Search NASA⌕ Search

SEARCH · Search NASA

Results for “exploration versus exploitation”

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.

7 records
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Efficient data acquisition and training of collisional-radiative model artificial neural network surrogates through adaptive parameter space sampling

Abstract Effective plasma transport modeling of magnetically confined fusion devices relies on having an accurate understanding of the ion composition and radiative power losses of the plasma. Generally, these quantities can be obtained from solutions of a collisional-radiative (CR) model at each time step within a plasma transport simulation. However, even compact, approximate CR models can be computationally onerous to evaluate, and in-situ evaluation of these models within a larger plasma transport code can lead to a rigid bottleneck. As a way to bypass this bottleneck, we propose deploying artificial neural network (ANN) surrogates to allow rapid evaluation of the necessary plasma quantities. However, one issue with training an accurate ANN surrogate is the reliance on a sufficiently large and representative training and validation data set, which can be time-consuming to generate. In this work we explore a data-driven active learning and training routine to allow autonomous adaptive sampling of the problem parameter space to ensure a sufficiently large and meaningful set of training data is assembled for the network training. As a result, we can demonstrate approximately order-of-magnitude savings in required training data samples to produce an accurate surrogate.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS2007

Automated Knowledge Discovery From Simulators

A computational method, SimLearn, has been devised to facilitate efficient knowledge discovery from simulators. Simulators are complex computer programs used in science and engineering to model diverse phenomena such as fluid flow, gravitational interactions, coupled mechanical systems, and nuclear, chemical, and biological processes. SimLearn uses active-learning techniques to efficiently address the "landscape characterization problem." In particular, SimLearn tries to determine which regions in "input space" lead to a given output from the simulator, where "input space" refers to an abstraction of all the variables going into the simulator, e.g., initial conditions, parameters, and interaction equations. Landscape characterization can be viewed as an attempt to invert the forward mapping of the simulator and recover the inputs that produce a particular output. Given that a single simulation run can take days or weeks to complete even on a large computing cluster, SimLearn attempts to reduce costs by reducing the number of simulations needed to effect discoveries. Unlike conventional data-mining methods that are applied to static predefined datasets, SimLearn involves an iterative process in which a most informative dataset is constructed dynamically by using the simulator as an oracle. On each iteration, the algorithm models the knowledge it has gained through previous simulation trials and then chooses which simulation trials to run next. Running these trials through the simulator produces new data in the form of input-output pairs. The overall process is embodied in an algorithm that combines support vector machines (SVMs) with active learning. SVMs use learning from examples (the examples are the input-output pairs generated by running the simulator) and a principle called maximum margin to derive predictors that generalize well to new inputs. In SimLearn, the SVM plays the role of modeling the knowledge that has been gained through previous simulation trials. Active learning is used to determine which new input points would be most informative if their output were known. The selected input points are run through the simulator to generate new information that can be used to refine the SVM. The process is then repeated. SimLearn carefully balances exploration (semi-randomly searching around the input space) versus exploitation (using the current state of knowledge to conduct a tightly focused search). During each iteration, SimLearn uses not one, but an ensemble of SVMs. Each SVM in the ensemble is characterized by different hyper-parameters that control various aspects of the learned predictor - for example, whether the predictor is constrained to be very smooth (nearby points in input space lead to similar output predictions) or whether the predictor is allowed to be "bumpy." The various SVMs will have different preferences about which input points they would like to run through the simulator next. SimLearn includes a formal mechanism for balancing the ensemble SVM preferences so that a single choice can be made for the next set of trials.

Burl, Michael↗
DOE OSTI2023

Salt Diapir‐Driven Recycling of Gas Hydrate

Abstract By harnessing both hypothetical, synthetic basin and gas hydrate (GH) system models and real‐world models of well‐studied salt diapir‐associated GH sites at Green Canyon (Gulf of Mexico) and Blake Ridge (U.S. Atlantic coast), we propose and demonstrate salt movement (and in particular, diapirism) to be a new mechanism for the recycling of marine GH. At Green Canyon, for example, we show that by considering this newly proposed diapir‐driven recycling mechanism in conjunction with previously proposed lithological control on sandy‐reservoir‐hosted hydrate at the base of the GH stability zone (BGHSZ; ∼bottom‐simulating reflector, BSR), modeled GH saturations match drilling data. Overall, salt diapir movement‐induced GH recycling provides a temperature‐driven mechanism by which GH saturations at the BGHSZ may reach >90 vol. % and by which GH volumes near and free gas volumes beneath the BGHSZ may be increased significantly through time. Interestingly, comparison of salt diapir‐driven recycling and sediment burial‐driven recycling scenarios suggests notably higher rates of recycling via diapir‐driven versus burial‐driven processes. Our results suggest that GH and associated free gas accumulations above salt diapir crests represent particularly attractive targets for unconventional and conventional hydrocarbon resource exploration and for scientific and academic drilling expeditions aimed at exploiting GH systems. Salt basins containing GH systems—including passive margin basins of the Gulf of Mexico, southeastern Brazil, and southwestern Africa—are therefore compelling localities for studying salt‐driven GH recycling and for salt diapir‐associated natural gas exploration.

58 GEOSCIENCES↗
DOE OSTI2026

Data for Spatial Analysis of Cell Patterning to Aid Genetic and Phenotypic Understanding of Grass Stomatal Density: A Case Study in Maize

Biological processes involve complex hierarchies where composite traits result from multiple component traits. However, holistically understanding of how sets of component traits interact to underpin genotype-to-phenotype relationships is generally lacking. Stomatal density (SD) is a tractable model system for exploring how high-throughput phenotyping (HTP) data could be exploited by a new spatial analysis approach to better understand a developmentally and functionally important trait. SD is a composite trait, resulting from various components related to cell identity and size, which are themselves governed by a series of spatio-developmental processes. Data from 192 recombinant inbred lines of maize [Zea mays (L.)] were analyzed by a new stomatal patterning phenotype (SPP) to (1) describe the average spatial probability distribution of the nearest neighboring stomata; (2) derive a core set of component traits related to cell size, cell packing, and positional probabilities; (3) build a structural equation model of component traits underlying SD; and (4) identify stomatal patterning quantitative trait loci (QTL). The core set of SPP-derived traits explained 74% of the variation in SD. Analyzing SPP component traits allowed some loci previously identified as generic SD QTL to be recognized as specific to lateral versus longitudinal elements of stomatal patterning. Therefore, this study highlights how novel insights can be gained by decomposing a composite trait (e.g., SD) into a set of component traits that were present in HTP data but not previously exploited.

AI/ML↗
NASA NTRS1995

Differenced Range Versus Integrated Doppler (DRVID) ionospheric analysis of metric tracking in the Tracking and Data Relay Satellite System (TDRSS)

The Differenced Range (DR) Versus Integrated Doppler (ID) (DRVID) method exploits the opposition of high-frequency signal versus phase retardation by plasma media to obtain information about the plasma's corruption of simultaneous range and Doppler spacecraft tracking measurements. Thus, DR Plus ID (DRPID) is an observable independent of plasma refraction, while actual DRVID (DR minus ID) measures the time variation of the path electron content independently of spacecraft motion. The DRVID principle has been known since 1961. It has been used to observe interplanetary plasmas, is implemented in Deep Space Network tracking hardware, and has recently been applied to single-frequency Global Positioning System user navigation This paper discusses exploration at the Goddard Space Flight Center (GSFC) Flight Dynamics Division (FDD) of DRVID synthesized from simultaneous two-way range and Doppler tracking for low Earth-orbiting missions supported by the Tracking and Data Relay Satellite System (TDRSS) The paper presents comparisons of actual DR and ID residuals and relates those comparisons to predictions of the Bent model. The complications due to the pilot tone influence on relayed Doppler measurements are considered. Further use of DRVID to evaluate ionospheric models is discussed, as is use of DRPID in reducing dependence on ionospheric modeling in orbit determination.

Radomski, M. S.↗
DOE OSTI2023

Terahertz inverse spin Hall effect in spintronic nanostructures with various ferromagnetic materials

Bilayers of ferromagnetic and heavy metal nanolayers excited with femtosecond laser pulses emit subpicosecond bursts of electromagnetic radiation with spectral frequencies up to several THz. We fabricated such spintronic THz emitters containing various ferromagnetic materials with vastly different magnetic remanence, saturation magnetization, and coercive fields. In all cases, the THz amplitude versus magnetic-field dependence follows the independently measured magnetization hysteresis loops and the THz polarization direction is perpendicular to the sample’s magnetization M, both consistent with the inverse spin Hall effect (ISHE) as the physical origin. The mV-level signals of observed THz transient also favor the ISHE over the anomalous Nernst effect (ANE). The M(T) variation governs the temperature dependence of the THz generation. Emitters with weakly remanent ferromagnets are magnetic-field tunable, while moderately and strongly remanent ferromagnets, once magnetized, allow intense THz generation even without an external field, nevertheless exhibiting low susceptibility to magnetic perturbations if the hysteresis loop is square. Hence, exploiting the magnetic properties of the ferromagnetic layer enables tailoring of weakly temperature-dependent spintronic THz emitters. Finally, we explored the applicability of perovskite oxide materials such as La-Sr-Mn-O (LSMO) for THz transient generation. We detected weak (<40-μV amplitude) THz emissions from both LSMO/Au and pure LSMO nanostructures with no sign flip upon the sample reversal. Furthermore, we excluded the ISHE and believe it was likely governed by the ANE in the thick-film regime, although we cannot exclude the transient demagnetization mechanism.

36 MATERIALS SCIENCE↗
DOE OSTI2026

Spatial analysis of cell patterning to aid genetic and phenotypic understanding of grass stomatal density: A case study in maize

Biological processes involve complex hierarchies where composite traits result from multiple component traits. However, holistically understanding of how sets of component traits interact to underpin genotype-to-phenotype relationships is generally lacking. Stomatal density (SD) is a tractable model system for exploring how high-throughput phenotyping (HTP) data could be exploited by a new spatial analysis approach to better understand a developmentally and functionally important trait. SD is a composite trait, resulting from various components related to cell identity and size, which are themselves governed by a series of spatio-developmental processes. Data from 192 recombinant inbred lines of maize [Zea mays (L.)] were analyzed by a new stomatal patterning phenotype (SPP) to (1) describe the average spatial probability distribution of the nearest neighboring stomata; (2) derive a core set of component traits related to cell size, cell packing, and positional probabilities; (3) build a structural equation model of component traits underlying SD; and (4) identify stomatal patterning quantitative trait loci (QTL). The core set of SPP-derived traits explained 74% of the variation in SD. Analyzing SPP component traits allowed some loci previously identified as generic SD QTL to be recognized as specific to lateral versus longitudinal elements of stomatal patterning. Therefore, this study highlights how novel insights can be gained by decomposing a composite trait (e.g., SD) into a set of component traits that were present in HTP data but not previously exploited.

59 BASIC BIOLOGICAL SCIENCES↗