Search NASA⌕ Search

SEARCH · Search NASA

Results for “Performance Optimized Space Networks”

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 235 records · Page 13
All sourcesNASA NTRSDOE OSTI
NASA NTRS1975

Sasaki's variational optimization analysis of temperature and moisture advection in a severe storm environment

Temperature and moisture advection patterns of dry air intrusion were analyzed by a variational optimization method to relate mesoscale advection to a severe thunderstorm environment. The analysis results presented here were obtained from a case study, May 26, 1973, when a tornado squall line developed in Oklahoma in conjunction with a clearly delineated dry air intrusion. Observations from the synoptic upper air network were used for inputs of horizontal wind components, potential temperature, and mixing ratio at each 50 mb increment from 950 to 100 mb. An objective analysis was performed to assign values of the variables to grid points of a 15 x 14 horizontal grid with approximate 125 km grid spacing at each pressure level. After the analysis the data fields were smoothed using a variational formalism.

Mcfarland, M. J.↗
NASA NTRS1992

ACTS Multibeam Communications Package - Description and Performance Characterization

To maintain its competitive position in the 1990s, satellite communications must evolve to provide new services and to reduce the costs of supplying existing services. NASA's Advanced Communications Technology Satellite (ACTS) program is directed toward developing the technology base that will support this evolution. The ACTS communications payload is optimized for digital communications using a time-division multiple-access format. Scanning spot-beam antennas provide high-performance, on-demand service to communications earth stations. On-board processing and switching allow full connectivity in a single-hop network architecture. ACTS employs Ka-Band (30/20 GHz) links for communications traffic and includes techniques for mitigating rain fades. The prelaunch test program has verified that all critical payload performance requirements have been realized, including G/T, EIRP, beam coverage, AM/AM, AM/PM, and BER vs. E(b)/N(o). Scheduled to be launched by a Space Shuttle in 1993, ACTS will provide in-orbit technology verification and support experimental communications networks for up to four years.

Cashman, William F.↗
DOE OSTI2022

Systematic quark/gluon identification with ratios of likelihoods

Discriminating between quark- and gluon-initiated jets has long been a central focus of jet substructure, leading to the introduction of numerous observables and calculations to high perturbative accuracy. At the same time, there have been many attempts to fully exploit the jet radiation pattern using tools from statistics and machine learning. We propose a new approach that combines a deep analytic understanding of jet substructure with the optimality promised by machine learning and statistics. After specifying an approximation to the full emission phase space, we show how to construct the optimal observable for a given classification task. This procedure is demonstrated for the case of quark and gluons jets, where we show how to systematically capture sub-eikonal corrections in the splitting functions, and prove that linear combinations of weighted multiplicity is the optimal observable. In addition to providing a new and powerful framework for systematically improving jet substructure observables, we demonstrate the performance of several quark versus gluon jet tagging observables in parton-level Monte Carlo simulations, and find that they perform at or near the level of a deep neural network classifier. Combined with the rapid recent progress in the development of higher order parton showers, we believe that our approach provides a basis for systematically exploiting subleading effects in jet substructure analyses at the Large Hadron Collider (LHC) and beyond.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2022

Automatic Generation of High-Performance Convolution Kernels on ARM CPUs for Deep Learning

In this work, we present FastConv, a template-based code auto-generation open source library that can automatically generate high-performance deep learning convolution kernels of arbitrary matrices/tensors shapes. FastConv is based on the Winograd algorithm, which is reportedly the highest performing algorithm for the time-consuming convolution layers of convolutional neural networks. ARM CPUs cover a wide range designs and specifications, from embedded devices to HPC-grade CPUs. The leads to the dilemma of how to consistently optimize Winograd-based convolution solvers for convolution layers of different shapes. FastConv addresses this problem by using templates to auto-generate multiple shapes of tuned kernels variants suitable for skinny tall matrices. As a performance portable library, FastConv transparently searches for the best combination of kernel shapes, cache tiles, scheduling of loop orders, packing strategies, access patterns, and online/offline computations. Auto-tuning is used to search the parameter configuration space for the best performance for a given target architecture and problem size. The experiments with layer-wise evaluation on the VGG--16 model confirms a 1.25x performance gains is got by tuning the Winograd library. Integrated comparison results shows 1.02x to 1.40x, 1.14x to 2.17x, and 1.22x and 2.48x speedup is achieved over NNPACK, Arm NN, and FeatherCNN on the Kunpeng 920 beside few cases. Furthermore, problem size performance portability experiments with various convolution shapes shows that FastConv achieves 1.2x to 1.7x speedup and 2x to 22x speedup over NNPACK and ARM NN inference engine using Winograd on Kunpeng 920 . CPU performance portability evaluation on the VGG--16 show an average speedup over NNPACK of 1.42x, 1.21x, 1.26x, 1.37x, 2.26x, and 11.02x is observed on Kunpeng 920, Snapdragon 835, 855, 888, Apple M1, and AWS Graviton2, respectively.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2017

Synthesis Study on Transitions in Signal Infrastructure and Control Algorithms for Connected and Automated Transportation

Documenting existing state of practice is an initial step in developing future control infrastructure to be co-deployed for heterogeneous mix of connected and automated vehicles with human drivers while leveraging benefits to safety, congestion, and energy. With advances in information technology and extensive deployment of connected and automated vehicle technology anticipated over the coming decades, cities globally are making efforts to plan and prepare for these transitions. CAVs not only offer opportunities to improve transportation systems through enhanced safety and efficient operations of vehicles. There are also significant needs in terms of exploring how best to leverage vehicle-to-vehicle (V2V) technology, vehicle-to-infrastructure (V2I) technology and vehicle-to-everything (V2X) technology. Both Connected Vehicle (CV) and Connected and Automated Vehicle (CAV) paradigms feature bi-directional connectivity and share similar applications in terms of signal control algorithm and infrastructure implementation. The discussion in our synthesis study assumes the CAV/CV context where connectivity exists with or without automated vehicles. Our synthesis study explores the current state of signal control algorithms and infrastructure, reports the completed and newly proposed CV/CAV deployment studies regarding signal control schemes, reviews the deployment costs for CAV/AV signal infrastructure, and concludes with a discussion on the opportunities such as detector free signal control schemes and dynamic performance management for intersections, and challenges such as dependency on market adaptation and the need to build a fault-tolerant signal system deployment in a CAV/CV environment. The study will serve as an initial critical assessment of existing signal control infrastructure (devices, control instruments, and firmware) and control schemes (actuated, adaptive, and coordinated-green wave). Also, the report will help to identify the future needs for the signal infrastructure to act as the nervous system for urban transportation networks, providing not only signaling, but also observability, surveillance, and measurement capacity. The discussion of the opportunities space includes network optimization and control theory perspectives, and the current states of observability for key system parameters (what can be detected, how frequently can it be reported) as well as controllability of dynamic parameters (this includes adjusting not only the signal phase and timing, but also the ability to alter vehicle trajectories through information or direct control). The perspective of observability and controllability of the dynamic systems provides an appropriate lens to discuss future directions as CAV/CV become more prevalent in the future.

33 ADVANCED PROPULSION SYSTEMS↗
NASA NTRS1999

Integrated Modeling Tools for Thermal Analysis and Applications

Integrated modeling of spacecraft systems is a rapidly evolving area in which multidisciplinary models are developed to design and analyze spacecraft configurations. These models are especially important in the early design stages where rapid trades between subsystems can substantially impact design decisions. Integrated modeling is one of the cornerstones of two of NASA's planned missions in the Origins Program -- the Next Generation Space Telescope (NGST) and the Space Interferometry Mission (SIM). Common modeling tools for control design and opto-mechanical analysis have recently emerged and are becoming increasingly widely used. A discipline that has been somewhat less integrated, but is nevertheless of critical concern for high precision optical instruments, is thermal analysis and design. A major factor contributing to this mild estrangement is that the modeling philosophies and objectives for structural and thermal systems typically do not coincide. Consequently the tools that are used in these discplines suffer a degree of incompatibility, each having developed along their own evolutionary path. Although standard thermal tools have worked relatively well in the past. integration with other disciplines requires revisiting modeling assumptions and solution methods. Over the past several years we have been developing a MATLAB based integrated modeling tool called IMOS (Integrated Modeling of Optical Systems) which integrates many aspects of structural, optical, control and dynamical analysis disciplines. Recent efforts have included developing a thermal modeling and analysis capability, which is the subject of this article. Currently, the IMOS thermal suite contains steady state and transient heat equation solvers, and the ability to set up the linear conduction network from an IMOS finite element model. The IMOS code generates linear conduction elements associated with plates and beams/rods of the thermal network directly from the finite element structural model. Conductances for temperature varying materials are accommodated. This capability both streamlines the process of developing the thermal model from the finite element model, and also makes the structural and thermal models compatible in the sense that each structural node is associated with a thermal node. This is particularly useful when the purpose of the analysis is to predict structural deformations due to thermal loads. The steady state solver uses a restricted step size Newton method, and the transient solver is an adaptive step size implicit method applicable to general differential algebraic systems. Temperature dependent conductances and capacitances are accommodated by the solvers. In addition to discussing the modeling and solution methods. applications where the thermal modeling is "in the loop" with sensitivity analysis, optimization and optical performance drawn from our experiences with the Space Interferometry Mission (SIM), and the Next Generation Space Telescope (NGST) are presented.

Milman, Mark H.↗
NASA NTRS2008

Analysis of a Dynamic Multi-Track Airway Concept for Air Traffic Management

The Dynamic Multi-track Airways (DMA) Concept for Air Traffic Management (ATM) proposes a network of high-altitude airways constructed of multiple, closely spaced, parallel tracks designed to increase en-route capacity in high-demand airspace corridors. Segregated from non-airway operations, these multi-track airways establish high-priority traffic flow corridors along optimal routes between major terminal areas throughout the National Airspace System (NAS). Air traffic controllers transition aircraft equipped for DMA operations to DMA entry points, the aircraft use autonomous control of airspeed to fly the continuous-airspace airway and achieve an economic benefit, and controllers then transition the aircraft from the DMA exit to the terminal area. Aircraft authority within the DMA includes responsibility for spacing and/or separation from other DMA aircraft. The DMA controller is responsible for coordinating the entry and exit of traffic to and from the DMA and for traffic flow management (TFM), including adjusting DMA routing on a daily basis to account for predicted weather and wind patterns and re-routing DMAs in real time to accommodate unpredicted weather changes. However, the DMA controller is not responsible for monitoring the DMA for traffic separation. This report defines the mature state concept, explores its feasibility and performance, and identifies potential benefits. The report also discusses (a) an analysis of a single DMA, which was modeled within the NAS to assess capacity and determine the impact of a single DMA on regional sector loads and conflict potential; (b) a demand analysis, which was conducted to determine likely city-pair candidates for a nationwide DMA network and to determine the expected demand fraction; (c) two track configurations, which were modeled and analyzed for their operational characteristic; (d) software-prototype airborne capabilities developed for DMA operations research; (e) a feasibility analysis of key attributes in the concept design; (f) a near-term, transitional application of the DMA concept as a proving ground for new airborne technologies; and (g) conclusions. The analysis indicates that the operational feasibility of a national DMA network faces significant challenges, especially for interactions between DMAs and between DMA and non-DMA traffic. Provided these issues are resolved, sectors near DMAs could experience significant local capacity benefits.

Wing, David J.↗
DOE OSTI2024

Role of depth in optical diffractive neural networks

Free-space all-optical diffractive neural networks have emerged as promising systems for neuromorphic scene classification. Understanding the fundamental properties of these systems is important to establish their ultimate performance. Here we consider the case of diffraction by subwavelength apertures and study the behavior of the system as a function of the number of diffractive layers by employing a co-design modeling approach. We show that adding depth allows the system to achieve high classification accuracies with a reduced number of diffractive features compared to a single layer, but that it does not allow the system to surpass the performance of an optimized single layer. The improvement from depth is found to be limited to the first few layers. These properties originate from the constraints imposed by the physics of light, in particular the weakening electric field with distance from the aperture.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
NASA NTRS2017

Medical Optimization Network for Space Telemedicine Resources

INTRODUCTION: Long-duration missions beyond low Earth orbit introduce new constraints to the space medical system such as the inability to evacuate to Earth, communication delays, and limitations in clinical skillsets. NASA recognizes the need to improve capabilities for autonomous care on such missions. As the medical system is developed, it is important to have an ability to evaluate the trade space of what resources will be most important. The Medical Optimization Network for Space Telemedicine Resources was developed for this reason, and is now a system to gauge the relative importance of medical resources in addressing medical conditions. METHODS: A list of medical conditions of potential concern for an exploration mission was referenced from the Integrated Medical Model, a probabilistic model designed to quantify in-flight medical risk. The diagnostic and treatment modalities required to address best and worst-case scenarios of each medical condition, at the terrestrial standard of care, were entered into a database. This list included tangible assets (e.g. medications) and intangible assets (e.g. clinical skills to perform a procedure). A team of physicians working within the Exploration Medical Capability Element of NASA's Human Research Program ranked each of the items listed according to its criticality. Data was then obtained from the IMM for the probability of occurrence of the medical conditions, including a breakdown of best case and worst case, during a Mars reference mission. The probability of occurrence information and criticality for each resource were taken into account during analytics performed using Tableau software. RESULTS: A database and weighting system to evaluate all the diagnostic and treatment modalities was created by combining the probability of condition occurrence data with the criticalities assigned by the physician team. DISCUSSION: Exploration Medical Capabilities research at NASA is focused on providing a medical system to support crew medical needs in the context of a Mars mission. MONSTR is a novel approach to performing a quantitative risk analysis that will assess the relative value of individual resources needed for the diagnosis and treatment of various medical conditions. It will provide the operational and research communities at NASA with information to support informed decisions regarding areas of research investment, future crew training, and medical supplies manifested as part of the exploration medical system.

Shah, R. V.↗
DOE OSTI2024

RandONets: Shallow networks with random projections for learning linear and nonlinear operators

Deep neural networks have been extensively used for the solution of both the forward and the inverse problem for dynamical systems. However, their implementation necessitates optimizing a high-dimensional space of parameters and hyperparameters. This fact, along with the requirement of substantial computational resources, pose a barrier to achieving high numerical accuracy, but also interpretability. Here, to address the above challenges, we present Random Projection-based Operator Networks (RandONets): shallow networks with random projections and tailor-made numerical analysis methods that learn accurately and fast linear and nonlinear operators. Building on previous works, we prove that RandOnets are universal approximators of linear and nonlinear operators. Due to their simplicity, RandONets provide a one-step transformation of the input space, facilitating interpretability. For the evaluation of their performance, we focus on operators of PDEs. We show, that RandONets outperform by several orders of magnitude, both in terms of numerical approximation accuracy and computational cost, the “vanilla” DeepONets. Hence, we believe that our method will trigger further developments in the field of scientific machine learning, for the development of new ‘’light”schemes that will provide high accuracy while reducing dramatically the computational cost. A MATLAB toolbox for RandONets, including demos, is available on GitHub at https://github.com/GianlucaFabiani/RandONets.

Interpretable machine learning↗
NASA NTRS1991

Description of real-time Ada software implementation of a power system monitor for the Space Station Freedom PMAD DC testbed

The authors describe the Ada language software developed to perform the electrical power system monitoring functions for the NASA Lewis Research Center's Power Management and Distribution (PMAD) DC testbed. The results of the effort to implement this monitor are presented. The PMAD DC testbed is a reduced-scale prototype of the electric power system to be used in Space Station Freedom. The power is controlled by smart switches known as power control components (or switchgear). The power control components are currently coordinated by five Compaq 386/20e computers connected through an 802.4 local area network. The power system monitor algorithm comprises several functions, including periodic data acquisition, data smoothing, system performance analysis, and status reporting. Data are collected from the switchgear sensors every 100 ms, then passed through a 2-Hz digital filter. System performance analysis includes power interruption and overcurrent detection. The system monitor required a hardware timer interrupt to activate the data acquisition function. The execution time of the code was optimized by using an assembly language routine. The routine allows direct vectoring of the processor to Ada language procedures that perform periodic control activities.

Ludwig, Kimberly↗
DOE OSTI2023

Quantitative Representativeness and Constituency of the Long-Term Agroecosystem Research Network and Analysis of Complementarity with Existing Ecological Networks

Abstract Studies conducted at sites across ecological research networks usually strive to scale their results to larger areas, trying to reach conclusions that are valid throughout larger enclosing regions. Network representativeness and constituency can show how well conditions at sampling locations represent conditions also found elsewhere and can be used to help scale-up results over larger regions. Multivariate statistical methods have been used to design networks and select sites that optimize regional representation, thereby maximizing the value of datasets and research. However, in networks created from already established sites, an immediate challenge is to understand how well existing sites represent the range of environments in the whole area of interest. We performed an analysis to show how well sites in the USDA Long-Term Agroecosystem Research (LTAR) Network represent all agricultural working lands within the conterminous United States (CONUS). Our analysis of 18 LTAR sites, based on 15 climatic and edaphic characteristics, produced maps of representativeness and constituency. Representativeness of the LTAR sites was quantified through an exhaustive pairwise Euclidean distance calculation in multivariate space, between the locations of experiments within each LTAR site and every 1 km cell across the CONUS. Network representativeness is from the perspective of all CONUS locations, but we also considered the perspective from each LTAR site. For every LTAR site, we identified the region that is best represented by that particular site—its constituency—as the set of 1 km grid locations best represented by the environmental drivers at that particular LTAR site. Representativeness shows how well the combination of characteristics at each CONUS location was represented by the LTAR sites’ environments, while constituency shows which LTAR site was the closest match for each location. LTAR representativeness was good across most of the CONUS. Representativeness for croplands was higher than for grazinglands, probably because croplands have more specific environmental criteria. Constituencies resemble ecoregions but have their environmental conditions “centered” on those at particular existing LTAR sites. Constituency of LTAR sites can be used to prioritize the locations of experimental research at or even within particular sites, or to identify the extents that can likely be included when generalizing knowledge across larger regions of the CONUS. Sites with a large constituency have generalist environments, while those with smaller constituency areas have more specialized environmental combinations. These “specialist” sites are the best representatives for smaller, more unusual areas. The potential of sharing complementary sites from the Long-Term Ecological Research (LTER) Network and the National Ecological Observatory Network (NEON) to boost representativeness was also explored. LTAR network representativeness would benefit from borrowing several NEON sites and the Sevilleta LTER site. Later network additions must include such specialist sites that are targeted to represent unique missing environments. While this analysis exhaustively considered principal environmental characteristics related to production on working lands, we did not consider the focal agronomic systems under study, or their socio-economic context.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2024

Semi-supervised Learning of Dynamical Systems with Neural Ordinary Differential Equations: A Teacher-Student Model Approach

Modeling dynamical systems is crucial for a wide range of tasks, but it remains challenging due to complex nonlinear dynamics, limited observations, or lack of prior knowledge. Recently, data-driven approaches such as Neural Ordinary Differential Equations (NODE) have shown promising results by leveraging the expressive power of neural networks to model unknown dynamics. However, these approaches often suffer from limited labeled training data, leading to poor generalization and suboptimal predictions. On the other hand, semi-supervised algorithms can utilize abundant unlabeled data and have demonstrated good performance in classification and regression tasks. We propose TS-NODE, the first semi-supervised approach to modeling dynamical systems with NODE. TS-NODE explores cheaply generated synthetic pseudo rollouts to broaden exploration in the state space and to tackle the challenges brought by lack of ground-truth system data under a teacher-student model. TS-NODE employs an unified optimization framework that corrects the teacher model based on the student's feedback while mitigating the potential false system dynamics present in pseudo rollouts. TS-NODE demonstrates significant performance improvements over a baseline Neural ODE model on multiple dynamical system modeling tasks.

Wang, Yu↗
NASA NTRS1991

Genetic algorithm based input selection for a neural network function approximator with applications to SSME health monitoring

A genetic algorithm is used to select the inputs to a neural network function approximator. In the application considered, modeling critical parameters of the space shuttle main engine (SSME), the functional relationship between measured parameters is unknown and complex. Furthermore, the number of possible input parameters is quite large. Many approaches have been used for input selection, but they are either subjective or do not consider the complex multivariate relationships between parameters. Due to the optimization and space searching capabilities of genetic algorithms they were employed to systematize the input selection process. The results suggest that the genetic algorithm can generate parameter lists of high quality without the explicit use of problem domain knowledge. Suggestions for improving the performance of the input selection process are also provided.

Peck, Charles C.↗
DOE OSTI2024

Toward machine-learning-assisted PW-class high-repetition-rate experiments with solid targets

We present progress in utilizing a machine learning (ML) assisted optimization framework to study the trends in a parameter space defined by spectrally shaped, high-intensity, petawatt-class (8 J, 45 fs) laser pulses interacting with solid targets and give the first simulation-based overview of predicted trends. A neural network (NN) incorporating uncertainty quantification is trained to predict the number of hot electrons generated by the laser–target interaction as a function of pulse shaping parameters. The predictions of this NN serve as the basis function for a Bayesian optimization framework to navigate this space. For post-experimental evaluation, we compare two separate neural network (NN) models. One is based solely on data from experiments, and the other is trained only on ensemble particle-in-cell simulations. Reviewing the predicted and observed trends across the experiment-capable laser parameter search space, we find that both ML models predict a maximal increase in hot electron generation at a level of approximately 12%–18%; however, no statistically significant enhancement was observed in experiments. On direct comparison of the NN models, the average discrepancy is 8.5%, with a maximum of 30%. Since shot-to-shot fluctuations in experiments affect the observations, we evaluate the behavior of our optimization framework by performing virtual experiments that vary the number of repeated observations and the noise levels. Here, we discuss the implications of such a framework for future autonomous exploration platforms in high-repetition-rate experiments.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
NASA NTRS1995

Optimization on a Network-based Parallel Computer System for Supersonic Laminar Wing Design

A set of Computational Fluid Dynamics (CFD) routines and flow transition prediction tools are integrated into a network based parallel numerical optimization routine. Through this optimization routine, the design of a 2-D airfoil and an infinitely swept wing will be studied in order to advance the design cycle capability of supersonic laminar flow wings. The goal of advancing supersonic laminar flow wing design is achieved by wisely choosing the design variables used in the optimization routine. The design variables are represented by the theory of Fourier series and potential theory. These theories, combined with the parallel CFD flow routines and flow transition prediction tools, provide a design space for a global optimal point to be searched. Finally, the parallel optimization routine enables gradient evaluations to be performed in a fast and parallel fashion.

Garcia, Joseph A.↗
DOE OSTI2022

Deep learning to estimate permeability using geophysical data

Time-lapse electrical resistivity tomography (ERT) is a popular geophysical method to estimate three-dimensional (3D) permeability fields from electrical potential difference measurements. Traditional inversion and data assimilation methods are used to ingest this ERT data into hydrogeophysical models to estimate permeability. Due to ill-posedness and the curse of dimensionality, existing inversion strategies provide poor estimates and low resolution of the 3D permeability field. Recent advances in deep learning provide us with powerful algorithms to overcome this challenge. This paper presents a deep learning (DL) framework to estimate the 3D subsurface permeability from time-lapse ERT data. To test the feasibility of the proposed framework, we train DL-enabled inverse models on simulation data. Each measurement in both synthetic and field data is standardized by removing the mean and scaling the time-series to unit variance. This pre-processing step is necessary to bring simulation data closer to field observations. Subsurface process models based on hydrogeophysics are used to generate this synthetic data. Training performed on limited simulation data resulted in the DL model over-fitting. An advanced data augmentation based on mixup is implemented to generate additional training samples to overcome this issue. This mixup technique creates weakly labeled (low-fidelity) samples from strongly labeled (high-fidelity) data. The weakly labeled training data is then used to develop DL-enabled inverse models and reduce over-fitting. As both time-lapse ERT (1133048 features/realization) and 3D permeability (585453 features/realization) data samples are from a high-dimensional space, principal component analysis (PCA) is employed to reduce dimensionality. Encoded ERT and encoded permeability are generated using the trained PCA estimators. A deep neural network is then trained to map the encoded ERT to encoded permeability. This mixup training and unsupervised learning allowed us to build a fast and reasonably accurate DL-based inverse model under limited simulation data. Results show that proposed weak supervised learning can capture salient spatial features in the 3D permeability field. Quantitatively, the average mean squared error (in terms of the natural log) on the strongly labeled training, validation, and test datasets is less than 0.5. The R 2 -score (global metric) is greater than 0.75, and the percent error in each cell (local metric) is less than 10%. Finally, an added benefit in terms of computational cost is that the proposed DL-based inverse model is at least O(10 4 ) times faster than running a forward model once it is trained. Data generation, DL model training, and hyperparameter tuning to identify optimal neural network architectures utilized high-performance computing resources while the DL inference is performed on a standard laptop. Approximately, O(10 5 ) processor hours are used for generating data and DL tuning and training. We acknowledge that the data generation and DL model development are expensive. But once a DL model is trained, it can be re-used for inversion rapidly for the given system, with set physics and domain. Note that traditional inversion may require multiple forward model simulations (e.g., in the order of 10 to 1000), which are very expensive. This computational savings ≈ O(10 5 ) – O(10 7 )) makes the proposed DL-based inverse model attractive for subsurface imaging and real-time ERT monitoring applications due to fast and yet reasonably accurate estimations of permeability field.

58 GEOSCIENCES↗
DOE OSTI2021

Optimizing a magnitude-limited spectroscopic training sample for photometric classification of supernovae

ABSTRACT In preparation for photometric classification of transients from the Legacy Survey of Space and Time (LSST) we run tests with different training data sets. Using estimates of the depth to which the 4-m Multi-Object Spectroscopic Telescope (4MOST) Time Domain Extragalactic Survey (TiDES) can classify transients, we simulate a magnitude-limited sample reaching rAB ≈ 22.5 mag. We run our simulations with the software snmachine, a photometric classification pipeline using machine learning. The machine-learning algorithms struggle to classify supernovae when the training sample is magnitude limited, in contrast to representative training samples. Classification performance noticeably improves when we combine the magnitude-limited training sample with a simulated realistic sample of faint high-redshift supernovae observed from larger spectroscopic facilities; the algorithms’ range of average area under receiver operator characteristic curve (AUC) scores over 10 runs increases from 0.547–0.628 to 0.946–0.969 and purity of the classified sample reaches 95 per cent in all runs for two of the four algorithms. By creating new, artificial light curves using the augmentation software avocado, we achieve a purity in our classified sample of 95 per cent in all 10 runs performed for all machine-learning algorithms considered. We also reach a highest average AUC score of 0.986 with the artificial neural network algorithm. Having ‘true’ faint supernovae to complement our magnitude-limited sample is a crucial requirement in optimization of a 4MOST spectroscopic sample. However, our results are a proof of concept that augmentation is also necessary to achieve the best classification results.

79 ASTRONOMY AND ASTROPHYSICS↗