Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “test vector”

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 1,729 records · Page 96
All sourcesNASA NTRSDOE OSTI
DOE OSTI2019

Hyperspectral vegetation identification at a legacy underground nuclear explosion test site

The detection, location, and identification of suspected underground nuclear explosions (UNEs) are global security priorities that rely on integrated analysis of multiple data modalities for uncertainty reduction in event analysis. Vegetation disturbances may provide complementary signatures that can confirm or build on the observables produced by prompt sensing techniques such as seismic or radionuclide monitoring networks. For instance, the emergence of non-native species in an area may be indicative of anthropogenic activity or changes in vegetation health may reflect changes in the site conditions resulting from an underground explosion. Previously, we collected high spatial resolution (10 cm) hyperspectral data from an unmanned aerial system at a legacy underground nuclear explosion test site and its surrounds. These data consist of visible and near-infrared wavebands over 4.3 km 2 of high desert terrain along with high spatial resolution (2.5 cm) RGB context imagery. In this work, we employ various spectral detection and classification algorithms to identify and map vegetation species in an area of interest containing the legacy test site. Here, we employed a frequentist framework for fusing multiple spectral detections across various reference spectra captured at different times and sampled from multiple locations. The spatial distribution of vegetation species is compared to the location of the underground nuclear explosion. We discover a difference in species abundance within a 130 m radius of the center of the test site.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2022

High-Gradient Wien Spin Rotators at Jefferson Lab

Nuclear physics experiments performed in the Continuous Electron Beam Accelerator Facility (CEBAF) at Jefferson Laboratory (JLab) require spin manipulation of electron beams. Two Wien spin rotators in the injector keV region are essential at CEBAF to establish longitudinal polarization at the end station target, and to flip the polarization direction by ? rad to rule out false asymmetries. In a Wien filter, the homogeneous and independent electric and magnetic fields, along with the velocity vectors of the electrons that traverse it, form a mutually orthogonal system. The magnitude of the electrostatic field, established by biasing two highly-polished electrodes, defines the desired spin angle at the target yet deviates the beam trajectory due to the Lorentz force. The beam trajectory in the Wien is then re-established by adjusting the magnetic field, induced by an electromagnet encasing the device vacuum chamber. This contribution describes the evolution design and high voltage testing of Wien filters for spin manipulation at increased beam energies in the keV injector region, required by high precision parity violation experiments like MOLLER.

Palacios Serrano, G. G.↗
DOE OSTI2020

Search for New Bosons in Gluon-Gluon and Vector Boson Fusion Processes at the LHC and Development of Silicon Module Assembly Techniques for the CMS High Granularity Calorimeter

A search for heavy resonances decaying to a pair of vectors bosons is presented which utilizes events in which a hadronically decaying Z or W boson is identified using jet substructure techniques and large missing transverse momentum is found. Data analyzed were recorded by the Compact Muon Solenoid experiment at the CERN Large Hadron Collider in 2016, 2017, and 2018, and correspond to 137 fb$^{-1}$ . The events are categorized as having arisen from Vector Boson Fusion process or not, and are characterized by their transverse mass distribution. The standard model backgrounds are estimated based on observed yields in control regions. No excess of events above the expected SM background are observed and limits are placed on the production cross section of Radion (spin-0), Wprime (spin-1) and Bulk graviton (spin2) particles. In adiditon, the high luminosity LHC and the CMS HGCAL upgrade are also presented. The assembly techniques and various tests for the Silicon module and the cassette are discussed.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2022

High-Gradient Wien Spin Rotators at Jefferson Lab

Nuclear physics experiments performed in the Continuous Electron Beam Accelerator Facility (CEBAF) at Jefferson Laboratory (JLab) require spin manipulation of electron beams. Two Wien spin rotators in the injector keV region are essential at CEBAF to establish longitudinal polarization at the end station target, and to flip the polarization direction by ? rad to rule out false asymmetries. In a Wien filter, the homogeneous and independent electric and magnetic fields, along with the velocity vectors of the electrons that traverse it, form a mutually orthogonal system. The magnitude of the electrostatic field, established by biasing two highly-polished electrodes, defines the desired spin angle at the target yet deviates the beam trajectory due to the Lorentz force. The beam trajectory in the Wien is then re-established by adjusting the magnetic field, induced by an electromagnet encasing the device vacuum chamber. This contribution describes the evolution design and high voltage testing of Wien filters for spin manipulation at increased beam energies in the keV injector region, required by high precision parity violation experiments like MOLLER.

Palacios Serrano, G. G.↗
NASA NTRS1972

Longitudinal handling qualities during approach and landing of a powered lift STOL aircraft

Longitudinal handling qualities evaluations were conducted on the Ames Research Center Flight Simulator for Advanced Aircraft (FSAA) for the approach and landing tasks of a powered lift STOL research aircraft. The test vehicle was a C-8A aircraft modified with a new wing incorporating internal blowing over an augmentor flap. The investigation included: (1) use of various flight path and airspeed control techniques for the basic vehicle; (2) assessment of stability and command augmentation schemes for pitch attitude and airspeed control; (3) determination of the influence of longitudinal and vertical force coupling for the power control; (4) determination of the influence of pitch axis coupling with the thrust vector control; and (5) evaluations of the contribution of stability and command augmentation to recovery from a single engine failure. Results are presented in the form of pilot ratings and commentary substantiated by landing approach time histories.

Franklin, J. A.↗
NASA NTRS1980

Performance improvement of delta wings at subsonic speeds due to vortex flaps

Subsonic wind tunnel tests were conducted to determine performance improvements possible from the use of leading edge vortex flaps (LEVF) on delta wings. Various flap sizes and deflection angles were examined and lift-to-drag ratio improvements of up to 40% were found at moderate angles of attack on 60 deg and 75 deg swept wings. The LEVF is concluded to be effective in moving the wing's leading edge vortex onto the flaps, tilting the vortex-induced force vector forward to produce a thrust or reduce the wing's drag while maintaining attached flow and lift on the wing's upper surface.

Marchman, J. F., III↗
NASA NTRS1992

A coupled Eulerian/Lagrangian method for the solution of three-dimensional vortical flows

A coupled Eulerian/Lagrangian method is presented for the reduction of numerical diffusion observed in solutions of three-dimensional rotational flows using standard Eulerian finite-volume time-marching procedures. A Lagrangian particle tracking method using particle markers is added to the Eulerian time-marching procedure and provides a correction of the Eulerian solution. In turn, the Eulerian solutions is used to integrate the Lagrangian state-vector along the particles trajectories. The Lagrangian correction technique does not require any a-priori information on the structure or position of the vortical regions. While the Eulerian solution ensures the conservation of mass and sets the pressure field, the particle markers, used as 'accuracy boosters,' take advantage of the accurate convection description of the Lagrangian solution and enhance the vorticity and entropy capturing capabilities of standard Eulerian finite-volume methods. The combined solution procedures is tested in several applications. The convection of a Lamb vortex in a straight channel is used as an unsteady compressible flow preservation test case. The other test cases concern steady incompressible flow calculations and include the preservation of turbulent inlet velocity profile, the swirling flow in a pipe, and the constant stagnation pressure flow and secondary flow calculations in bends. The last application deals with the external flow past a wing with emphasis on the trailing vortex solution. The improvement due to the addition of the Lagrangian correction technique is measured by comparison with analytical solutions when available or with Eulerian solutions on finer grids. The use of the combined Eulerian/Lagrangian scheme results in substantially lower grid resolution requirements than the standard Eulerian scheme for a given solution accuracy.

Felici, Helene Marie↗
NASA NTRSDate not supplied

Transcriptomics-based Machine Learning (ML) Analysis Predicts Space-Exposed Murine Livers

Limited sample sizes, high data dimensionality, and sensitivity to technical and biological variability of next generation sequencing (NGS), typically limits machine learning (ML) approaches in spaceflight studies that include radiation effects. However, pooling smaller studies while addressing intra- and inter-study variabilities allows for ML predictive modeling. Here, integration methods were applied to whole transcriptome shotgun sequencing (RNA-seq) data from six mouse liver GeneLab datasets (GLDS) (n ranging from 6 to 39 samples) from with a total of 81 spaceflight and ground-control samples to determine top features (i.e. genes) relevant to spaceflight including the effect of radiation exposure. RNASeq counts were normalized for each study, then merged and scaled across all datasets. Data dimensionality was reduced using a minimum redundancy maximum relevance (MRMR) methodology. Redundancy and relevance were computed using the Pearson correlation and F-statistic, respectively. The top 100 MRMR features were used to predict spaceflight vs. ground-control samples using Random Forest (RF), Support Vector Machine (SVM), and Linear Discriminant Analysis (LDA) classifiers with 5-fold cross validation (CV). Principal component analysis (PCA) on the complete feature set versus the MRMR features shows separation between spaceflight samples and ground controls (Figure 1A). The ML-based gene sets were compared against differential gene expression results obtained with DESeq2 from individual GLDS. Using all features or randomly sampled subsets at matching set sizes with MRMR, a maximum classifier accuracy of 69% on the test set over 5 folds. For all classifiers, CV training using at least the top 30 MRMR genes show minimum 89% accuracy and 0.95 AUC value on the test set over 5 folds (Figure 1B). Baseline set analysis on differentially expressed genes (DEGs) identified using padj ≤ 0.05 show 295 DEGs that overlap at least two studies and 13 DEGs that overlap three studies (Figure 1C). Set analysis between the top 100 MRMR features and the DEGs showed 47 genes that overlap at least one study and 24 genes that overlap two studies. Over-representation analysis showed overlapping biological processes related to fatty acid and lipid metabolism which may indicate these processes in the response to spaceflight stressors. MRMR feature selection for the selected ML methods improve performance relative to a classifier built on all features or randomly sampled subsets. Permutation feature importance within the decorrelated MRMR features showed concordance in feature ranking between ML methods. A challenge of applying ML methods across heterogeneous NGS data is accounting for signal:noise. Here, signal validation across studies was shown by intersecting sets between top MRMR genes and DEGs from DESeq2 analysis. Non-intersecting sets introduce opportunity to explore genes relevant to differentiating space flight exposed groups and implementing ML methods across existing NGS datasets may overcome sample size limitations.

Machine Learning↗
DOE OSTI2014

Blue tensor spectrum from particle production during inflation

We discuss a mechanism of particle production during inflation that can result in a blue gravitational wave (GW) spectrum, compatible with the BICEP2 result and with the r < 0.11 limit on the tensor-to-scalar ratio at the Planck pivot scale. The mechanism is based on the production of vector quanta from a rolling pseudo-scalar field. Both the vector and the pseudo-scalar are only gravitationally coupled to the inflaton, to keep the production of inflaton quanta at an unobservable level (the overproduction of non-gaussian scalar perturbations is a generic difficulty for mechanisms that aim to generate a visible GW signal from particle production during inflation). This mechanism can produce a detectable amount of GWs for any inflationary energy scale. The produced GWs are chiral and non-gaussian; both these aspects can be tested with large-scale polarization data (starting from Planck). We study how to reconstruct the pseudo-scalar potential from the GW spectrum.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2026

Feature-Based PMU Event Classification under Variable PMU Participation and Overlapping Events

Danovo Energy Solution's presented its paper named: Feature-Based PMU Event Classification under Variable PMU Participation and Overlapping Events at the 2026 Georgia Tech Fault & Disturbance Analysis Conference. The full paper can be found at OSTI ID# 3169150 Paper Abstract—Phasor Measurement Units (PMUs) stream time synchronized, high-resolution measurements from the grid, enabling data-driven techniques for event detection and classification. Accurate event classification improves grid reliability and stability. Events can be detected by varying numbers of PMUs and exhibit different durations depending on the event type. This variability challenges standard classifiers that require uniform input sizes. Moreover, multiple events may coincide, which increases classification complexity. Standard classifiers assign each instance to the class with the highest predicted probability, whereas overlapping events may exhibit comparable probabilities across multiple classes. In this study, to handle data size variability, we extract a wide range of time–frequency domain features from all available PMUs for each event into a fixed-length vector, facilitating the application of standard machine learning classifiers, including Random Forest, XGBoost, LightGBM, Support Vector Machine, and Multilayer Perceptron. To account for overlapping events, a probabilistic post-processing step is applied. For a given data instance, if multiple predicted class probabilities exceed 30% and the differences between them are less than 10%, the event is assigned to multiple classes. Experiments using real-world PMU data demonstrate that the Random Forest and XGBoost models achieve the highest accuracy, while the proposed post-processing method yields perfect classification performance on external unseen test sets.

Nematirad, Reza [Danova Energy Solutions]↗
NASA NTRS1975

Scanning thermal plumes

Over a three-year period 800 thermal line scans of power plant plumes were made by an airborne scanner, with ground truth measured concurrently at the plants. Computations using centered finite differences in the thermal scanning imagery show a lower bound in the horizontal temperature gradient in excess of 1.6 C/m. Gradients persist to 3 m below the surface. Vector plots of the velocity of thermal fronts are constructed by tracing the front motion in successive thermal images. A procedure is outlined for the two-point ground calibration of a thermal scanner from an equation describing the scanner signal and the voltage for two known temperatures. The modulation transfer function is then calculated by input of a thermal step function and application of digital time analysis techniques using Fast Fourier Transforms to the voltage output. Field calibration tests are discussed. Data accuracy is limited by the level of ground truth effort chosen.

Scarpace, F. L.↗
NASA NTRS2017

Development of the European Service Module Propulsion Subsystem for the Multi-Purpose Crew Vehicle

2013, NASA and the European Space Agency (ESA) entered into an international partnership to develop the European Service Module (ESM) for use on NASA's Multi-Purpose Crew Vehicle (MPCV), also known as Orion. The MPCV will be used as the principal spacecraft for future human space exploration missions beyond low earth orbit. The ESM Propulsion Subsystem (PSS) is a pressure-fed, bi-propellant propulsion system, being developed by Airbus Defense and Space under contract to ESA. For this effort, NASA is responsible for the traditional role of insight/oversight to ensure that the PSS delivered by Airbus meets all MPCV Program requirements. In addition, the NASA Propulsion team also has some unique responsibilities that are a result of the Implementing Agreement (IA) between NASA and ESA for development of the ESM. These responsibilities include: (1) providing the main engine and Thrust Vector Control (TVC) assembly for the PSS. This is being accomplished through the delta qualification and re-use the Space Shuttle Orbital Maneuvering System (OMS) engine and TVC assembly; (2) procurement and delivery of the Auxiliary engines (R-4Ds) for the PSS. These engines are being procured by NASA from Aerojet-Rocketdyne via Lockheed Martin, the prime contractor for the MPCV, per an Airbus-provided specification; and (3) conducting the integrated systems hot-fire test which will qualify the end-to-end PSS for flight on MPCV. This test is being conducted at the NASA White Sands Test Facility (WSTF) using an Airbus-provided test article known as the Propulsion Qualification Model (PQM).

Multi-Purpose Crew Vehicle↗
DOE OSTI2026

Feature-Based PMU Event Classification under Variable PMU Participation and Overlapping Events

This paper is the basis for a presentation help at the 2026 Georgia Tech Fault & Disturbance Analysis Conference, which can be found at OSTI # 3168287 Paper Abstract—Phasor Measurement Units (PMUs) stream time synchronized, high-resolution measurements from the grid, enabling data-driven techniques for event detection and classification. Accurate event classification improves grid reliability and stability. Events can be detected by varying numbers of PMUs and exhibit different durations depending on the event type. This variability challenges standard classifiers that require uniform input sizes. Moreover, multiple events may coincide, which increases classification complexity. Standard classifiers assign each instance to the class with the highest predicted probability, whereas overlapping events may exhibit comparable probabilities across multiple classes. In this study, to handle data size variability, we extract a wide range of time–frequency domain features from all available PMUs for each event into a fixed-length vector, facilitating the application of standard machine learning classifiers, including Random Forest, XGBoost, LightGBM, Support Vector Machine, and Multilayer Perceptron. To account for overlapping events, a probabilistic post-processing step is applied. For a given data instance, if multiple predicted class probabilities exceed 30% and the differences between them are less than 10%, the event is assigned to multiple classes. Experiments using real-world PMU data demonstrate that the Random Forest and XGBoost models achieve the highest accuracy, while the proposed post-processing method yields perfect classification performance on external unseen test sets.

Nematirad, Reza [Danovo Energy Solutions]↗
DOE OSTI2022

J-PLUS: Support vector machine applied to STAR-GALAXY-QSO classification

Context. In modern astronomy, machine learning has proved to be efficient and effective in mining big data from the newest telescopes. Aims. In this study, we construct a supervised machine-learning algorithm to classify the objects in the Javalambre Photometric Local Universe Survey first data release (J-PLUS DR1). Methods. The sample set is featured with 12-waveband photometry and labeled with spectrum-based catalogs, including Sloan Digital Sky Survey spectroscopic data, the Large Sky Area Multi-Object Fiber Spectroscopic Telescope, and VERONCAT – the Veron Catalog of Quasars & AGN. The performance of the classifier is presented with the applications of blind test validations based on RAdial Velocity Extension, the Kepler Input Catalog, the Two Micron All Sky Survey Redshift Survey, and the UV-bright Quasar Survey. A new algorithm was applied to constrain the potential extrapolation that could decrease the performance of the machine-learning classifier. Results. The accuracies of the classifier are 96.5% in the blind test and 97.0% in training cross-validation. The F1-scores for each class are presented to show the balance between the precision and the recall of the classifier. We also discuss different methods to constrain the potential extrapolation.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2015

Erratum to: Constraining couplings of top quarks to the Z boson in $ t\overline{t} $ + Z production at the LHC

We study top quark pair production in association with a Z boson at the Large Hadron Collider (LHC) and investigate the prospects of measuring the couplings of top quarks to the Z boson. To date these couplings have not been constrained in direct measurements. Such a determination will be possible for the first time at the LHC. Our calculation improves previous coupling studies through the inclusion of next-to-leading order (NLO) QCD corrections in production and decays of all unstable particles. We treat top quarks in the narrow-width approximation and retain all NLO spin correlations. To determine the sensitivity of a coupling measurement we perform a binned log-likelihood ratio test based on normalization and shape information of the angle between the leptons from the Z boson decay. The obtained limits account for statistical uncertainties as well as leading theoretical systematics from residual scale dependence and parton distribution functions. We use current CMS data to place the first direct constraints on the ttbZ couplings. We also consider the upcoming high-energy LHC run and find that with 300 inverse fb of data at an energy of 13 TeV the vector and axial ttbZ couplings can be constrained at the 95% confidence level to C_V=0.24^{+0.39}_{-0.85} and C_A=-0.60^{+0.14}_{-0.18}, where the central values are the Standard Model predictions. This is a reduction of uncertainties by 25% and 42%, respectively, compared to an analysis based on leading-order predictions. We also translate these results into limits on dimension-six operators contributing to the ttbZ interactions beyond the Standard Model.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2016

Planck 2015 results: XVII. Constraints on primordial non-Gaussianity

We report that the Planck full mission cosmic microwave background (CMB) temperature and E-mode polarization maps are analysed to obtain constraints on primordial non-Gaussianity (NG). Using three classes of optimal bispectrum estimators – separable template-fitting (KSW), binned, and modal – we obtain consistent values for the primordial local, equilateral, and orthogonal bispectrum amplitudes, quoting as our final result from temperature alone ƒ local NL = 2.5 ± 5.7, ƒ equil NL = -16 ± 70, , and ƒ ortho NL = -34 ± 32 (68% CL, statistical). Combining temperature and polarization data we obtain ƒ local NL = 0.8 ± 5.0, ƒ equil NL = -4 ± 43, and ƒ ortho NL = -26 ± 21 (68% CL, statistical). The results are based on comprehensive cross-validation of these estimators on Gaussian and non-Gaussian simulations, are stable across component separation techniques, pass an extensive suite of tests, and are consistent with estimators based on measuring the Minkowski functionals of the CMB. The effect of time-domain de-glitching systematics on the bispectrum is negligible. In spite of these test outcomes we conservatively label the results including polarization data as preliminary, owing to a known mismatch of the noise model in simulations and the data. Beyond estimates of individual shape amplitudes, we present model-independent, three-dimensional reconstructions of the Planck CMB bispectrum and derive constraints on early universe scenarios that generate primordial NG, including general single-field models of inflation, axion inflation, initial state modifications, models producing parity-violating tensor bispectra, and directionally dependent vector models. We present a wide survey of scale-dependent feature and resonance models, accounting for the “look elsewhere” effect in estimating the statistical significance of features. We also look for isocurvature NG, and find no signal, but we obtain constraints that improve significantly with the inclusion of polarization. The primordial trispectrum amplitude in the local model is constrained to be g local NL = (-0.9 ± 7.7 ) X 10 4 (68% CL statistical), and we perform an analysis of trispectrum shapes beyond the local case. The global picture that emerges is one of consistency with the premises of the ΛCDM cosmology, namely that the structure we observe today was sourced by adiabatic, passive, Gaussian, and primordial seed perturbations.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2015

Downregulation of GAUT12 in Populus deltoides by RNA silencing results in reduced recalcitrance, increased growth and reduced xylan and pectin in a woody biofuel feedstock

The inherent recalcitrance of woody bioenergy feedstocks is a major challenge for their use as a source of second-generation biofuel. Secondary cell walls that constitute the majority of hardwood biomass are rich in cellulose, xylan, and lignin. The interactions among these polymers prevent facile accessibility and deconstruction by enzymes and chemicals. Plant biomass that can with minimal pretreatment be degraded into sugars is required to produce renewable biofuels in a cost-effective manner. The following are the results: GAUT12/IRX8 is a putative glycosyltransferase proposed to be involved in secondary cell wall glucuronoxylan and/or pectin biosynthesis based on concomitant reductions of both xylan and the pectin homogalacturonan (HG) in Arabidopsis irx8 mutants. Two GAUT12 homologs exist in Populus trichocarpa, PtGAUT12.1 and PtGAUT12.2. Knockdown expression of both genes simultaneously has been shown to reduce xylan content in Populus wood. We tested the proposition that RNA interference (RNAi) downregulation of GAUT12.1 alone would lead to increased sugar release in Populus wood, that is, reduced recalcitrance, based on the hypothesis that GAUT12 synthesizes a wall structure required for deposition of xylan and that cell walls with less xylan and/or modified cell wall architecture would have reduced recalcitrance. Using an RNAi approach, we generated 11 Populus deltoides transgenic lines with 50 to 67% reduced PdGAUT12.1 transcript expression compared to wild type (WT) and vector controls. Ten of the eleven RNAi lines yielded 4 to 8% greater glucose release upon enzymatic saccharification than the controls. The PdGAUT12.1 knockdown (PdGAUT12.1-KD) lines also displayed 12 to 52% and 12 to 44% increased plant height and radial stem diameter, respectively, compared to the controls. Knockdown of PdGAUT12.1 resulted in a 25 to 47% reduction in galacturonic acid and 17 to 30% reduction in xylose without affecting total lignin content, revealing that in Populus wood as in Arabidopsis, GAUT12 affects both pectin and xylan formation. Finally, analyses of the sugars present in sequential cell wall extracts revealed a reduction of glucuronoxylan and pectic HG and rhamnogalacturonan in extracts from PdGAUT12.1-KD lines.

09 BIOMASS FUELS↗
DOE OSTI2022

Nonlocal Kernel Network (NKN): a Stable and Resolution-Independent Deep Neural Network.

Neural operators have recently become popular tools for designing solution maps between function spaces in the form of neural networks. Differently from classical scientific machine learning approaches that learn parameters of a known partial differential equation (PDE) for a single instance of the input parameters at a fixed resolution, neural operators approximate the solution map of a family of PDEs [6, 7]. Despite their success, the uses of neural operators are so far restricted to relatively shallow neural networks and confined to learning hidden governing laws. In this work, we propose a novel nonlocal neural operator, which we refer to as nonlocal kernel network (NKN), that is resolution independent, characterized by deep neural networks, and capable of handling a variety of tasks such as learning governing equations and classifying images. Our NKN stems from the interpretation of the neural network as a discrete nonlocal diffusion reaction equation that, in the limit of infinite layers, is equivalent to a parabolic nonlocal equation, whose stability is analyzed via nonlocal vector calculus. The resemblance with integral forms of neural operators allows NKNs to capture long-range dependencies in the feature space, while the continuous treatment of node-to-node interactions makes NKNs resolution independent. The resemblance with neural ODEs, reinterpreted in a nonlocal sense, and the stable network dynamics between layers allow for generalization of NKN’s optimal parameters from shallow to deep networks. This fact enables the use of shallow-to-deep initialization techniques [8]. Our tests show that NKNs outperform baseline methods in both learning governing equations and image classification tasks and generalize well to different resolutions and depths.

97 MATHEMATICS AND COMPUTING↗