Search NASA⌕ Search

SEARCH · Search NASA

Results for “feature engineering”

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 127 records · Page 7

A generalized machine learning workflow to visualize mechanical discontinuity

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

42 ENGINEERING↗

Machine Learning for Materials Scientists: An Introductory Guide toward Best Practices

This Methods/Protocols article is intended for materials scientists interested in performing machine learning-centered research. Herein, we cover broad guidelines and best practices regarding the obtaining and treatment of data, feature engineering, model training, validation, evaluation and comparison, popular repositories for materials data and benchmarking data sets, model and architecture sharing, and finally publication. In addition, we include interactive Jupyter notebooks with example Python code to demonstrate some of the concepts, workflows, and best practices discussed. Overall, the data-driven methods and machine learning workflows and considerations are presented in a simple way, allowing interested readers to more intelligently guide their machine learning research using the suggested references, best practices, and their own materials domain expertise.

36 MATERIALS SCIENCE↗

Bayesian Optimization of Catalysis with In-Context Learning

Large language models (LLMs) can perform accurate classification with zero or few examples through in-context learning (ICL), allowing the model to observe query-relevant examples at inference time and eliminating the need for additional weight updates to generalize beyond its original training data. We extend this capability to regression with uncertainty estimation using frozen LLMs (e.g., GPT-4o, Gemini), enabling Bayesian optimization (BO) in natural language without explicit model training or feature engineering. We apply this to materials discovery by representing materials as synthesis and testing procedures for use in natural language prompts. This Bayesian, design-first approach prioritizes optimization toward target material properties before detailed characterization, in contrast to conventional experimental workflows that often emphasize characterization of suboptimal materials. On benchmarks like aqueous solubility and oxidative coupling of methane (OCM), BO-ICL matches or outperforms Gaussian processes. In live experiments on the reverse water–gas shift (RWGS) reaction, BO-ICL identifies multimetallic catalysts that approach equilibrium CO yield within 6 and 10 iterations from a pool of 3,700 and 360,000 candidates, respectively. Our method redefines materials representation and accelerates discovery, with broad applications across catalysis, materials science, and AI.

Calibration↗

Investigating Microtopographic and Soil Controls on a Mountainous Meadow Plant Community Using High-Resolution Remote Sensing and Surface Geophysical Data

Our study aims to explore the microtopographic controls that dictate the heterogeneity of plant communities in a mountainous floodplain-hillslope system, using remote sensing and surface geophysical techniques. Working within a lower montane floodplain-hillslope study site (750 m × 750 m) in the Upper Colorado River Basin, we developed a new data fusion framework, based on machine learning and feature engineering, that exploits remote sensing optical and light detection and ranging (LiDAR) data to estimate the distribution of key plant meadow communities at submeter resolution. We collected surface electrical resistivity tomography data to explore the variability in soil properties along a floodplain-hillslope transect at 0.50-m resolution and extracted LiDAR-derived metrics to model the rapid change in microtopography. Furthermore, we investigated the covariability among the estimated plant community distributions, soil information, and topographic metrics. Results show that our framework estimated the distribution of nine plant communities with higher accuracy (87% versus 80% overall; 85% versus 60% for shrubs) compared to conventional classification approaches. Analysis of the covariabilities reveals a strong correlation between plant community distribution, soil electric conductivity, and slope, indicating that soil moisture is a primary control on heterogeneous spatial distribution. At the same time, microtopography plays an important role in creating particular ecosystem niches for some of the communities. Such relationships could be exploited to provide information about the spatial variability of soil properties. This highly transferable framework can be employed within long-term monitoring to capture community-specific physiological responses to perturbations, offering the possibility of bridging local plot-scale observations with large landscape monitoring.

54 ENVIRONMENTAL SCIENCES↗

AtomSets as a hierarchical transfer learning framework for small and large materials datasets

Abstract Predicting properties from a material’s composition or structure is of great interest for materials design. Deep learning has recently garnered considerable interest in materials predictive tasks with low model errors when dealing with large materials data. However, deep learning models suffer in the small data regime that is common in materials science. Here we develop the AtomSets framework, which utilizes universal compositional and structural descriptors extracted from pre-trained graph network deep learning models with standard multi-layer perceptrons to achieve consistently high model accuracy for both small compositional data (<400) and large structural data (>130,000). The AtomSets models show lower errors than the graph network models at small data limits and other non-deep-learning models at large data limits. They also transfer better in a simulated materials discovery process where the targeted materials have property values out of the training data limits. The models require minimal domain knowledge inputs and are free from feature engineering. The presented AtomSets model framework can potentially accelerate machine learning-assisted materials design and discovery with less data restriction.

Chen, Chi (ORCID:0000000180087043)↗

Enhanced energy coupling for indirectly driven inertial confinement fusion

The indirect drive scheme for inertial confinement fusion (ICF) employs x-rays generated within a high-Z cavity (hohlraum) to implode an embedded low-Z capsule filled with deuterium- tritium (DT) fuel to reach the high density and temperature required for thermonuclear burn. Recently, several cylinder hohlraum experiments on the National Ignition Facility (NIF) have reached the alpha-heating regime [1] where the self-heating by fusion products becomes dominant, producing neutron yields above 1 1016 [2{ 4]. However, there are still challenges on the path towards ignition, such as symmetry control, instability minimization and engineering feature mitigation [5]. Energy coupling from a cylindrical hohlraum to the capsule is typically less than 10%, or 150 kJ for a 1.9 MJ laser drive [1], thereby limiting the energy available in the final hot spot for fusion. Here we report the first experiment on NIF demonstrating 30% energy coupling to an Al capsule in a rugby-shaped Au hohlraum [6]. This high coupling efficiency can substantially enhance the performance margin and improve the prospects for ignition, both in mainline single- shell hot-spot designs and potential double-shell targets.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

A generalized deep learning approach for local structure identification in molecular simulations

Identifying local structure in molecular simulations is of utmost importance. The most common existing approach to identify local structure is to calculate some geometrical quantity referred to as an order parameter. In simple cases order parameters are physically intuitive and trivial to develop (e.g., ion-pair distance), however in most cases, order parameter development becomes a much more difficult endeavor (e.g., crystal structure identification). Using ideas from computer vision, we adapt a specific type of neural network called a PointNet to identify local structural environments in molecular simulations. A primary challenge in applying machine learning techniques to simulation is selecting the appropriate input features. This challenge is system-specific and requires significant human input and intuition. In contrast, our approach is a generic framework that requires no system-specific feature engineering and operates on the raw output of the simulations, i.e., atomic positions. We demonstrate the method on crystal structure identification in Lennard-Jones (four different phases), water (eight different phases), and mesophase (six different phases) systems. The method achieves as high as 99.5% accuracy in crystal structure identification. The method is applicable to heterogeneous nucleation and it can even predict the crystal phases of atoms near external interfaces. We demonstrate the versatility of our approach by using our method to identify surface hydrophobicity based solely upon positions and orientations of surrounding water molecules. Our results suggest the approach will be broadly applicable to many types of local structure in simulations.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Exploring End-to-end Deep Learning Applications for Event Classification at CMS

An essential part of new physics searches at the Large Hadron Collider (LHC) at CERN involves event classification, or distinguishing potential signal events from those coming from background processes. Current machine learning techniques accomplish this using traditional hand-engineered features like particle 4-momenta, motivated by our understanding of particle decay phenomenology. While such techniques have proven useful for simple decays, they are highly dependent on our ability to model all aspects of the phenomenology and detector response. Meanwhile, powerful deep learning algorithms are capable of not only training on high-level features, but of performing feature extraction. In computer vision, convolutional neural networks have become the state-of-the-art for many applications. Motivated by their success, we apply deep learning algorithms to low-level detector data from the 2012 CMS Simulated Open Data to directly learn useful features, in what we call, end-to-end event classification. We demonstrate the power of this approach in the context of a physics search and offer solutions to some of the inherent challenges, such as image construction, image sparsity, combining multiple sub-detectors, and de-correlating the classifier from the search observable, among others.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Observation of Wakefield Generation in Left-Handed Band of Metamaterial-Loaded Waveguide

We report on a design of a TM-mode based metamaterial-loaded waveguide. Network analyzer measurements demonstrated a left-handed propagation region for the TM11 mode at around 10 GHz. A beamline experiment was performed with the metamaterial-loaded waveguide. In this experiment, a 6 MeV electron beam passes through the waveguide and generates a wakefield, via the Cherenkov radiation mechanism. We detected a signal in the left-handed frequency band at 10 GHz. This is an indirect demonstration of reverse Cherenkov radiation as discussed in several papers. Cherenkov radiation in artificially constructed materials (metamaterials, MTM) can provide unusual, engineered features that can be advantageous for particle detector design.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

High-resolution modeling of indirectly driven high-convergence layered inertial confinement fusion capsule implosions

In this study, we present the results of high-resolution simulations of the implosion of high-convergence layered indirect-drive inertial confinement fusion capsules of the type fielded on the National Ignition Facility using the xRAGE radiation-hydrodynamics code. In order to evaluate the suitability of xRAGE to model such experiments, we benchmark simulation results against available experimental data, including shock-timing, shock-velocity, and shell trajectory data, as well as hydrodynamic instability growth rates. We discuss the code improvements that were necessary in order to achieve favorable comparisons with these data. Due to its use of adaptive mesh refinement and Eulerian hydrodynamics, xRAGE is particularly well suited for high-resolution study of multi-scale engineering features such as the capsule support tent and fill tube, which are known to impact the performance of high-convergence capsule implosions. High-resolution two-dimensional (2D) simulations including accurate and well-resolved models for the capsule fill tube, support tent, drive asymmetry, and capsule surface roughness are presented. These asymmetry seeds are isolated in order to study their relative importance and the resolution of the simulations enables the observation of details that have not been previously reported. We analyze simulation results to determine how the different asymmetries affect hotspot reactivity, confinement, and confinement time and how these combine to degrade yield. Yield degradation associated with the tent occurs largely through decreased reactivity due to the escape of hot fuel mass from the hotspot. Drive asymmetries and the fill tube, however, degrade yield primarily via burn truncation, as associated instability growth accelerates the disassembly of the hotspot. Finally, modeling all of these asymmetries together in 2D leads to improved agreement with experiment but falls short of explaining the experimentally observed yield degradation, consistent with previous 2D simulations of such capsules.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

A near one-dimensional indirectly driven implosion at convergence ratio 30

Inertial confinement fusion cryogenic-layered implosions at the National Ignition Facility, while successfully demonstrating self-heating due to alpha-particle deposition, have fallen short of the performance predicted by one-dimensional (1D) multi-physics implosion simulations. The current understanding, from experimental evidence as well as simulations, suggests that engineering features such as the capsule tent and fill tube, as well as time-dependent low-mode asymmetry, are to blame for the lack of agreement. A short series of experiments designed specifically to avoid these degradations to the implosion are described here in order to understand if, once they are removed, a high-convergence cryogenic-layered deuterium-tritium implosion can achieve the 1D simulated performance. The result is a cryogenic layered implosion, round at stagnation, that matches closely the performance predicted by 1D simulations. This agreement can then be exploited to examine the sensitivity of approximations in the model to the constraints imposed by the data.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Mitigation of X-ray shadow seeding of hydrodynamic instabilities on inertial confinement fusion capsules using a reduced diameter fuel fill-tube

We report a reduced X-ray shadow imprint of hydrodynamic instabilities on the high-density carbon ablator surface of inertial confinement fusion (ICF) capsules using a reduced diameter fuel fill tube on the National Ignition Facility (NIF). The perturbation seed mass from hydrodynamic instabilities was reduced by approximately an order of magnitude by reducing both the diameter and wall thickness of the fill tube by ~2×, consistent with analytical estimates. Overall, this work demonstrates a successful mitigation strategy for engineered features for ICF implosions on the NIF.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Machine learning prediction of self-diffusion in Lennard-Jones fluids

In this work, different machine learning (ML) methods were explored for the prediction of self-diffusion in Lennard-Jones (LJ) fluids. Using a database of diffusion constants obtained from the molecular dynamics simulation literature, multiple Random Forest (RF) and Artificial Neural Net (ANN) regression models were developed and characterized. The role and improved performance of feature engineering coupled to the RF model development was also addressed. The performance of these different ML models was evaluated by comparing the prediction error to an existing empirical relationship used to describe LJ fluid diffusion. It was found that the ANN regression models provided superior prediction of diffusion in comparison to the existing empirical relationships.

74 ATOMIC AND MOLECULAR PHYSICS↗

Magnetization around mix jets entering inertial confinement fusion fuel

Engineering features are known to cause jets of ablator materials to enter the fuel hot-spot in inertial confinement fusion implosions. The Biermann battery mechanism wraps them in a self-generated magnetic field. Here we show that higher-Z jets have an additional thermoelectric magnetic source term that is not present for hydrogen jets, verified here through a kinetic simulation. It has similar magnitude to the Biermann term. We then include this in an extended magneto-hydrodynamic approach to post process an xRAGE radiation-hydrodynamic implosion simulation. The simulation includes an accurate model for the capsule fill tube, producing a dense carbon jet that becomes wrapped in a 4000 T magnetic field. A simple spherical carbon mix model shows that this insulates the electron heat conduction enough to cause contraction of the jet to an optically thick equilibrium. The denser magnetized jet hydrodynamics could change its core penetration and therefore the final mix mass, which is known to be well correlated with fusion yield degradation. Fully exploring this will require self-consistent magneto-hydrodynamic simulations. Experimental signatures of this self-magnetization may emerge in the high energy neutron spectrum.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

A mechanism for reduced compression in indirectly driven layered capsule implosions

High-yield implosions on the National Ignition Facility rely on maintaining low entropy in the deuterium–tritium fuel, quantified by its adiabat, in order to efficiently couple energy to the hot spot through high compression of the fuel layer. We present very-high-resolution xRAGE simulation results that study the impacts of interfacial mixing and the jetting of materials due to surface defects, defects on internal interfaces, voids, and engineering features on fuel layer compression. Defects and voids are typically neglected in implosion simulations due to their small size and three-dimensional geometry. Our results showed that supersonic jets of material arise through weak spots in the shell at peak implosion velocity that prevent uniform compression of the fuel layer even when they do not introduce contaminant into the hot spot. This occurs despite maintaining low fuel entropy, since the formation of the weak spots involves nonradial displacement of fuel mass. In contrast, simulations show that fuel–ablator mixing due to interfacial instabilities has a much smaller impact on compression. We show that defects on interior interfaces of plastic capsules decrease compression by 15% to 25% and interfacial mixing between the ablator and fuel decreases compression by less than 1% for implosions with plastic or high-density carbon (HDC) ablators. For low adiabat implosions, the impact of jetting seeded by the support tent can also decrease the compression by 25%. We demonstrate that the inclusion of interior defects in simulations can explain the inferred compression in two fielded plastic capsule implosions and that the inclusion of voids, for which available characterization has large uncertainties, in simulations of HDC capsule implosions has a qualitatively consistent impact. This mechanism offers a potential explanation for persistently overestimated fuel compression in design simulations of layered implosions on the National Ignition Facility.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Detailed simulations of the first deuterium-tritium-filled double shell implosions on the National Ignition Facility

The first indirectly driven, liquid DT-filled double shell inertial confinement fusion (ICF) implosions have recently been successfully performed on the National Ignition Facility (NIF). Double shells are a class of alternative designs that use a low-Z outer shell to compress a foam cushion that accelerates a high-Z inner shell to efficiently compress a liquid DT core. Double shells are challenging to fabricate, field, and model. Important engineering features enabling double shell fabrication include a fill-tube penetrating all shells and a carefully designed and very narrow (few μm) step-joint in the ablator. Due to the higher density materials involved, high Atwood number instabilities are also important at many material interfaces. In this paper, numerical simulations of double shell implosions using the Los Alamos National Laboratory multi-physics radiation-hydrodynamics code xRAGE will be discussed. An extensive effort has been under way for several years to develop the code capabilities for ICF simulations in a common modeling framework to allow ease of simulation setup and standardization of the computational methodology. This paper will present a wide range of simulation results capturing, quantifying, and comparing the impact of all these degradation mechanisms on implosion performance. Brief comparisons with recent experimental results and suggestions for future improvements will also be discussed. Our results suggest that capsule surface roughness and the step-joint gap have the largest impact on implosion performance. Initial experimental data may suggest that the sensitivity to the step-joint gap could provide the dominant explanation for DT-filled double shell experiments that have been fielded on NIF thus far.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

First indirectly driven liquid-DT filled double shell implosions at the National Ignition Facility

Double shell implosions aim to explore material mixing under fusion conditions in a volume burn geometry using high-Z metal pushers. High-Z pushers are more compressible than low-Z pushers enabling high stagnation pressure which reduces the required implosion speed while maintaining a low pusher adiabat despite strong shock heating. Additionally, the use of a small fuel mass reduces the fuel internal energy required for ignition, thus achieving stable platforms with reasonable fusion output to conduct controlled experiments. These factors make volume burn in a double shell implosion highly promising. Recently, a series of liquid-DT filled, indirectly driven, double shell implosions were conducted at the National Ignition Facility with laser drives reaching up to 1.5 MJ. These experiments achieved a maximum DT neutron yield of 1.67 × 10 14 (yield—479 J), DT ion temperature of 2.6 keV, fuel areal density (ρR) of 0.15 g/cm 2 , and stagnation pressure of 79 Gbar. Over the course of these shots, the DT neutron yield has increased by an order of magnitude largely from improved mitigation of outer shell assembly joint driven instability growth using thicker gold plating at the assembly joint. Further performance improvements are expected by enhancing outer-to-inner shell kinetic energy transfer and refined mitigation of degradations from engineering features such as the outer shell joint, fill tube, and surface roughness.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Unsupervised Learning for Improved Gamma-Ray Spectrometry in Pixelated Cadmium Zinc Telluride (CZT) Detectors

Machine learning has been found to be ubiquitously useful across many industries, presenting an opportunity to improve radiation detection performance using data-driven algorithms. Improved detector resolution can aid in the detection, identification, and quantification of radionuclides. Here, in this work, a novel, data-driven, unsupervised learning approach is developed to improve detector spectral characteristics by learning, and subsequently rejecting, poorly performing regions of the pixelated detector. Feature engineering is used to fit individual characteristic photo peaks to a Doniach lineshape with a linear background model. Then, principal component analysis is used to learn a lower-dimension latent space representation of each photo peak where the pixels are clustered, and subsequently ranked, based on the cluster mean distance to an optimal point. Pixels within the worst cluster(s) are rejected to improve the full-width at half-maximum (FWHM) by 10% to 15% (relative to the bulk detector) at 50% net efficiency when applied to training data obtained from measurements of a 100 μCi 154 Eu source using a H3D M400i pixelated cadmium zinc telluride detector. These results compare well with, but do not outperform, a greedy algorithm that accumulates pixels in order of FWHM from lowest to highest used as a benchmark. In the future, this approach can be extended to include the detector energy and angular response. Finally, the model is applied to newly seen natural and enriched uranium spectra relevant for nuclear safeguards applications.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗