Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “deep”

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 883 records · Page 49

DEEP GALEX OBSERVATIONS OF THE COMA CLUSTER: SOURCE CATALOG AND GALAXY COUNTS

We present a source catalog from a deep 26 ks Galaxy Evolution Explorer (GALEX) observation of the Coma cluster in the far-UV (FUV; 1530 A) and near-UV (NUV; 2310 A) wavebands. The observed field is centered {approx}0.{sup 0}9 (1.6 Mpc) southwest of the Coma core in a well-studied region of the cluster known as 'Coma-3'. The entire field is located within the apparent virial radius of the Coma cluster, and has optical photometric coverage with Sloan Digital Sky Survey (SDSS) and deep spectroscopic coverage to r {approx} 21. We detect GALEX sources to NUV = 24.5 and FUV = 25.0, which corresponds to a star formation rate of {approx}10{sup -3} M {sub sun} yr{sup -1} for galaxies at the distance of Coma. We have assembled a catalog of 9700 galaxies with GALEX and SDSS photometry, including 242 spectroscopically confirmed Coma member galaxies that span a large range of galaxy types from giant spirals and elliptical galaxies to dwarf irregular and early-type galaxies. The full multi-wavelength catalog (cluster plus background galaxies) is {approx}80% complete to NUV = 23 and FUV = 23.5. The GALEX images presented here are very deep and include detections of many resolved cluster members superposed on a dense field of unresolved background galaxies. This required a two-fold approach to generating a source catalog: we used a Bayesian deblending algorithm to measure faint and compact sources (using SDSS coordinates as position prior), and used the GALEX pipeline catalog for bright and/or extended objects. We performed simulations to assess the importance of systematic effects (e.g., object blends, source confusion, Eddington Bias) that influence the source detection and photometry when using both methods. The Bayesian deblending method roughly doubles the number of source detections and provides reliable photometry to a few magnitudes deeper than the GALEX pipeline catalog. This method is free from source confusion over the UV magnitude range studied here; we estimate that the GALEX pipeline catalogs are confusion limited at NUV {approx} 23 and FUV {approx} 24. We have measured the UV field galaxy counts using our catalog and report a {approx}50% (30%) excess of counts across FUV = 22-23.5 (NUV = 21.5-23) relative to other GALEX studies. Our number counts are a better match to deeper UV galaxy counts measured with Hubble Space Telescope.

79 ASTRONOMY AND ASTROPHYSICS↗

CANDELS MULTIWAVELENGTH CATALOGS: SOURCE IDENTIFICATION AND PHOTOMETRY IN THE CANDELS UKIDSS ULTRA-DEEP SURVEY FIELD

We present the multiwavelength-ultraviolet to mid-infrared-catalog of the UKIRT Infrared Deep Sky Survey (UKIDSS) Ultra-Deep Survey field observed as part of the Cosmic Assembly Near-infrared Deep Extragalactic Legacy Survey (CANDELS). Based on publicly available data, the catalog includes the CANDELS data from the Hubble Space Telescope (near-infrared WFC3 F125W and F160W data and visible ACS F606W and F814W data); u-band data from CFHT/Megacam; B, V, R{sub c} , i', and z' band data from Subaru/Suprime-Cam; Y and K{sub s} band data from VLT/HAWK-I; J, H, and K band data from UKIDSS (Data Release 8); and Spitzer/IRAC data (3.6, 4.5 {mu}m from SEDS; 5.8 and 8.0 {mu}m from SpUDS). The present catalog is F160W-selected and contains 35, 932 sources over an area of 201.7 arcmin{sup 2} and includes radio- and X-ray-detected sources and spectroscopic redshifts available for 210 sources.

79 ASTRONOMY AND ASTROPHYSICS↗

THE HST EXTREME DEEP FIELD (XDF): COMBINING ALL ACS AND WFC3/IR DATA ON THE HUDF REGION INTO THE DEEPEST FIELD EVER

The eXtreme Deep Field (XDF) combines data from 10 years of observations with the Hubble Space Telescope Advanced Camera for Surveys (ACS) and the Wide-Field Camera 3 Infra-Red (WFC3/IR) into the deepest image of the sky ever in the optical/near-IR. Since the initial observations of the Hubble Ultra-Deep Field (HUDF) in 2003, numerous surveys and programs, including supernovae follow-up, HUDF09, CANDELS, and HUDF12, have contributed additional imaging data across this region. However, these images have never been combined and made available as one complete ultra-deep image dataset. We combine them now with the XDF program. Our new and improved processing techniques provide higher quality reductions of the total dataset. All WFC3/IR and optical ACS data sets have been fully combined and accurately matched, resulting in the deepest imaging ever taken at these wavelengths, ranging from 29.1 to 30.3 AB mag (5σ in a 0.''35 diameter aperture) in 9 filters. The combined image therefore reaches to 31.2 AB mag 5σ (32.9 at 1σ) for a flat f {sub ν} source. The gains in the optical for the four filters done in the original ACS HUDF correspond to a typical improvement of 0.15 mag, with gains of 0.25 mag in the deepest areas. Such gains are equivalent to adding ∼130 to ∼240 orbits of ACS data to the HUDF. Improved processing alone results in a typical gain of ∼0.1 mag. Our 5σ (optical+near-IR) SExtractor catalogs reveal about 14,140 sources in the full field and about 7121 galaxies in the deepest part of the XDF.

79 ASTRONOMY AND ASTROPHYSICS↗

Deep chalcogen donors and electron localization in aluminum nitride

Abstract We show with hybrid density functional theory calculations that chalcogen donors other than oxygen (i.e. S N , Se N , and Te N ) give rise to deep donor states in aluminum nitride. These donors trap a localized electron in their neutral charge state, leading to deep ( + / 0 ) donor levels that are 0.45 eV or more from the conduction-band edge. As such, this behavior is distinct from the DX behavior leads to deep (+/−) levels which affects other donors such as O N and Si Al . We highlight how these results hint at the formation of small electron polarons in AlN, which are found to be unstable in the bulk, but metastable when bound to donor dopants like Si Al and the chalocogens, with activation energies on the order of 0.2–0.3 eV. These results indicate that S, Se, and Te are not shallow donor dopants in aluminum nitride and identify origins of the experimentally observed ∼200–300 meV activation energies for dopant activation in donor-doped samples.

aluminum nitride↗

Slicing with deep learning models at ProtoDUNE-SP

DUNE is a cutting-edge experiment aiming to study neutrinos in detail, with a special focus on the flavor oscillation mechanism. The prototype of the DUNE Far Detector Single Phase TPC (ProtoDUNE-SP) was built and operated at CERN with a full set of reconstruction tools. To implement these reconstruction tools, Pandora, a multi-algorithm framework, has been developed. A large number of these algorithms, some of them being exploiting traditional clustering, detector physics and deep learning approaches, have been applied to images to gradually build up a picture out of singular events. One of such algorithms is the Pandora slicing algorithm which aims to partition the detector hits of an event in sets called slices. Each slice represents a single interaction in the detector and should identify all the hits related to the interacting particle and its subsequent decay products. We expect the order of tens of slices per event in ProtoDUNE-SP. In this paper we present a deep learning approach to the problem, designing a model able to outperform the state-of-the-art slicing algorithm which is currently implemented within Pandora. We assess the performance of our tool in terms of efficiency and accuracy, while exploiting hardware accelerating setups. The ultimate goal is to incorporate this deep learning approach in the Pandora reconstruction tool.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Reducing model bias in a deep learning classifier using domain adversarial neural networks in the MINERvA experiment

We present a simulation-based study using deep convolutional neural networks (DCNNs) to identify neutrino interaction vertices in the MINERvA passive targets region, and illustrate the application of domain adversarial neural networks (DANNs) in this context. DANNs are designed to be trained in one domain (simulated data) but tested in a second domain (physics data) and utilize unlabeled data from the second domain so that during training only features which are unable to discriminate between the domains are promoted. MINERvA is a neutrino-nucleus scattering experiment using the NuMI beamline at Fermilab. A-dependent cross sections are an important part of the physics program, and these measurements require vertex finding in complicated events. To illustrate the impact of the DANN we used a modified set of simulation in place of physics data during the training of the DANN and then used the label of the modified simulation during the evaluation of the DANN. In conclusion, we find that deep learning based methods offer significant advantages over our prior track-based reconstruction for the task of vertex finding, and that DANNs are able to improve the performance of deep networks by leveraging available unlabeled data and by mitigating network performance degradation rooted in biases in the physics models used for training.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Deep decarbonization and U.S. biofuels production: a coordinated analysis with a detailed structural model and an integrated multisectoral model

Scenarios for deep decarbonization involve biomass for biofuels, biopower, and bioproducts, and they often include negative emissions via carbon capture and storage or utilization. However, critical questions remain about the feasibility of rapid growth to high levels of biomass utilization, given biomass and land availability as well as historical growth rates of the biofuel industry. We address these questions through a unique coordinated analysis and comparison of carbon pricing effects on biomass utilization growth in the United States using a multisectoral integrated assessment model, the Global Change Analysis Model (GCAM), and a biomass-to-biofuels system dynamics model, the Bioenergy Scenario Model (BSM). We harmonized and varied key factors—such as carbon prices, vehicle electrification, and arable land availability—in the two models. We varied the rate of biorefinery construction, the fungibility of feedstock types across conversion processes, and policy incentives in BSM. The rate of growth in biomass deployment under a carbon price in both models is within the range of current literature. However, the reallocation of land to biomass feedstocks would need to overcome bottlenecks to achieve growth consistent with deep decarbonization scenarios. Investments as a result of near-term policy incentives can develop technology and expand capacity—reducing costs, enabling flexibility in feedstock use, and improving stability—but if biomass demand is high, these investments might not overcome land reallocation bottlenecks. Biomass utilization for deep decarbonization relies on extraordinary growth in biomass availability and industrial capacity. In this paper, we quantify and describe the potential challenges of this rapid change.

09 BIOMASS FUELS↗

Feature learning and generalization in deep networks with orthogonal weights

Fully-connected deep neural networks with weights initialized from independent Gaussian distributions can be tuned to criticality, which prevents the exponential growth or decay of signals propagating through the network. However, such networks still exhibit fluctuations that grow linearly with the depth of the network, which may impair the training of networks with width comparable to depth. We show analytically that rectangular networks with tanh activations and weights initialized from the ensemble of orthogonal matrices have corresponding preactivation fluctuations which are independent of depth, to leading order in inverse width. Moreover, we demonstrate numerically that, at initialization, all correlators involving the neural tangent kernel (NTK) and its descendants at leading order in inverse width—which govern the evolution of observables during training—saturate at a depth of ~20, rather than growing without bound as in the case of Gaussian initializations. We speculate that this structure preserves finite-width feature learning while reducing overall noise, thus improving both generalization and training speed in deep networks with depth comparable to width. We provide some experimental justification by relating empirical measurements of the NTK to the superior performance of deep non-linear orthogonal networks trained under full-batch gradient descent on the MNIST and CIFAR-10 classification tasks.

97 MATHEMATICS AND COMPUTING↗

Deep structural clustering for single-cell RNA-seq data jointly through autoencoder and graph neural network

Abstract Single-cell RNA sequencing (scRNA-seq) permits researchers to study the complex mechanisms of cell heterogeneity and diversity. Unsupervised clustering is of central importance for the analysis of the scRNA-seq data, as it can be used to identify putative cell types. However, due to noise impacts, high dimensionality and pervasive dropout events, clustering analysis of scRNA-seq data remains a computational challenge. Here, we propose a new deep structural clustering method for scRNA-seq data, named scDSC, which integrate the structural information into deep clustering of single cells. The proposed scDSC consists of a Zero-Inflated Negative Binomial (ZINB) model-based autoencoder, a graph neural network (GNN) module and a mutual-supervised module. To learn the data representation from the sparse and zero-inflated scRNA-seq data, we add a ZINB model to the basic autoencoder. The GNN module is introduced to capture the structural information among cells. By joining the ZINB-based autoencoder with the GNN module, the model transfers the data representation learned by autoencoder to the corresponding GNN layer. Furthermore, we adopt a mutual supervised strategy to unify these two different deep neural architectures and to guide the clustering task. Extensive experimental results on six real scRNA-seq datasets demonstrate that scDSC outperforms state-of-the-art methods in terms of clustering accuracy and scalability. Our method scDSC is implemented in Python using the Pytorch machine-learning library, and it is freely available at https://github.com/DHUDBlab/scDSC.

Gan, Yanglan↗

Deep graph representations embed network information for robust disease marker identification

We report that the accurate disease diagnosis and prognosis based on omics data rely on the effective identification of robust prognostic and diagnostic markers that reflect the states of the biological processes underlying the disease pathogenesis and progression. In this article, we present GCNCC, a Graph Convolutional Network-based approach for Clustering and Classification, that can identify highly effective and robust network-based disease markers. Based on a geometric deep learning framework, GCNCC learns deep network representations by integrating gene expression data with protein interaction data to identify highly reproducible markers with consistently accurate prediction performance across independent datasets possibly from different platforms. GCNCC identifies these markers by clustering the nodes in the protein interaction network based on latent similarity measures learned by the deep architecture of a graph convolutional network, followed by a supervised feature selection procedure that extracts clusters that are highly predictive of the disease state. By benchmarking GCNCC based on independent datasets from different diseases (psychiatric disorder and cancer) and different platforms (microarray and RNA-seq), we show that GCNCC outperforms other state-of-the-art methods in terms of accuracy and reproducibility.

59 BASIC BIOLOGICAL SCIENCES↗

Deep reinforcement learning of cell movement in the early stage of C.elegans embryogenesis

Motivation: Cell movement in the early phase of Caenorhabditis elegans development is regulated by a highly complex process in which a set of rules and connections are formulated at distinct scales. Previous efforts have demonstrated that agent-based, multi-scale modeling systems can integrate physical and biological rules and provide new avenues to study developmental systems. However, the application of these systems to model cell movement is still challenging and requires a comprehensive understanding of regulatory networks at the right scales. Recent developments in deep learning and reinforcement learning provide an unprecedented opportunity to explore cell movement using 3D time-lapse microscopy images. Results: We present a deep reinforcement learning approach within an agent-based modeling system to characterize cell movement in the embryonic development of C.elegans. Our modeling system captures the complexity of cell movement patterns in the embryo and overcomes the local optimization problem encountered by traditional rule-based, agent-based modeling that uses greedy algorithms. Here, we tested our model with two real developmental processes: the anterior movement of the Cpaaa cell via intercalation and the rearrangement of the superficial left-right asymmetry. In the first case, the model results suggested that Cpaaa’s intercalation is an active directional cell movement caused by the continuous effects from a longer distance (farther than the length of two adjacent cells), as opposed to a passive movement caused by neighbor cell movements. In the second case, a leader-follower mechanism well explained the collective cell movement pattern in the asymmetry rearrangement. These results showed that our approach to introduce deep reinforcement learning into agent-based modeling can test regulatory mechanisms by exploring cell migration paths in a reverse engineering perspective. Lastly, this model opens new doors to explore the large datasets generated by live imaging. Availability and implementation: Source code is available at https://github.com/zwang84/drl4cellmovement.

59 BASIC BIOLOGICAL SCIENCES↗

Dark Energy Survey Year 3 Results: Deep Field optical + near-infrared images and catalogue

ABSTRACT We describe the Dark Energy Survey (DES) Deep Fields, a set of images and associated multiwavelength catalogue (ugrizJHKs) built from Dark Energy Camera (DECam) and Visible and Infrared Survey Telescope for Astronomy (VISTA) data. The DES Deep Fields comprise 11 fields (10 DES supernova fields plus COSMOS), with a total area of ∼30 sq. deg. in ugriz bands and reaching a maximum i-band depth of 26.75 (AB, 10σ, 2 arcsec). We present a catalogue for the DES 3-yr cosmology analysis of those four fields with full 8-band coverage, totalling 5.88 sq. deg. after masking. Numbering 2.8 million objects (1.6 million post-masking), our catalogue is drawn from images coadded to consistent depths of r = 25.7, i = 25, and z = 24.3 mag. We use a new model-fitting code, built upon established methods, to deblend sources and ensure consistent colours across the u-band to Ks-band wavelength range. We further detail the tight control we maintain over the point-spread function modelling required for the model fitting, astrometry and consistency of photometry between the four fields. The catalogue allows us to perform a careful star–galaxy separation and produces excellent photometric redshift performance (NMAD = 0.023 at i < 23). The Deep-Fields catalogue will be made available as part of the cosmology data products release, following the completion of the DES 3-yr weak lensing and galaxy clustering cosmology work.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Practical galaxy morphology tools from deep supervised representation learning

Astronomers have typically set out to solve supervised machine learning problems by creating their own representations from scratch. We show that deep learning models trained to answer every Galaxy Zoo DECaLS question learn meaningful semantic representations of galaxies that are useful for new tasks on which the models were never trained. We exploit these representations to outperform several recent approaches at practical tasks crucial for investigating large galaxy samples. The first task is identifying galaxies of similar morphology to a query galaxy. Given a single galaxy assigned a free text tag by humans (e.g. ‘#diffuse’), we can find galaxies matching that tag for most tags. The second task is identifying the most interesting anomalies to a particular researcher. Our approach is 100 per cent accurate at identifying the most interesting 100 anomalies (as judged by Galaxy Zoo 2 volunteers). The third task is adapting a model to solve a new task using only a small number of newly labelled galaxies. Models fine-tuned from our representation are better able to identify ring galaxies than models fine-tuned from terrestrial images (ImageNet) or trained from scratch. We solve each task with very few new labels; either one (for the similarity search) or several hundred (for anomaly detection or fine-tuning). This challenges the longstanding view that deep supervised methods require new large labelled data sets for practical use in astronomy. To help the community benefit from our pretrained models, we release our fine-tuning code zoobot. Zoobot is accessible to researchers with no prior experience in deep learning.

79 ASTRONOMY AND ASTROPHYSICS↗

Deep Learning and Likelihood Approaches for Viral Phylogeography Converge on the Same Answers Whether the Inference Model Is Right or Wrong

Abstract Analysis of phylogenetic trees has become an essential tool in epidemiology. Likelihood-based methods fit models to phylogenies to draw inferences about the phylodynamics and history of viral transmission. However, these methods are often computationally expensive, which limits the complexity and realism of phylodynamic models and makes them ill-suited for informing policy decisions in real-time during rapidly developing outbreaks. Likelihood-free methods using deep learning are pushing the boundaries of inference beyond these constraints. In this paper, we extend, compare, and contrast a recently developed deep learning method for likelihood-free inference from trees. We trained multiple deep neural networks using phylogenies from simulated outbreaks that spread among 5 locations and found they achieve close to the same levels of accuracy as Bayesian inference under the true simulation model. We compared robustness to model misspecification of a trained neural network to that of a Bayesian method. We found that both models had comparable performance, converging on similar biases. We also implemented a method of uncertainty quantification called conformalized quantile regression that we demonstrate has similar patterns of sensitivity to model misspecification as Bayesian highest posterior density (HPD) and greatly overlap with HPDs, but have lower precision (more conservative). Finally, we trained and tested a neural network against phylogeographic data from a recent study of the SARS-Cov-2 pandemic in Europe and obtained similar estimates of region-specific epidemiological parameters and the location of the common ancestor in Europe. Along with being as accurate and robust as likelihood-based methods, our trained neural networks are on average over 3 orders of magnitude faster after training. Our results support the notion that neural networks can be trained with simulated data to accurately mimic the good and bad statistical properties of the likelihood functions of generative phylogenetic models.

Evolutionary Biology↗

Some recent advances in understanding the mineralogy of Earth's deep mantle

Understanding planetary structure and evolution requires a detailed knowledge of the properties of geological materials under the conditions of deep planetary interiors. Experiments under the extreme pressure-temperature conditions of the deep mantle are challenging, and many fundamental properties remain poorly constrained or are inferred only through uncertain extrapolations from lower pressure-temperature states. Nevertheless, the last several years have witnessed a number of new developments in this area, and a broad overview of the current understanding of the Earth's lower mantle is presented here. Some recent experimental and theoretical advances related to the lowermost mantle are highlighted. Measurements of the equation of state and deformation behaviour of (Mg,Fe)SiO{sub 3} in the CaIrO{sub 3}-type (post-perovskite) structure yield insights into the nature of the core-mantle boundary region. Theoretical studies of the behaviour of MgSiO3 liquids under high pressure-temperature conditions provide constraints on melt volumes, diffusivities and viscosities that are relevant to understanding both the early Earth (e.g. deep magma oceans) and seismic structure observed in the present Earth (e.g. ultra-low-velocity zones).

58 GEOSCIENCES↗

Dark Energy Survey Year 3 results: Simulation-based 𝑤CDM inference from weak lensing and galaxy clustering maps with deep learning: Analysis design

Data-driven approaches using deep learning are emerging as powerful techniques to extract non-Gaussian information from cosmological large-scale structure. Here, this work presents the first simulation-based inference (SBI) pipeline that combines weak lensing and galaxy clustering maps in a realistic Dark Energy Survey Year 3 (DES Y3) configuration and serves as preparation for a forthcoming analysis of the survey data. We develop a scalable forward model based on the CosmoGridV1 suite of N-body simulations to generate over one million self-consistent mock realizations of DES Y3 at the map level. Leveraging this large dataset, we train deep graph convolutional neural networks on the full survey footprint in spherical geometry to learn low-dimensional features that approximately maximize mutual information with target parameters. These learned compressions enable neural density estimation of the implicit likelihood via normalizing flows in a ten-dimensional parameter space spanning cosmological 𝑤CDM, intrinsic alignment, and linear galaxy bias parameters, while marginalizing over baryonic, photometric redshift, and shear bias nuisances. To ensure robustness, we extensively validate our inference pipeline using synthetic observations derived from both systematic contaminations in our forward model and independent Buzzard galaxy catalogs. Our forecasts yield significant improvements in cosmological parameter constraints, achieving 2−3× higher figures of merit in the 𝛺 𝑚 − 𝑆 8 plane relative to our implementation of baseline two-point statistics and effectively breaking parameter degeneracies through probe combination. These results demonstrate the potential of SBI analyses powered by deep learning for upcoming Stage-IV wide-field imaging surveys.

Thomsen, A. [Zurich, ETH] (ORCID:0000000203099021)↗

Semi-inclusive deep-inelastic scattering on a polarized spin-1 target. I. Cross section and spin observables

We develop the theoretical framework for semi-inclusive deep-inelastic scattering on a polarized spin-1 target and apply it to scattering on the polarized deuteron with spectator nucleon tagging. In Part I (this article) we present the general form of the semi-inclusive cross section and polarization observables for the spin-1 target. A relativistically covariant formulation in terms of four-vectors and invariant polarization parameters is employed. The target polarization is described by a spin density matrix with vector and tensor polarization. The spin and azimuthal angle dependence of the semi-inclusive cross section is derived and parametrized in terms of invariant structure functions. To validate the result, the structure functions are expressed as photon-target helicity amplitudes with known symmetry properties. The expressions presented here are kinematic (no assumptions about particle production dynamics) and valid in all regions of the deep-inelastic final state (current and target fragmentation regions). In Part II (following article), we consider deep-inelastic scattering on the polarized deuteron with spectator nucleon tagging as a special case of target fragmentation. The semi-inclusive structure functions are computed by separating nuclear and hadronic structure, and the polarization observables are explored as functions of the tagged nucleon momentum.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Coherence-Induced Deep Thermalization Transition in Random Permutation Quantum Dynamics

We report a phase transition in the projected ensemble—the collection of postmeasurement wave functions of a local subsystem obtained by measuring its complement. The transition emerges in systems undergoing random permutation dynamics, a type of quantum time evolution wherein computational basis states are shuffled without creating superpositions. It separates a phase exhibiting deep thermalization, where the projected ensemble is distributed over Hilbert space in a maximally entropic fashion (Haar random), from a phase where it is minimally entropic (“classical bit-string ensemble”). Crucially, this deep thermalization transition is invisible to the subsystem’s density matrix, which always exhibits thermalization to infinite temperature across the phase diagram. Through a combination of analytical arguments and numerical simulations, we show that the transition is tuned by the total amount of injected by the input state and the measurement basis, and is exhibited robustly across different microscopic models. Our findings represent a novel form of ergodicity-breaking universality in quantum many-body dynamics, characterized not by a failure of regular thermalization, but rather by a failure of deep thermalization.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗