Search NASA⌕ Search

SEARCH · Search NASA

Results for “Convolution”

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 1,621 records · Page 90

Noise signal identification in time projection chamber data using deep learning model

Deep learning has been employed in various scientific fields and has provided promising results. Here, in this study, a deep learning classifier was implemented to improve the quality of data obtained from a time projection chamber. Digital waveforms of the detected signals were classified into the following three categories: particles, noises, and particles piled up with noises. A simple 1-dimensional convolutional neural network was developed for the classification. The model demonstrated an excellent performance on the test dataset. Its practical performance was also examined using track images and particle identification plots by comparing the original and clean data without the noise signals. The comparison clearly showed that the deep learning model improved the quality of data. The current study presents an effective application of the deep learning model for the time projection chamber data.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Time-interval distributions in a digital gamma multi-channel analyzer at extreme input rates

Here, this study explores the relationship between experimentally observed inter-event time-interval distributions (TIDs) and measured dead times for pulses generated by a HPGe gamma detector and processed by a digital spectrometer. The system utilizes a fast and slow channel, pile-up rejector, trapezoid filtering, and flash analog-to-digital converter. The experimentally derived TIDs were compared with theory for validation. The results demonstrate that the theoretical model reliably describes measured TIDs up to 40% dead time. However, significant distortion effects become increasingly pronounced at higher input rates. It appears that the deviation between the measured and calculated time-interval distributions take the shape of higher-order convolutions of the TID, which represent the time difference between counts for more than two successive events. In this work, theoretical functions for the TID are expanded to reproduce measurements up to 90%. This refinement in the interpretation and treatment of the measured TIDs provides improved accuracy and precision in the prediction of the true event rate and measured dead time in the counter. Even though it reflects occasional count loss that is specific to the set-up, it is expected that a similar adjustment may be applicable to other detection systems as well.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Unsupervised anomaly detection in MeV ultrafast electron diffraction

MeV ultrafast electron diffraction (MUED) is a pump-probe technique used to study the dynamic structural evolution of materials. An ultrashort laser pulse triggers structural changes, which are then probed by an ultrashort relativistic electron beam. To overcome low signal-to-noise ratios, diffraction patterns are averaged over thousands of shots. However, shot-to-shot instabilities in the electron beam can distort individual patterns, introducing uncertainty. Improving MUED accuracy requires detecting and removing these anomalous patterns from large datasets. In this work, we developed a fully unsupervised methodology for the detection of anomalous diffraction patterns. Using a convolutional autoencoder, we calculate the reconstruction mean squared error of the diffraction patterns. Based on the statistical analysis of this error, we provide the user an estimation of the probability that the pattern is normal, which also allows a posterior visual inspection of the images that are difficult to classify. This method has been trained with only 100 diffraction patterns and tested on 1521 patterns, resulting in a false positive rate between 0.2% and 0.4%, with a training time of 10 s per image and a test time of about 1 s per image. Here, the proposed methodology can also be applied to other diffraction techniques in which large datasets are collected that include faulty images due to instrumental instabilities.

43 PARTICLE ACCELERATORS↗

Experimental observations of bifurcated power decay lengths in the near Scrape-Off Layer of ST40 High Field Spherical Tokamak

The scrape-off layer parallel heat flux decay lengths measured at ST40, a high field, low aspect ratio spherical tokamak, have been observed to bifurcate into two groups. The wide group follows established H-mode scalings (ranging between 2 to 8 mm) while the narrow group falls up to 10 times below these scalings (between 0.2 and 0.8 mm), being comparable to the ion total Larmor radius rather than the ion poloidal Larmor radius. The heat flux profiles of the latter group can only be described by a multi-exponential function, rather than the single exponential function convoluted with a Gaussian. The onset of the narrow scrape-off layer width is observed to be associated with suppressed magnetic fluctuations, suggesting reduced electromagnetic turbulence levels in the SOL.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Beyond interpolation: Physics-inspired gating transformers for extrapolating irradiation conditions to novel nuclear fuels

The qualification of advanced nuclear fuels relies on irradiation experiments in test reactors that emulate commercial conditions. Designing these tests requires accurate prediction of key irradiation quantities, particularly heat generation rate and burnup, yet obtaining them typically involves computationally expensive multi-step simulation workflows. We propose a physics-inspired gating transformer (PIGT) that integrates an inverse-square, distance-based attenuation into the encoder representation to bias attention toward physically relevant spatial relationships while retaining data-driven flexibility. Using MiniFuel irradiation data from the High Flux Isotope Reactor at Oak Ridge National Laboratory, we benchmark against ensemble methods, feedforward and recurrent networks, convolutional models, and standard transformers. While baseline models perform well under interpolation, they exhibit a pronounced generalization gap when evaluated on fuels not included in the training set. The proposed model consistently improves extrapolative accuracy and stability, yielding the strongest performance on unseen fuel configurations. These results indicate that a lightweight physics structure embedded within attention mechanisms can substantially improve robustness, enabling more reliable surrogate predictions to accelerate the design of nuclear fuel irradiation experiments.

Fuel qualification↗

Two-gluon correlations in heavy–light ion collisions: Energy and geometry dependence, IR divergences, and k T -factorization

Here, we study the properties of the cross section for two-gluon production in heavy–light ion collisions derived in our previous paper in the saturation/Color Glass Condensate framework. Concentrating on the energy and geometry dependence of the corresponding correlation functions we find that the two-gluon correlator is a much slower function of the center-of-mass energy than the one- and two-gluon production cross sections. The geometry dependence of the correlation function leads to stronger azimuthal near- and away-side correlations in the tip-on-tip U + U collisions than in the side-on-side U + U collisions, an exactly opposite behavior from the correlations generated by the elliptic flow of the quark–gluon plasma: a study of azimuthal correlations in the U + U collisions may thus help to disentangle the two sources of correlations. We demonstrate that the cross section for two-gluon production in heavy–light ion collisions contains a power-law infrared (IR) divergence even for fixed produced gluon momenta: while saturation effects in the target regulate some of the power-law IR-divergent terms in the lowest-order expression for the two-gluon correlator, other power-law IR-divergent terms remain, possibly due to absence of saturation effects in the dilute projectile. Finally we rewrite our result for the two-gluon production cross-section in a k T -factorized form, obtaining a new factorized expression involving a convolution of one- and two-gluon Wigner distributions over both the transverse momenta and impact parameters. Lastly, we show that the two-gluon production cross-section depends on two different types of unintegrated two-gluon Wigner distribution functions.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Classify QCD phase transition with deep learning

The state-of-the-art pattern recognition method in machine learning (deep convolution neural network) is used to identify the equation of state (EoS) employed in the relativistic hydrodynamic simulations of heavy ion collisions. High-level correlations of particle spectra in transverse momentum and azimuthal angle learned by the network act as an effective EoS-meter in deciphering the nature of the phase transition in QCD. The EoS-meter is model independent and insensitive to other simulation inputs including the initial conditions and shear viscosity for hydrodynamic simulations. Through this study we demonstrate that there is a traceable encoder of the dynamical information from the phase structure that survives the evolution and exists in the final snapshot of heavy ion collisions and one can exclusively and effectively decode these information from the highly complex final output with machine learning when traditional methods fail. Besides the deep neural network, the performance of traditional machine learning classifiers are also provided.

74 ATOMIC AND MOLECULAR PHYSICS↗

Identifying the nature of the QCD transition in heavy-ion collisions with deep learning

In this proceeding, we review our recent work using deep convolutional neural network (CNN) to identify the nature of the QCD transition in a hybrid modeling of heavy-ion collisions. Within this hybrid model, a viscous hydrodynamic model is coupled with a hadronic cascade “after-burner”. As a binary classification setup, we employ two different types of equations of state (EoS) of the hot medium in the hydrodynamic evolution. The resulting final-state pion spectra in the transverse momentum and azimuthal angle plane are fed to the neural network as the input data in order to distinguish different EoS. To probe the effects of the fluctuations in the event-by-event spectra, we explore different scenarios for the input data and make a comparison in a systematic way. We observe a clear hierarchy in the predictive power when the network is fed with the event-by-event, cascade-coarse-grained and event-fine-averaged spectra. The carefully-trained neural network can extract high-level features from pion spectra to identify the nature of the QCD transition in a realistic simulation scenario.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Parton Energy Loss in the Generalized High-Twist Approach

In this work, we calculate the radiative parton energy loss in the deeply inelastic scattering (DIS) off a large nucleus within a generalized high-twist approach. The final gluon radiation spectrum is a convolution of the hard partonic part and the transverse momentum dependent (TMD) quark-gluon correlation function, without the twist expansion in the transverse momentum of the initial gluons used in original high twist approach. The TMD quark-gluon correlation function can be factorized approximately as the product of initial quark distribution and TMD gluon distribution which can be used to define the generalized or TMD jet transport coefficient. The radiation spectrum will recover the Gylassy-Levai-Vitev (GLV) result in the first order of the opacity expansion, under the static scattering center and soft gluon radiation approximation. We also investigate numerically the difference as a result of the soft gluon radiation approximation, under the static scattering center approximation.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Toward prediction of turbulent atmospheric flows over propagating oceanic waves via machine-learning augmented large-eddy simulation

Wind-wave interactions have important effects on the energy harvesting of offshore wind farms. High-fidelity large-eddy simulation (LES) is a powerful approach for investigating wind-wave interactions in turbulent oceanic environments. Due to the large scale of the flow domain and the high grid resolution required to resolve multi-scale flow motions, however, brute-force LES of wind-wave interactions is computationally very expensive. We propose augmenting brute-force LES via machine-learning data-driven modeling (ML-LES) to dramatically reduce the computational time required to obtain converged turbulence statistics when the brute-force approach is employed. Namely, we employ a convolutional neural network (CNN) autoencoder trained and validated with LES data sets to develop a highly efficient ML-LES approach for computing turbulence statistics from just a few snapshots of instantaneous LES flow fields. Further, our results demonstrate the accuracy and efficiency of ML-LES in predicting the mean velocity, velocity fluctuations, and turbulence kinetic energy in highly stretched computational grid systems required to carry out simulations in real-life oceanic environments.

42 ENGINEERING↗

Frame invariant neural network closures for Kraichnan turbulence

Numerical simulations of geophysical and atmospheric flows have to rely on parameterizations of subgrid scale processes due to their limited spatial resolution. Despite substantial progress in developing parameterization (or closure) models for subgrid scale (SGS) processes using physical insights and mathematical approximations, they remain imperfect and can lead to inaccurate predictions. In recent years, machine learning has been successful in extracting complex patterns from high-resolution spatio-temporal data, leading to improved parameterization models, and ultimately better coarse grid prediction. However, the inability to satisfy known physics and poor generalization hinders the application of these models for real-world problems. In this work, we put forth a frame invariant closure approach to improve the accuracy and generalizability of deep learning-based subgrid scale closure models by embedding physical symmetries directly into the structure of the neural network. Specifically, we utilized specialized layers within the convolutional neural network in such a way that desired constraints are theoretically guaranteed without the need for any regularization terms. We demonstrate our framework for a two-dimensional decaying turbulence test case mostly characterized by the forward enstrophy cascade. We show that our frame invariant SGS model (i) accurately predicts the subgrid scale source term, (ii) respects the physical symmetries such as translation, Galilean, and rotation invariance, and (iii) is numerically stable when implemented in coarse-grid simulation with generalization to different initial conditions and Reynolds number. This work opens up a possibility of connecting physics-based theories and data-driven modeling paradigms, and thus represents a promising step towards the development of physically consistent data-driven turbulence closure models.

42 ENGINEERING↗

Assessments of epistemic uncertainty using Gaussian stochastic weight averaging for fluid-flow regression

Here, we use Gaussian stochastic weight averaging (SWAG) to assess the epistemic uncertainty associated with neural-network-based function approximation relevant to fluid flows. SWAG approximates a posterior Gaussian distribution of each weight, given training data, and a constant learning rate. Having access to this distribution, it is able to create multiple models with various combinations of sampled weights, which can be used to obtain ensemble predictions. The average of such an ensemble can be regarded as the 'mean estimation', whereas its standard deviation can be used to construct 'confidence intervals', which enable us to perform uncertainty quantification (UQ) with regard to the training process of neural networks. We utilize representative neural-network-based function approximation tasks for the following cases: (i) a two-dimensional circular-cylinder wake; (ii) the DayMET dataset (maximum daily temperature in North America); (iii) a three-dimensional square-cylinder wake; and (iv) urban flow, to assess the generalizability of the present idea for a wide range of complex datasets. SWAG-based UQ can be applied regardless of the network architecture, and therefore, we demonstrate the applicability of the method for two types of neural networks: (i) global field reconstruction from sparse sensors by combining convolutional neural network (CNN) and multi-layer perceptron (MLP); and (ii) far-field state estimation from sectional data with two-dimensional CNN. We find that SWAG can obtain physically-interpretable confidence-interval estimates from the perspective of epistemic uncertainty. This capability supports its use for a wide range of problems in science and engineering.

97 MATHEMATICS AND COMPUTING↗

Particle multiplicities in the central region of high-energy collisions from $k_T$-factorization with running coupling corrections

Horowitz and Kovchegov have derived a $k_T$-factorization formula for particle production at small x which includes running coupling corrections. We perform a first numerical analysis to confront the theory with data on the energy and centrality dependence of particle multiplicities at midrapidity in high-energy p+A (and A+A) collisions. Moreover, we point out a strikingly different dependence of the multiplicity per participant on $N$ part in p+Pb vs. Pb+Pb collisions at LHC energies, and argue that the observed behavior follows rather naturally from the convolution of the gluon distributions of an asymmetric vs. symmetric projectile and target.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Probing criticality with deep learning in relativistic heavy-ion collisions

Systems with different interactions could develop the same critical behavior due to the underlying symmetry and universality. Using this principle of universality, we can embed critical correlations modeled on the 3D Ising model into the simulated data of heavy-ion collisions, hiding weak signals of a few inter-particle correlations within a large particle cloud. Employing a point cloud network with dynamical edge convolution, we are able to identify events with critical fluctuations through supervised learning, and pick out a large fraction of signal particles used for decision-making in each single event.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Search for a boosted Higgs boson decaying to bottom quark pairs in association with a W or Z boson in proton-proton collisions at $\sqrt{s}$ = 13TeV

A search is conducted for standard model Higgs bosons with large transverse momentum ($p_\mathrm{T}$) decaying to bottom quark pairs and produced in association with a hadronically decaying W or Z boson at the LHC. The result is based on a dataset of proton-proton collisions at a center-of-mass energy of 13 TeV collected with the CMS detector in 2016$-$2018, corresponding to an integrated luminosity of 138 fb$^{-1}$. Boosted Higgs, W, and Z boson decays are reconstructed using large-radius jets with $p_\mathrm{T}$$\gt$ 450 GeV and identified with heavy-flavor classifiers based on a graph convolutional neural network. The observed signal strength relative to the standard model expectation is $μ$ = 0.72 $^{+0.75}_{-0.71}$ including statistical and systematic uncertainties.

CMS↗

Strength statistics of single crystals and metallic glasses under small stressed volumes

It has been well documented that plastic deformation of crystalline and amorphous metals/alloys shows a general trend of “smaller is stronger”. The majority of the experimental and modeling studies along this line have been focused on finding and reasoning the scaling slope or exponent in the logarithmic plot of strength versus size. In contrast to this view, here we show that the universal picture should be the thermally activated nucleation mechanisms in small stressed volume, the stochastic behavior as to find the weakest links in intermediate sizes of the stressed volume, and the convolution of these two mechanisms with respect to variables such as indenter radius in nanoindentation pop-in, crystallographic orientation, pre-strain level, sample length as in uniaxial tests, and others. Furthermore, experiments that cover the entire spectrum of length scales and a unified model that treats both thermal activation and spatial stochasticity have discovered new perspectives in understanding and correlating the strength statistics in a vast of observations in nanoindentation, micro-pillar compression, and fiber/whisker tension tests of single crystals and metallic glasses.

36 MATERIALS SCIENCE↗

Development of a plug-and-play anti-noise module for fault diagnosis of rotating machines in nuclear power plants

The health of rotating machines is crucial to the stable operation and safety of nuclear power plants. However, research on machine learning-based fault diagnosis of rotating machines in the nuclear industry is still in its infancy. The signal noise generated in the plant may negatively affect the effectiveness of the analysis. In this paper, a plug-and-play anti-noise machine learning module is proposed to fill the knowledge and capability gap. The modules are loaded into a convolutional neural network called deep residual network (ResNet) to obtain a new model with noise reduction capability. The basic idea is that the module is able to identify noise features and include them in the subsequent analysis, effectively filtering noise at the feature level of the network. Nine variants of the new model are compared with the original ResNet as well as four classical machine learning models to test the effectiveness of the module and to examine the impact of the module's loading modes on the performance of the new model. Finally, this research helps facilitate the application of machine learning in the plant noise environment.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Carbonyl Identification and Quantification Uncertainties for Oxidative Polymer Degradation

The most revealing indicator for oxidative processes or state of degraded plastics is usually carbonyl formation, a key step in materials degradation as part of the carbon cycle for man-made materials. Hence, the identification and quantification of carbonyl species with infrared spectroscopy have been the method of choice for generations, thanks to their strong absorbance and being an essential intermediate in carbon oxidation pathways. Despite their importance, precise identification and quantification can be challenging and rigorous fully traceable data are surprisingly rare in the existing literature. An overview of the complexity of carbonyl quantification is presented by the screening of reference compounds in solution with transmission and polymer films with ATR IR spectroscopy, and systematic data analyses. Significant variances in existing data and their past use have been recognized. Guidance is offered how better measurements and data reporting could be accomplished. Experimental variances depend on the combination of uncertainty in exact carbonyl species, extinction coefficient, contributions from neighboring convoluting peaks, matrix interaction phenomena and instrumental variations in primary IR spectral acquisition (refractive index and penetration depth for ATR measurements). In addition, diverging sources for relevant extinction coefficients may exist, based on original spectral acquisition. For common polymer degradation challenges, a relative comparison of carbonyl yields for a material is easily accessible, but quantification for other purposes, such as degradation rates and spatially dependent interpretation, requires thorough experimental validation. All variables featured in this overview demonstrate the significant error margins in carbonyl quantification, with exact carbonyl species and extinction coefficients already being major contributors on their own.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗