT modeling and applied to a single cloudy pixel, where cloud geometrydefaults to that of a plane-parallel slab. Incident unpolarized sunlight, once scattered by cloud droplets, changesits polarization state according to droplet size. Therefore, polarimetric measurements in the rainbow and gloryangular regions can be used to infer the droplet size distribution. This work defines and derives a framework for afull 3D tomography of cloud droplets for both their mass concentration in space and their distribution across arange of sizes. This gridded 3D retrieval of key microphysical properties is made tractable by our novel approachthat involves a restructuring and partial linearization of an open-source polarized 3D RT code to accommodate aspecial two-step iterative optimization technique. Physically-realistic synthetic clouds are used to demonstrate themethodology with rigorous uncertainty quantification, while a real-world cloud imaged by AirMSPI is processedto illustrate the new remote sensing capability

Schechner, Yoav Y.↗

Characterizing Regional Methane Emissions from Natural Gas Liquids Unloading

A “bottom-up” probabilistic model was developed using engineering first-principles to quantify annualized throughput normalized methane emissions (TNME) from natural gas liquid unloading activities for 18 basins in the United States in 2016. For each basin, six discrete liquid-unloading scenarios are considered, consisting of combinations of well types (conventional and unconventional) and liquid-unloading systems (nonplunger, manual plunger lift, and automatic plunger lift). While previous work has advocated that operational malfunctions and abnormal process conditions explain the existence of super-emitters in the natural gas supply chain, this work finds that super-emitters can arise naturally due to variability in underlying component processes. Additionally, average cumulative methane emissions from liquids unloading, attributed to the natural gas supply chain, across all basins are ~4.8 times higher than those inferred from the 2016 Greenhouse Gas Reporting Program (GHGRP).

Zaimes, George G.↗

Machine learning pipeline for denoising low signal-to-noise ratio and out-of-distribution transmission electron microscopy datasets

High-resolution transmission electron microscopy (HRTEM) is crucial for observing material’s structural and morphological evolution at Angstrom scales, but the electron beam can alter these processes. Devices such as CMOS-based direct-electron detectors operating in electron-counting mode can be utilized to substantially reduce the electron dosage. However, the resulting images often lead to a low signal-to-noise ratio, which requires frame integration that sacrifices temporal resolution. Several machine learning (ML) models have been recently developed to successfully denoise HRTEM images. Yet, these models are often computationally expensive, and their inference speeds on GPUs are outpaced by the imaging speed of advanced detectors, precluding in situ analysis. Furthermore, the performance of these denoising models on datasets with imaging conditions that deviate from the training datasets has not been evaluated. To mitigate these gaps, we propose a new self-supervised ML denoising pipeline specifically designed for time-series HRTEM images. This pipeline integrates a blind-spot convolution neural network with pre-processing and post-processing steps, including drift correction and low-pass filtering. Results demonstrate that our model outperforms various other ML and non-ML denoising methods in noise reduction and contrast enhancement, leading to improved visual clarity of atomic features. Additionally, the model is drastically faster than U-Net-based ML models and demonstrates excellent out-of-distribution generalization. The model’s computational inference speed is in the order of milliseconds per image, rendering it suitable for application in in-situ HRTEM experiments.

36 MATERIALS SCIENCE↗

Adversarial super-resolution of climatological wind and solar data

Accurate and high-resolution data reflecting different climate scenarios are vital for policy makers when deciding on the development of future energy resources, electrical infrastructure, transportation networks, agriculture, and many other societally important systems. However, state-of-the-art long-term global climate simulations are unable to resolve the spatiotemporal characteristics necessary for resource assessment or operational planning. We introduce an adversarial deep learning approach to super resolve wind velocity and solar irradiance outputs from global climate models to scales sufficient for renewable energy resource assessment. Using adversarial training to improve the physical and perceptual performance of our networks, we demonstrate up to a 50 × resolution enhancement of wind and solar data. In validation studies, the inferred fields are robust to input noise, possess the correct small-scale properties of atmospheric turbulent flow and solar irradiance, and retain consistency at large scales with coarse data. An additional advantage of our fully convolutional architecture is that it allows for training on small domains and evaluation on arbitrarily-sized inputs, including global scale. We conclude with a super-resolution study of renewable energy resources based on climate scenario data from the Intergovernmental Panel on Climate Change’s Fifth Assessment Report.

14 SOLAR ENERGY↗

A comprehensive final summary of the MCNP calculations performed concerning the UW 137 Cs release event

This document combines and supersedes earlier reports (e.g. LA-UR-19-32358 and LA-UR-20- 22216) that discussed MCNP-calculated conversion factors (count rate per unit activity per unit area) for collimated 2”x2” NaI detectors (Ludlum model 44-10 and Thermo model SPA3) in contact with various surfaces of interest to the University of Washington 137 Cs release event. These surfaces include: Exterior surface of round and rectangular ducts, Exterior surface of outdoor brick work, Exterior surface of cast iron pipes, Exterior surface of an O 2 filter assembly For each surface, the calculated conversion factors allowed inferences on 137 Cs activity on interior surfaces based on exterior contact measurements of the 662 keV photopeak count rate. The MCNP calculations included the effect of attached collimators (i.e. either a Pb/Sb-based Ludlum or a W-based “Rees” model) on detector response. As a consequence, the conversion factors only apply to the appropriate collimated 2”x2” NaI detector operated in pulse height analysis (PHA) mode. Other assumptions affecting the calculations and valid application of the conversion factors will be outlined below.

54 ENVIRONMENTAL SCIENCES↗

Distance-Weighted Graph Neural Networks on FPGAs for Real-Time Particle Reconstruction in High Energy Physics

Graph neural networks have been shown to achieve excellent performance for several crucial tasks in particle physics, such as charged particle tracking, jet tagging, and clustering. An important domain for the application of these networks is the FGPA-based first layer of real-time data filtering at the CERN Large Hadron Collider, which has strict latency and resource constraints. We discuss how to design distance-weighted graph networks that can be executed with a latency of less than one μs on an FPGA. To do so, we consider a representative task associated to particle reconstruction and identification in a next-generation calorimeter operating at a particle collider. We use a graph network architecture developed for such purposes, and apply additional simplifications to match the computing constraints of Level-1 trigger systems, including weight quantization. Using the hls4ml library, we convert the compressed models into firmware to be implemented on an FPGA. Performance of the synthesized models is presented both in terms of inference accuracy and resource usage.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

FINDS: A fault inferring nonlinear detection system programmers manual, version 3.0

Detailed software documentation of the digital computer program FINDS (Fault Inferring Nonlinear Detection System) Version 3.0 is provided. FINDS is a highly modular and extensible computer program designed to monitor and detect sensor failures, while at the same time providing reliable state estimates. In this version of the program the FINDS methodology is used to detect, isolate, and compensate for failures in simulated avionics sensors used by the Advanced Transport Operating Systems (ATOPS) Transport System Research Vehicle (TSRV) in a Microwave Landing System (MLS) environment. It is intended that this report serve as a programmers guide to aid in the maintenance, modification, and revision of the FINDS software.

Lancraft, R. E.↗