Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “algorithms and data structure”

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 595 records · Page 33
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Achieving precise multiparameter measurements with distributed optical fiber sensor using wavelength diversity and deep neural networks

The development of advanced distributed optical fiber sensing systems that are capable of performing accurate and spatially resolved multiparameter measurements is of great interest to a wide range of scientific and industrial applications. Here, in this paper, we propose and experimentally demonstrate a wavelength diversity based advanced distributed optical fiber sensor system to accomplish multiparameter sensing while greatly enhancing measurement accuracy. A suite of deep neural network (DNN) algorithms are developed and verified for data denoising, rapid Brillouin frequency shift estimation, and vibration data event classification. As a proof-of-concept, we demonstrate the effectiveness of the proposed advanced wavelength diversity distributed fiber sensor system assisted by DNN for simultaneous, independent measurements of static strain, temperature, and acoustic vibrations over a 25 km long sensing fiber at 3 m spatial resolution. These results suggest the potential for an intelligent multiparameter monitoring system with enhanced performance in advanced structural health monitoring applications.

47 OTHER INSTRUMENTATION↗
DOE OSTI2022

To image, or not to image: class-specific diffractive cameras with all-optical erasure of undesired objects

Abstract Privacy protection is a growing concern in the digital era, with machine vision techniques widely used throughout public and private settings. Existing methods address this growing problem by, e.g., encrypting camera images or obscuring/blurring the imaged information through digital algorithms. Here, we demonstrate a camera design that performs class-specific imaging of target objects with instantaneous all-optical erasure of other classes of objects. This diffractive camera consists of transmissive surfaces structured using deep learning to perform selective imaging of target classes of objects positioned at its input field-of-view. After their fabrication, the thin diffractive layers collectively perform optical mode filtering to accurately form images of the objects that belong to a target data class or group of classes, while instantaneously erasing objects of the other data classes at the output field-of-view. Using the same framework, we also demonstrate the design of class-specific permutation and class-specific linear transformation cameras, where the objects of a target data class are pixel-wise permuted or linearly transformed following an arbitrarily selected transformation matrix for all-optical class-specific encryption, while the other classes of objects are irreversibly erased from the output image. The success of class-specific diffractive cameras was experimentally demonstrated using terahertz (THz) waves and 3D-printed diffractive layers that selectively imaged only one class of the MNIST handwritten digit dataset, all-optically erasing the other handwritten digits. This diffractive camera design can be scaled to different parts of the electromagnetic spectrum, including, e.g., the visible and infrared wavelengths, to provide transformative opportunities for privacy-preserving digital cameras and task-specific data-efficient imaging.

47 OTHER INSTRUMENTATION↗
DOE OSTI2022

Aircraft Icing Severity Evaluation

Aircraft icing refers to the ice buildup on the surface of an aircraft flying in icing conditions. The ice accretion on the aircraft alters the original aerodynamic configuration and degrades the aerodynamic performances and may lead to unsafe flight conditions. Evaluating the flow structure, icing mechanism and consequences is of great importance to the development of an anti/deicing technique. Studies have shown computational fluid dynamics (CFD) and machine learning (ML) to be effective in predicting the ice shape and icing severity under different flight conditions. CFD solves a set of partial differential equations to obtain the air flow fields, water droplets trajectories and ice shape. ML is a branch of artificial intelligence and, based on the data, the self-improved computer algorithms can be effective in finding the nonlinear mapping relationship between the input flight conditions and the output aircraft icing severity features.

Li, Sibo (ORCID:000000021705844X)↗
DOE OSTI2022

Advanced modeling and simulation of research reactors using dynamic mode decomposition

Full text of publication follows. Due to the ever-increasing safety requirements, the current trend of nuclear reactor analysis is shifting towards high-fidelity multi-physics models, which have a very high computational cost and modelling complexity. As the cost of even a single model run makes it impossible to analyse the behaviour and performance of these models on large-scale commercial plants, it has become even more significant to provide suitable benchmarks to validate and test them extensively. In this sense, research reactors offer a promising solution for the initial validation of high-fidelity models, as they are significantly smaller than commercial reactors and their characteristics are well known. In particular, the reactors of the TRIGA family have been used to assess and validate models and methods for Generation-IV designs, as they have some similar features (such as the dominance of natural convection as cooling mechanism and the difficulties in performing sub-channel analysis using standard codes). Still, the computational requirements of high-fidelity models make them unsuitable for real-time analysis, even following their assessment on research reactors. In this sense, Model Order Reduction (MOR) techniques give an additional strategy to reduce the computational cost of high-fidelity models (whilst preserving sufficient accuracy). In particular, this work focuses on Dynamic Mode Decomposition (DMD), a non-intrusive MOR technique that aims at representing models with explicit temporal dynamics by extracting the time-varying characteristics and the governing structures based only on a set of available data, thus without needing any underlying knowledge of the governing equations. In addition, DMD also computes a low-dimensional surrogate of the dynamic matrix of the system, making it suited for stability analysis and real-time evaluations. This work focuses on the application and validation of the DMD method on the Computational Fluid-Dynamics (CFD) model TRIGA Mark II reactor, also discussing in detail the potentiality of this algorithm as an advanced modelling tool for nuclear reactor analysis. (author)

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
DOE OSTI2025

Search for medium effects using jet axis decorrelation in inclusive jets from PbPb collisions at $\sqrt{{s}_{\text{NN}}}$ = 5.02 TeV

The jet axis decorrelation in inclusive jets is studied using lead-lead (PbPb) collisions at a center-of-mass energy per nucleon pair of 5.02 TeV. The jet axis decorrelation is defined as the angular difference between two definitions of the jet axis. It is obtained by applying two recombination schemes on all the constituents of a given jet reconstructed by the anti-kT sequential algorithm with a distance parameter of R = 0.4. The data set, corresponding to an integrated luminosity of 0.66 nb −1 , was collected in 2018 with the CMS detector at the CERN LHC. The jet axis decorrelations are examined across collision centrality selections and intervals of jet transverse momentum. A centrality dependent evolution of the measured distributions is observed, with a progressive narrowing seen in more central events. This narrowing could result from medium-induced modification of the internal jet structure or reflect color charge effects in energy loss. This new measurement probes jet substructure in previously unexplored kinematic domains and show great promise for providing new insights on the color charge dependence of energy loss to jet-quenching models.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2023

Long–short-term memory encoder–decoder with regularized hidden dynamics for fault detection in industrial processes

The ability of recurrent neural networks (RNN) to model nonlinear dynamics of high dimensional process data has enabled data-driven RNN-based fault detection algorithms. Previous studies have focused on detecting faults by identifying the discrepancies in data distribution between the faulty and normal data, as reflected in prediction errors generated by RNN models. However, in industrial processes, variations in data distribution can also result from changes in normal control setpoints and compensatory control adjustments in response to disturbances, making it hard to differentiate between normal and faulty conditions. This paper proposes a fault detection method utilizing a long short-term memory (LSTM) encoder–decoder structure with regularized hidden dynamics and reversible instance normalization (RevIN) to compactly represent high-dimensional measurements for effective monitoring. During training, the hidden states of the model are regularized to form a low-dimensional latent space representation of the original multivariate time series data. As a result, the prediction errors of the latent states can be used to monitor the abnormal dynamic variations, while the reconstruction errors of the measured variables are used to monitor the abnormal static variations. Furthermore, the proposed indices can reflect operating conditions, even when the distribution of test data changes, which helps distinguish faults from normal adjustments and disturbances that controllers can settle. Here, data from numerical simulation and the Tennessee Eastman process are used to illustrate the effectiveness of the proposed fault detection method.

42 ENGINEERING↗
DOE OSTI2020

Constraining the distance to the North Polar Spur with Gaia DR2

ABSTRACT The North Polar Spur (NPS) is one of the largest structures observed in the Milky Way in both the radio and soft X-rays. While several predictions have been made regarding the origin of the NPS, modelling the structure is difficult without precise distance constraints. In this paper, we determine accurate distances to the southern terminus of the NPS and towards latitudes ranging up to 55°. First, we fit for the distance and extinction to stars towards the NPS using optical and near-infrared photometry and Gaia Data Release 2 astrometry. We model these per-star distance–extinction estimates as being caused by dust screens at unknown distances, which we fit for using a nested sampling algorithm. We then compare the extinction to the Spur derived from our 3D dust modelling with integrated independent measures from XMM–Newton X-ray absorption and H i column density measures. We find that we can account for nearly 100 per cent of the total column density of the NPS as lying within 140 pc for latitudes >26° and within 700 pc for latitudes <11°. Based on the results, we conclude that the NPS is not associated with the Galactic Centre or the Fermi bubbles. Instead, it is likely associated, especially at higher latitudes, with the Scorpius–Centaurus association.

Das, Kaustav K.↗
DOE OSTI2021

Prediction of DIII-D Pedestal Structure From Externally Controllable Parameters

The sharp increase of pressure at the edge of a high confinement mode (H-mode) plasma, the pedestal, strongly impacts overall plasma performance. Predicting the pedestal is a necessity to control and optimize tokamak operations. Here, an experimental data-driven machine learning (ML) approach is presented that predicts the pedestal heights and widths of electron density (n e ) and electron temperature (T e ) profiles as well as the separatrix ne from externally controllable parameters such as the plasma shape, heating method and power, and gas puff rate and integrated gas puff. The OMFIT framework was used with DIII-D data to efficiently, robustly, and automatically build a database of pedestal parameters to train machine learning models. Database creation was enabled by the search engine tool for DIII-D data, TokSearch, which parallelizes data fetching, enabling fast searches through basic signals of thousands of DIII-D shots and selection of relevant time intervals. Principal Component Analysis (PCA) separated the database into three clusters that represent classes of plasma shapes that are regularly used in DIII-D. The most important parameters for setting the pedestal structure were plasma current (I p ), toroidal magnetic field (B Φ ), neutral beam heating power (P NBI ) and shaping quantities. The Deep Jointly Informed Neural Networks (DJINN) algorithm was applied to identify suitable neural network (NN) architectures that appropriately capture the features of the pedestal database. Separate NNs were implemented for each pedestal parameter, and ensembling methods were used to improve the prediction accuracy and allowed estimation of the prediction uncertainty. The pedestal predictions of the test dataset lie within the measurement uncertainties of the pedestal parameters. The NN outperformed simple Linear Regression (LR) analysis, indicating non-linear dependencies in the pedestal structure. The presented achievements illustrate a promising path for future research, using feature extraction to infer experimental trends and thereby improve pedestal models as well as deploying NN for a fast pedestal prediction in DIII-D scenario development.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2023

The direct and indirect drivers shaping RNA viral communities in grassland soil

Recent studies have revealed diverse RNA viral communities in soils. Yet, how environmental factors influence soil RNA viruses remains largely unknown. Here, we recovered RNA viral communities from 24 metatranscriptomes sequenced from grassland soils managed under a range of environmental conditions including 1) water content: 100% and 25% water holding capacity, 2) plant presence: planted with tall wheatgrass (Thinopyrum ponticum) and bare soil, 3) cultivar type: Alkar and Jose, and 4) soil depth: 0-5 cm and 15-25 cm. The recovered RNA viral communities were novel with nearly one-third of the RNA viral contigs uniquely detected in the studied grassland. The classified RNA viral contigs are mostly known as eukaryotic RNA viruses (74.7%) belonging to Phyla Duplornaviricota, Kitrinoviricota, Lenarviricota, and Pisuviricota. Eukaryotic RNA viruses of Family Mitoviridae as well as their natural hosts, Fungi, are one of the most dominant taxa. Consistent with the results of nonmetric multidimensional scaling analysis, the four environmental conditions (water content, plant presence, cultivar, and soil depth) significantly influence the assemblages of soil RNA viral communities as suggested by the correlation analysis and the random forest algorithm. The modularity analysis of the factor network and the structural equation modeling further support the hierarchical associations among the four environmental factors and the community factors representing the co-existing eukaryotic, prokaryotic, and RNA viral communities. The soil water content, plant presence, and type of cultivar demonstrate a significant positive impact on eukaryotic RNA viral richness directly as well as indirectly on eukaryotic RNA viral abundance via influencing the co-existing eukaryotic members in this soil. Our data also provide statistical support for the negative influence of soil depth on soil eukaryotic richness and abundances resulting in its indirect impact on soil eukaryotic RNA viral communities. This study provides field-relevant information on how environmental and community factors collectively shape soil RNA communities and contribute to ecological understanding of RNA viral survival under various environmental conditions and virus-host interactions in soil.

Wu, Ruonan↗
DOE OSTI2020

Scalable All-pairs Shortest Paths for Huge Graphs on Multi-GPU Clusters

We present an optimized Floyd-Warshall (Floyd-Warshall) algorithm that computes the All-pairs shortest path (APSP) for GPU accelerated clusters. The Floyd-Warshall algorithm due to its structural similarities to matrix-multiplication is well suited for highly parallel GPU architectures. To achieve high parallel efficiency, we address two key algorithmic challenges: reducing high communication overhead and addressing limited GPU memory. To reduce high communication costs, we redesign the parallel (a) to expose more parallelism, (b) aggressively overlap communication and computation with pipelined and asynchronous scheduling of operations, and (c) tailored MPI-collective. To cope with limited GPU memory, we employ an offload model, where the data resides on the host and is transferred to GPU on-demand. The proposed optimizations are supported with detailed performance models for tuning. Our optimized parallel Floyd-Warshall implementation is up to 5x faster than a strong baseline and achieves 8.1 PetaFLOPS/sec on 256~nodes of the Summit supercomputer at Oak Ridge National Laboratory. This performance represents 70% of the theoretical peak and 80% parallel efficiency. The offload algorithm can handle 2.5x larger graphs with a 20% increase in overall running time.

Sao, Piyush↗
DOE OSTI2021

Nonlinear Matrix Approximation with Radial Basis Function Components

We introduce and investigate matrix approximation by decomposition into a sum of radial basis function (RBF) components. An RBF component is a generalization of the outer product between a pair of vectors, where an RBF function replaces the scalar multiplication between individual vector elements. Even though the RBF functions are positive definite, the summation across components is not restricted to convex combinations and allows us to compute the decomposition for any real matrix that is not necessarily symmetric or positive definite. We formulate the problem of seeking such a decomposition as an optimization problem with a nonlinear and non-convex loss function. Several modern versions of the gradient descent method, including their scalable stochastic counterparts, are used to solve this problem. We provide extensive empirical evidence of the effectiveness of the RBF decomposition and that of the gradient-based fitting algorithm. While being conceptually motivated by singular value decomposition (SVD), our proposed nonlinear counterpart outperforms SVD by drastically reducing the memory required to approximate a data matrix with the same L2 error for a wide range of matrix types. For example, it leads to 2 to 6 times memory save for Gaussian noise, graph adjacency matrices, and kernel matrices. Moreover, this proximity-based decomposition can offer additional interpretability in applications that involve, e.g., capturing the inner low-dimensional structure of the data, retaining graph connectivity structure, and preserving the acutance of images.

Rebrova, Elizaveta↗
DOE OSTI2022

Mesoflow: An Open-Source Reacting Flow Solver for Catalysis at Mesoscale

We present the capabilities and software performance metrics of our open-source continuum solver for catalysis, Mesoflow, developed specifically for modeling transport and chemistry at the mesoscale. Our solver utilizes Cartesian block-structured adaptive mesh refinement to resolve complex catalyst surface morphologies directly obtained from X-ray tomography data. An immersed boundary based formulation enables rapid representation of complex geometries prevalent in most mesoporous catalyst interfaces. The solver is developed on top of open-source performance portable library, AMReX, providing parallel execution capabilities on current and upcoming high-performance-computing (HPC) architectures. Our flexible software framework enables integration of complex chemical mechanisms at heterogenous interfaces and time-split algorithms for circumventing highly disparate reaction and flow time-scales. Our current studies indicate a ten-fold performance gain by using graphics-processing-units (GPUs) compared to a single processor for representative problem sizes (2 million cell mesh). We will also present a brief introduction on how to build and use this software for application problems pertaining to catalytic upgrading and gas transport within porous catalyst particles.

adaptive meshing↗
DOE OSTI2020

Imaging of Calibrated Defects in Additively Manufactured Materials: Pulsed Thermal Tomography Nondestructive Examination of Additively Manufactured Reactor Materials and Components

Additive manufacturing (AM, or 3D printing) for commercial nuclear energy applications is an emerging method for cost-efficient manufacturing aimed at replacing aging nuclear reactor parts and reducing costs for new construction. Known material flaws in AM include low-density regions consisting of non-sintered powder, which have to be detected to ensure the safety of long-term performance nuclear reactor components. Currently, limited options are available for nondestructive evaluation (NDE), either during or post manufacturing. As a solution to NDE of AM, we are developing pulsed thermal imaging which is non-contact, one-sided, and scalable to arbitrary size and shape of the AM parts. Pulsed thermography (PT) system utilizes a high intensity flash lamp to rapidly heat surface of sample, and a high sensitivity fast-frame megapixel infrared (IR) camera to capture data of surface temperature variations. The acquired data cube consists of a stack of surface temperature images taken at different times. Information about material internal defects is extracted by analyzing the data cube. This report provides results of preliminary performance evaluation of pulsed thermal imaging capability in detection of imprinted flaws in AM metallic structures. The flaws were introduced into AM parts as imprinted hemispherical low density regions, consisting of trapped un-sintered metallic powder. Specimens for developed for this study consisted of AM stainless steel 316 and Inconel 718 plates. The diameters of imprinted defects varied from 1mm to 8mm, and their depths below the plate flat surface varied between 1mm and 6mm. Pulsed thermal tomography (PTT) processes the measured data cube to obtain 3D reconstructions of material effusivity using a unique inversion algorithm developed at Argonne. PTT has been previously used in imaging of similar size flat bottom hole (FBH) simulated defects in stainless steel 316 and Inconel 718 specimens. In the study involving AM specimens, PTT imaging results have shown that 1mm-diameter defects located 1mm and 2mm below the surface of specimens were detectable. Larger size defects were detectable at greater depth. We also explored an alternative approach to detection of material flaws in PT data cube, which is using neural learning-based approach to blind source separation. Detection of small material defects requires finding features in the data cube which have signal contrast levels approaching sensitivity limit of IR camera. In this study, an optimized Neural Learning based Blind Source Separation (NLBSS) algorithm, including Principal Component Analysis (PCA), and Independent Component Analysis (ICA) is demonstrated to automatically extract principal temporal and spatial features of thermography frames to enhance flaw detection. By using the NLBSS algorithm, material internal defects can be automatically detected. Furthermore, this processing approach compensates for experimental thermal imaging artifacts, such as noise and uneven heating. By merging artificial intelligence with phtotothermics, the NDE system detects internal calibrated defects of various sizes and depths in AM nuclear-grade metallic alloys.

36 MATERIALS SCIENCE↗
DOE OSTI2026

Enhancing weak lensing redshift distribution characterization by optimizing the Dark Energy Survey Self-Organizing Map Photo-z method

Characterization of the redshift distribution of ensembles of galaxies is pivotal for large scale structure cosmological studies. In this work, we focus on improving the Self-Organizing Map (SOM) methodology for photometric redshift estimation (SOMPZ), specifically in anticipation of the Dark Energy Survey Year 6 (DES Y6) data. This data set, featuring deeper and fainter galaxies than DES Year 3 (DES Y3), demands adapted techniques to ensure accurate recovery of the underlying redshift distribution. We investigate three strategies for enhancing the existing SOM-based approach used in DES Y3: 1) Replacing the Y3 SOM algorithm with one tailored for redshift estimation challenges; 2) Incorporating $\textit{g}$-band flux information to refine redshift estimates (i.e. using $\textit{griz}$ fluxes as opposed to only $\textit{riz}$); 3) Augmenting redshift data for galaxies where available. These methods are applied to DES Y3 data, and results are compared to the Y3 fiducial ones. Our analysis indicates significant improvements with the first two strategies, notably reducing the overlap between redshift bins. By combining strategies 1 and 2, we have successfully managed to reduce redshift bin overlap in DES Y3 by up to 66$\%$. Conversely, the third strategy, involving the addition of redshift data for selected galaxies as an additional feature in the method, yields inferior results and is abandoned. Our findings contribute to the advancement of weak lensing redshift characterization and lay the groundwork for better redshift characterization in DES Year 6 and future stage IV surveys, like the Rubin Observatory.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2024

Uncertainty-Aware, Structure-Preserving Machine Learning Approach for Domain Shift Detection From Nonlinear Dynamic Responses of Structural Systems

Complex structural systems deployed for aerospace, civil, or mechanical applications must operate reliably under varying operational conditions. Structural health monitoring (SHM) systems help ensure the reliability of these systems by providing continuous monitoring of the state of the structure. SHM relies on synthesizing measured data with a predictive model to make informed decisions about structural states. However, these models—which may be thought of as a form of a digital twin—need to be updated continuously as structural changes (e.g., due to damage) arise. We propose an uncertainty-aware machine learning model that enforces distance preservation of the original input state space and then encodes a distance-aware mechanism via a Gaussian process (GP) kernel. The proposed approach leverages the spectral-normalized neural GP algorithm to combine the flexibility of neural networks with the advantages of GP, subjected to structure-preserving constraints, to produce an uncertainty-aware model. This model is used to detect domain shift due to structural changes that cannot be observed directly because they may be spatially isolated (e.g., inside a joint or localized damage). This work leverages detection theory to detect domain shift systematically given statistical features of the prediction variance produced by the model. The proposed approach is demonstrated on a nonlinear structure being subjected to damage conditions. In conclusion, it is shown that the proposed approach is able to rely on distances of the transformed input state space to predict increased variance in shifted domains while being robust to normative changes.

Algorithms↗
DOE OSTI2024

Manifold Learning: What, How, and Why

Manifold learning (ML), also known as nonlinear dimension reduction, is a set of methods to find the low-dimensional structure of data. Dimension reduction for large, high-dimensional data is not merely a way to reduce the data; the new representations and descriptors obtained by ML reveal the geometric shape of high-dimensional point clouds and allow one to visualize, denoise, and interpret them. This review presents the underlying principles of ML, its representative methods, and their statistical foundations, all from a practicing statistician's perspective. It describes the trade-offs and what theory tells us about the parameter and algorithmic choices we make in order to obtain reliable conclusions.

Mathematics↗
DOE OSTI2024

Toward Accelerated Nuclear-physics Parameter Estimation from Binary Neutron Star Mergers: Emulators for the Tolman–Oppenheimer–Volkoff Equations

Abstract Gravitational-wave observations of binary neutron-star (BNS) mergers have the potential to revolutionize our understanding of the nuclear equation of state (EOS) and the fundamental interactions that determine its properties. However, Bayesian parameter estimation frameworks do not typically sample over microscopic nuclear-physics parameters that determine the EOS. One of the major hurdles in doing so is the computational cost involved in solving the neutron-star structure equations, known as the Tolman–Oppenheimer–Volkoff (TOV) equations. In this paper, we explore approaches to emulating solutions for the TOV equations: multilayer perceptrons (MLPs), Gaussian processes, and a data-driven variant of the reduced basis method (RBM). We implement these emulators for three different parameterizations of the nuclear EOS, each with a different degree of complexity represented by the number of model parameters. We find that our MLP-based emulators are generally more accurate than the other two algorithms, whereas the RBM results in the largest speedup with respect to the full high-fidelity TOV solver. We employ these emulators for a simple parameter inference using a potentially loud BNS observation and show that the posteriors predicted by our emulators are in excellent agreement with those obtained from the full TOV solver.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2025

Learning nuclear cross sections across the chart of nuclides with graph neural networks

We explore the use of deep learning techniques to learn how nuclear cross sections change as we add or remove protons and neutrons. As a proof of principle, we focus on the neutron-induced reactions in the fast energy regime. Our approach follows a two-stage learning framework. First, we apply representation learning to encode cross section data into a latent space using either variational autoencoders (VAEs) or implicit neural representations (INRs). Then, we train graph neural networks (GNNs) on the resulting embeddings to predict missing values across the nuclear chart by leveraging the topological structure of neighboring isotopes. We demonstrate accurate cross section predictions within a 9 × 9 block of missing nuclei. We also find that the optimal GNN training strategy depends on the type of latent representation used, with VAE embeddings performing best under end-to-end optimization in the original space, while INR embeddings achieve better results when the GNN is trained only in the latent space. Furthermore, using clustering algorithms, we map groups of latent vectors into regions of the nuclear chart and show that VAEs and INRs can discover some of the neutron magic numbers. These findings suggest that deep-learning models based on the representation encoding of cross sections combined with graph neural networks hold significant potential in augmenting nuclear theory models, e.g., by providing reliable estimates of covariances of cross sections, including cross-material covariances.

Machine learning↗