Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “CONVOLUTION THEORY”

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 145 records · Page 8
All sourcesNASA NTRSDOE OSTI
DOE OSTI2023

Defect graph neural networks for materials discovery in high-temperature clean-energy applications

We present a graph neural network approach that fully automates the prediction of defect formation enthalpies for any crystallographic site from the ideal crystal structure, without the need to create defected atomic structure models as input. Here we used density functional theory reference data for vacancy defects in oxides, to train a defect graph neural network (dGNN) model that replaces the density functional theory supercell relaxations otherwise required for each symmetrically unique crystal site. Interfaced with thermodynamic calculations of reduction entropies and associated free energies, the dGNN model is applied to the screening of oxides in the Materials Project database, connecting the zero-kelvin defect enthalpies to high-temperature process conditions relevant for solar thermochemical hydrogen production and other energy applications. The dGNN approach is applicable to arbitrary structures with an accuracy limited principally by the amount and diversity of the training data, and it is generalizable to other defect types and advanced graph convolution architectures. In conclusion, it will help to tackle future materials discovery problems in clean energy and beyond.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Correlator convolutional neural networks as an interpretable architecture for image-like quantum matter data

Image-like data from quantum systems promises to offer greater insight into the physics of correlated quantum matter. However, the traditional framework of condensed matter physics lacks principled approaches for analyzing such data. Machine learning models are a powerful theoretical tool for analyzing image-like data including many-body snapshots from quantum simulators. Recently, they have successfully distinguished between simulated snapshots that are indistinguishable from one and two point correlation functions. Thus far, the complexity of these models has inhibited new physical insights from such approaches. Here, we develop a set of nonlinearities for use in a neural network architecture that discovers features in the data which are directly interpretable in terms of physical observables. Applied to simulated snapshots produced by two candidate theories approximating the doped Fermi-Hubbard model, we uncover that the key distinguishing features are fourth-order spin-charge correlators. Our approach lends itself well to the construction of simple, versatile, end-to-end interpretable architectures, thus paving the way for new physical insights from machine learning studies of experimental and numerical data.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2022

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↗
DOE OSTI2024

Theory for Equivariant Quantum Neural Networks

Quantum neural network architectures that have little to no inductive biases are known to face trainability and generalization issues. Inspired by a similar problem, recent breakthroughs in machine learning address this challenge by creating models encoding the symmetries of the learning task. This is materialized through the usage of equivariant neural networks the action of which commutes with that of the symmetry. In this work, we import these ideas to the quantum realm by presenting a comprehensive theoretical framework to design equivariant quantum neural networks (EQNNs) for essentially any relevant symmetry group. We develop multiple methods to construct equivariant layers for EQNNs and analyze their advantages and drawbacks. Our methods can find unitary or general equivariant quantum channels efficiently even when the symmetry group is exponentially large or continuous. As a special implementation, we show how standard quantum convolutional neural networks (QCNNs) can be generalized to group-equivariant QCNNs where both the convolution and pooling layers are equivariant to the symmetry group. We then numerically demonstrate the effectiveness of a S U ( 2 ) -equivariant QCNN over symmetry-agnostic QCNN on a classification task of phases of matter in the bond-alternating Heisenberg model. Our framework can be readily applied to virtually all areas of quantum machine learning. Lastly, we discuss about how symmetry-informed models such as EQNNs provide hopes to alleviate central challenges such as barren plateaus, poor local minima, and sample complexity. Published by the American Physical Society 2024

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
NASA NTRS1988

Autocorrelation and regularization in digital images. I - Basic theory

Spatial structure occurs in remotely sensed images when the imaged scenes contain discrete objects that are identifiable in that their spectral properties are more homogeneous within than between them and other scene elements. The spatial structure introduced is manifest in statistical measures such as the autocovariance function and variogram associated with the scene, and it is possible to formulate these measures explicitly for scenes composed of simple objects of regular shapes. Digital images result from sensing scenes by an instrument with an associated point spread function (PSF). Since there is averaging over the PSF, the effect, termed regularization, induced in the image data by the instrument will influence the observable autocovariance and variogram functions of the image data. It is shown how the autocovariance or variogram of an image is a composition of the underlying scene covariance convolved with an overlap function, which is itself a convolution of the PSF. The functional form of this relationship provides an analytic basis for scene inference and eventual inversion of scene model parameters from image data.

Jupp, David L. B.↗
NASA NTRS1992

Rate 8/9 coded 8-PSK system for downlink applications

An advanced Coded Trellis Modulation (CTM) System which achieves a 2 bits/s/Hz bandwidth efficiency at an information rate of 200 Mbit/s while minimizing satellite power requirements, was developed for downlink earth station applications. The CTM system employs a high-speed rate 8/9 convolutional code with Viterbi decoding and an 8-Phase Shift Keying (PSK) modem. The minimum Euclidean distance between the modulated waveforms corresponding to the information sequences are maximized in order to maximize the noise immunity of the system. Nyquist filters with a square-root of 40 percent roll-off are used at the transmit and receive sides of the modem in order to minimize intersymbol interference, adjacent channel interference, and distortion at the nonlinear satellite power amplifier. The use of a coded system here also minimizes the effects of co-channel interference. The developed performance of the hardware system was measured to achieve within 1.5 dB from theory at a bit error rate of 5 x 10(exp -7) over an additive white Gaussian noise channel.

Fang, Russell↗
DOE OSTI2020

Isomer-Dependent Reaction Mechanisms of Cyclic Ether Intermediates: cis -23-Dimethyloxirane and trans -23-Dimethyloxirane.

Oxiranes are a class of cyclic ethers formed in abundance during low-temperature combustion of hydrocarbons and biofuels, either via chain-propagating steps that occur from unimolecular decomposition of β-hydroperoxyalkyl radicals (β-˙QOOH) or from reactions of HÒO with alkenes. The cis- and trans-isomers of 2,3-dimethyloxirane are intermediates of n-butane oxidation, and while rate coefficients for β-˙QOOH → 2,3-dimethyloxirane + OH are reported extensively, subsequent reaction mechanisms of the cyclic ethers are not. As a result, chemical kinetics mechanisms commonly adopt simplified chemistry to describe the consumption of 2,3-dimethyloxirane by convoluting several elementary reactions into a single step, which may introduce mechanism truncation error—uncertainty derived from missing or incomplete chemistry. The present research examines the isomerdependence of 2,3-dimethyloxirane reaction mechanisms in support of ongoing efforts to minimize mechanism truncation error. Reaction mechanisms are inferred via the detection of products from Cl-initiated oxidation of both cis-2,3-dimethyloxirane and trans-2,3-dimethyloxirane using multiplexed photoionization mass spectrometry (MPIMS). The experiments were conducted at 10 Torr and temperatures of 650 K and 800 K. To complement the experiments, the enthalpies of stationary points on the ˙R + O 2 surfaces were computed at the ccCA-PS3 level of theory. In total, 28 barrier heights were computed on the 2,3-dimethyloxiranylperoxy surfaces. Two notable aspects are low-lying pathways that form resonance-stabilized ketohydroperoxide-type radicals caused by ˙QOOH ring-opening when the unpaired electron is localized adjacent to the ether group, and cis-trans isomerization of ˙R and ˙QOOH radicals, via inversion, which enable reaction pathways otherwise restricted by stereochemistry. Several species were identified in the MPIMS experiments from ring opening of 2,3-dimethyloxiranyl radicals. Neither of the two conjugate alkene isomers prototypical of ˙R + O 2 reactions were detected. Products were also identified from decomposition of ketohydroperoxide-type radicals. The present work provides the first analysis of 2,3-dimethyloxirane oxidation chemistry and reveals that consumption pathways are complex and require the expansion of submechanisms in chemical kinetics mechanisms.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2025

Development of a data-driven neural network model for electron thermal transport in NSTX

A data-driven electron thermal transport neural network (ETT-NN) model, trained on TRANSP interpretative analysis results of National Spherical Torus Experiment (NSTX), was developed to enable faster and more accurate ETT computation for spherical tokamaks (STs). The model incorporates both convolutional NNs and recurrent NNs, allowing it to simultaneously account for the spatial and temporal non-localities and multi-scale features of turbulent transport, which have been considered only in a limited manner in conventional models. The model was validated through interpretative analysis and predictive simulations using Tokamak Reactor Integrated Automated Suite for Simulation and Computation, demonstrating relatively high accuracy. Additionally, parameter scans were performed on test discharges known to exhibit specific turbulent modes, such as microtearing mode, trapped electron mode, kinetic ballooning mode, and electron temperature gradient mode. The scanning results revealed that the ETT-NN model exhibits the same trends as those observed in conventional gyrokinetic simulations or theories, while also capturing the global nature of turbulent transport, indicating that the data-driven model accurately reflects the underlying physical characteristics. Furthermore, due to the dimensionless nature of the model, we can feasibly expand its applicability by incorporating data from other devices and uncovering the characteristics of ETT in STs in the future.

NSTX↗
DOE OSTI2021

Noise and error analysis and optimization in particle-based kinetic plasma simulations

In this paper we analyze the noise in macro-particle methods used in plasma physics and fluid dynamics, leading to approaches for minimizing the total error, focusing on electrostatic models in one dimension. We begin by describing kernel density estimation for continuous values of the spatial variable x, expressing the kernel in a form in which its shape and width are represented separately. The covariance matrix of the noise in the density is computed, first for uniform true density. The bandwidth of the covariance matrix C(x,y) is related to the width of the kernel. A feature that stands out is the presence of constant negative terms in the elements of the covariance matrix both on and off-diagonal. These negative correlations are related to the fact that the total number of particles is fixed at each time step; they also lead to the property ∫C(x,y)dy = 0. We investigate the effect of these negative correlations on the electric field computed by Gauss's law, finding that the noise in the electric field is related to a process called the Ornstein-Uhlenbeck bridge, leading to a covariance matrix of the electric field with variance significantly reduced relative to that of a Brownian process. For non-constant density, p(x), still with continuous x, we analyze the total error in the density estimation and discuss it in terms of bias-variance optimization (BVO). For some characteristic length l, determined by the density and its second derivative, and kernel width h, having too few particles within h leads to too much variance; for h that is large relative to l, there is too much smoothing of the density. The optimum between these two limits is found by BVO. For kernels of the same width, it is shown that this optimum (minimum) is weakly sensitive to the kernel shape. Next, we repeat the analysis for x discretized on a grid. In this case the charge deposition rule is determined by a particle shape. An important property to be respected in the discrete system is the exact preservation of total charge on the grid; this property is necessary to ensure that the electric field is equal at both ends, consistent with periodic boundary conditions. We find that if the particle shapes satisfy a partition of unity property, the particle charge deposited on the grid is conserved exactly. Further, if the particle shape is expressed as the convolution of a kernel with another kernel that satisfies the partition of unity, then the particle shape obeys the partition of unity. This property holds for kernels of arbitrary width, including widths that are not integer multiples of the grid spacing. Furthermore, we show results relaxing the approximations used to do BVO optimization analytically, by doing numerical computations of the total error as a function of the kernel width, on a grid in x. The comparison between numerical and analytical results shows good agreement over a range of particle shapes. We discuss the practical implications of our results, including the criteria for design and implementation of computationally efficient particle shapes that take advantage of the developed theory.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
NASA NTRS1988

Observations of comets and asteroids

All available ground-based observational techniques are used to study the chemical and physical properties of the small bodies of the solar system, primarily comets and secondarily asteroids. The ultimate goal is to use these bodies to understand the formation and evolution of the solar system. The analysis of charge coupled device (CCD) images of comet P/Halley continued. This included a mojor revision of a paper on the periodicities of CN jets in Halley showing that their periodicity is 7.3 days, not 2.2 days. It also included an analysis of the continuum images which showed that the dust in jets is much redder than in the ambient coma. Calculations with Mie theory suggest that the particles that show the strongest effects of radiation pressure. Much effort was expended devising methods to estimate the level of sky background of the many images in which comet extends to the edge of the chip. Examined images of comet Wilson were taken over many months in 1987. Wilson, a dynamically new comet, shows none of the short-term variability or jet structure (either dusty jets or jets of radicals seen in Halley. This is presumably due to the lack of a mantle or the nucleus. Median-imaging of sky was as a technique for removing residual flat-fielding errors in CCD images. A Monte Carlo model and a convolution model were developed for studying temporal variability of gaseous daughter products.

Ahearn, Michael F.↗
DOE OSTI2024

Symmetry-Based Structured Matrices for Efficient Approximately Equivariant Networks

There has been much recent interest in designing symmetry-aware neural networks (NNs) exhibiting relaxed equivariance. Such NNs aim to interpolate between being exactly equivariant and being fully flexible, affording consistent performance benefits. In a separate line of work, certain structured parameter matrices -- those with displacement structure, characterized by low displacement rank (LDR) -- have been used to design small-footprint NNs. Displacement structure enables fast function and gradient evaluation, but permits accurate approximations via compression primarily to classical convolutional neural networks (CNNs). In this work, we propose a general framework -- based on a novel construction of symmetry-based structured matrices -- to build approximately equivariant NNs with significantly reduced parameter counts. Our framework integrates the two aforementioned lines of work via the use of so-called Group Matrices (GMs), a forgotten precursor to the modern notion of regular representations of finite groups. GMs allow the design of structured matrices -- resembling LDR matrices -- which generalize the linear operations of a classical CNN from cyclic groups to general finite groups and their homogeneous spaces. We show that GMs can be employed to extend all the elementary operations of CNNs to general discrete groups. Further, the theory of structured matrices based on GMs provides a generalization of LDR theory focussed on matrices with cyclic structure, providing a tool for implementing approximate equivariance for discrete groups. We test GM-based architectures on a variety of tasks in the presence of relaxed symmetry. We report that our framework consistently performs competitively compared to approximately equivariant NNs, and other structured matrix-based compression frameworks, sometimes with a one or two orders of magnitude lower parameter count.

Samudre, Ashwin↗
NASA NTRS1997

First Test of Stochastic Growth Theory for Langmuir Waves in Earth's Foreshock

This paper presents the first test of whether stochastic growth theory (SGT) can explain the detailed characteristics of Langmuir-like waves in Earth's foreshock. A period with unusually constant solar wind magnetic field is analyzed. The observed distributions P(logE) of wave fields E for two intervals with relatively constant spacecraft location (DIFF) are shown to agree well with the fundamental prediction of SGT, that P(logE) is Gaussian in log E. This stochastic growth can be accounted for semi-quantitatively in terms of standard foreshock beam parameters and a model developed for interplanetary type III bursts. Averaged over the entire period with large variations in DIFF, the P(logE) distribution is a power-law with index approximately -1; this is interpreted in terms of convolution of intrinsic, spatially varying P(logE) distributions with a probability function describing ISEE's residence time at a given DIFF. Wave data from this interval thus provide good observational evidence that SGT can sometimes explain the clumping, burstiness, persistence, and highly variable fields of the foreshock Langmuir-like waves.

Cairns, Iver H.↗
DOE OSTI2021

Understanding Superatomic Ag Nanohydrides

Abstract Bulk Ag hydrides are extremely challenging to make even at very high pressures, but they may become stable as the particle size shrinks to the nanometer regime. Here, the formation and electronic structure of Ag nanohydrides are investigated from a superatomic perspective by density functional theory. It is found that as the coverage increases, adsorption energy of hydrogen atoms on Ag 38 cluster to form Ag 38 H 2 n nanohydride ( n is from 1 to 15) can be energetically favorable with respect to bare Ag 38 and H 2 . Furthermore, the adsorbed hydrogen atoms contribute their 1s electrons to the superatom electron count and behave as a metal instead of a ligand. The electronic structure of the silver nanohydrides follows the superatomic complex model, leading to magic or relatively more stable compositions such as Ag 38 H 2 , Ag 38 H 20 , and Ag 38 H 30 , which correspond to 40‐electron, 58‐electron, and 68‐electron shell closings, respectively. Angular momentum analyses of the superatomic orbitals suggest a convoluted interaction of geometry, symmetry, and orbital splitting.

He, Xiang↗
DOE OSTI2025

Web-Based Tools for Data-Informed Remedy Optimization: Software Theory and User Guide

This report documents the development and application of two web-based decision-support tools for pump-and-treat (P&T) groundwater remediation systems: PTOLEMY (Pump-and-Treat Optimized Location Evaluation to Maximize Yields) and OPTIMA (Optimization for Pump-and-Treat Implementation, Management, & Assessment). These tools enhance remedy design and management by leveraging advanced computational methods – specifically deep learning and multi-objective optimization – within a user-friendly platform. By integrating data-driven models with established hydrogeological knowledge, PTOLEMY and OPTIMA enable more efficient evaluation of well placement and operational strategies, helping site managers balance multiple remediation objectives under complex conditions. Both tools are implemented as modules within the SOCRATES (Suite Of Comprehensive Rapid Analysis Tools for Environmental Sites) web platform, which provides data access, visualization, and analytics to support remedy optimization across sites in the U.S. Department of Energy Office of Environmental Management complex. PTOLEMY is a rapid screening module designed to identify promising locations for new extraction wells. It employs a multi-channel three-dimensional convolutional neural network (MC3D-CNN) trained on high-fidelity simulation data to predict the relative performance (in terms of contaminant mass recovery) of potential well sites. Through an interactive web interface, PTOLEMY visualizes the probability of high performance across a site, highlighting areas where an extraction well is likely to yield above-threshold contaminant removal over a multi-year period. PTOLEMY’s map-based displays and exportable results support transparent communication of screening analyses. By focusing attention on the most favorable candidate locations, the tool augments traditional engineering judgment and physics-based modeling, providing a data informed basis for subsequent detailed evaluations. OPTIMA is a multi objective optimization module designed to find wellfield layouts and operating schedules that meet various cleanup goals. It quickly evaluates thousands of candidate setups – combinations of well locations, timing, and rates – and returns a small set of best trade-off options for comparison. At its core, OPTIMA uses a U-Net-based surrogate model – a deep-learning emulator of a groundwater flow and transport simulator – to dramatically accelerate scenario evaluations. Coupling this fast surrogate with the NSGA-II (Non-dominated Sorting Genetic Algorithm II) evolutionary algorithm, OPTIMA explores a wide decision space of well locations and schedules to identify Pareto-optimal solutions that trade off key objectives (e.g., minimizing cleanup time, maximizing contaminant mass removal, and minimizing plume extent). The tool outputs a family of optimal configurations and visualizes their trade-offs (Pareto frontiers of cleanup metrics and maps of optimized well placements). Site managers can use these results to understand the range of viable strategies and to select candidate designs for more detailed verification. OPTIMA is currently under active development and not yet fully released; this guide provides early documentation to support planning and gather user feedback.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2020

Predicting Elastic Properties of Materials from Electronic Charge Density Using 3D Deep Convolutional Neural Networks

Materials representation plays a key role in machine learning-based prediction of materials properties and new materials discovery. Currently both graph and three-dimensional (3D) voxel representation methods are based on the heterogeneous elements of the crystal structures. Here, we propose to use electronic charge density (ECD) as a generic unified 3D descriptor for materials property prediction with the advantage of possessing close relation with the physical and chemical properties of materials. We developed an ECD-based 3D convolutional neural networks (CNNs) for predicting the elastic properties of materials, in which CNNs can learn effective hierarchical features with multiple convolving and pooling operations. Extensive benchmark experiments over 2170 $Fm\bar3m$ face-centered-cubic materials show that our ECD-based CNNs can achieve good performance for elasticity prediction. Especially, our CNN models based on the fusion of elemental Materials-Agnostic Platform for Informatics and Exploration features and ECD descriptors achieved the best fivefold cross-validation performance. More importantly, we showed that our ECD-based CNN models can achieve significantly better extrapolation performance when evaluated over nonredundant data sets, where there are few neighbor-training samples around test samples. As an additional validation, we evaluated the predictive performance of our models on 329 materials of space group $Fm\bar3m$ by comparing to density functional theory calculated values, which shows a better prediction power of our model for bulk modulus than shear modulus. Because of the unified representation power of ECD, it is expected that our ECD-based CNN approach can also be applied to predict other physical and chemical properties of crystalline materials.

36 MATERIALS SCIENCE↗
DOE OSTI2022

Predicting the oxidation states of Mn ions in the oxygen-evolving complex of photosystem II using supervised and unsupervised machine learning

Abstract Serial Femtosecond Crystallography at the X-ray Free Electron Laser (XFEL) sources enabled the imaging of the catalytic intermediates of the oxygen evolution reaction of Photosystem II (PSII). However, due to the incoherent transition of the S-states, the resolved structures are a convolution from different catalytic states. Here, we train Decision Tree Classifier and K-means clustering models on Mn compounds obtained from the Cambridge Crystallographic Database to predict the S-state of the X-ray, XFEL, and CryoEM structures by predicting the Mn’s oxidation states in the oxygen-evolving complex. The model agrees mostly with the XFEL structures in the dark S 1 state. However, significant discrepancies are observed for the excited XFEL states (S 2 , S 3, and S 0 ) and the dark states of the X-ray and CryoEM structures. Furthermore, there is a mismatch between the predicted S-states within the two monomers of the same dimer, mainly in the excited states. We validated our model against other metalloenzymes, the valence bond model and the Mn spin densities calculated using density functional theory for two of the mismatched predictions of PSII. The model suggests designing a more optimized sample delivery and illumiation systems are crucial to precisely resolve the geometry of the advanced S-states to overcome the noncoherent S-state transition. In addition, significant radiation damage is observed in X-ray and CryoEM structures, particularly at the dangler Mn center (Mn4). Our model represents a valuable tool for investigating the electronic structure of the catalytic metal cluster of PSII to understand the water splitting mechanism.

Plant Sciences↗
DOE OSTI2023

Architecture-Preserving Provable Repair of Deep Neural Networks

Deep neural networks (DNNs) are becoming increasingly important components of software, and are considered the state-of-the-art solution for a number of problems, such as image recognition. However, DNNs are far from infallible, and incorrect behavior of DNNs can have disastrous real-world consequences. This paper addresses the problem of architecture-preserving V-polytope provable repair of DNNs. A V-polytope defines a convex bounded polytope using its vertex representation. V-polytope provable repair guarantees that the repaired DNN satisfies the given specification on the infinite set of points in the given V-polytope. An architecture-preserving repair only modifies the parameters of the DNN, without modifying its architecture. The repair has the flexibility to modify multiple layers of the DNN, and runs in polynomial time. It supports DNNs with activation functions that have some linear pieces, as well as fully-connected, convolutional, pooling and residual layers. To the best our knowledge, this is the first provable repair approach that has all of these features. We implement our approach in a tool called APRNN. Using MNIST, ImageNet, and ACAS Xu DNNs, we show that it has better efficiency, scalability, and generalization compared to PRDNN and REASSURE, prior provable repair methods that are not architecture preserving.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Deep learning insights into cosmological structure formation

The evolution of linear initial conditions present in the early Universe into extended halos of dark matter at late times can be computed using cosmological simulations. However, a theoretical understanding of this complex process remains elusive; in particular, the role of anisotropic information in the initial conditions in establishing the final mass of dark matter halos remains a long-standing puzzle. Here, we build a deep learning framework to investigate this question. We train a three-dimensional convolutional neural network to predict the mass of dark matter halos from the initial conditions, and quantify in full generality the amounts of information in the isotropic and anisotropic aspects of the initial density field about final halo masses. We find that anisotropies add a small, albeit statistically significant amount of information over that contained within spherical averages of the density field about final halo mass. However, the overall scatter in the final mass predictions does not change qualitatively with this additional information, only decreasing from 0.9 dex to 0.7 dex. Given such a small improvement, our results demonstrate that isotropic aspects of the initial density field essentially saturate the relevant information about final halo mass. Therefore, instead of searching for information directly encoded in initial conditions anisotropies, a more promising route to accurate, fast halo mass predictions is to add approximate dynamical information based e.g. on perturbation theory. More broadly, our results indicate that deep learning frameworks can provide a powerful tool for extracting physical insight into cosmological structure formation. Published by the American Physical Society 2024

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗