SEARCH · Search NASA
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.
Abstract—The failure of cutting tools is investigated. A generalized approach to describing failure is proposed, on the basis of the theory of random point processes. Two mechanisms by which the abrasive particles act on the tool’s cutting edge are considered. The corresponding dynamics of damage accumulation is described.
Fragmentation plays an important role in a variety of physical, chemical, and geological processes. Examples include atomization in sprays, crushing of rocks, explosion and impact of solids, polymer degradation, etc. Although each individual action of fragmentation is a complex process, the number of these elementary actions is large. It is natural to abstract a simple 'effective' scenario of fragmentation and to represent its essential features. One of the models is the fragmentation under the scaling symmetry: each breakup action reduces the typical length of fragments, r (right arrow) alpha r, by an independent random multiplier alpha (0 < alpha < 1), which is governed by the fragmentation intensity spectrum q(alpha), integral(sup 1)(sub 0) q(alpha)d alpha = 1. This scenario has been proposed by Kolmogorov (1941), when he considered the breakup of solid carbon particle. Describing the breakup as a random discrete process, Kolmogorov stated that at latest times, such a process leads to the log-normal distribution. In Gorokhovski & Saveliev, the fragmentation under the scaling symmetry has been reviewed as a continuous evolution process with new features established. The objective of this paper is twofold. First, the paper synthesizes and completes theoretical part of Gorokhovski & Saveliev. Second, the paper shows a new application of the fragmentation theory under the scale invariance. This application concerns the turbulent cascade with intermittency. We formulate here a model describing the evolution of the velocity increment distribution along the progressively decreasing length scale. The model shows that when the turbulent length scale gets smaller, the velocity increment distribution has central growing peak and develops stretched tails. The intermittency in turbulence is manifested in the same way: large fluctuations of velocity provoke highest strain in narrow (dissipative) regions of flow.
This work is devoted to the study of processes generated by random substitutions over a finite alphabet. We prove, under mild conditions on the substitution’s rule, the existence of a unique process which remains invariant under the substitution, and which exhibits a polynomial decay of correlations. For constant-length substitutions, we go further by proving that the invariant state is precisely a Gibbs measure which can be obtained as the projective limit of its natural Markovian approximations. We end up the paper by studying a class of substitutions whose invariant state is the unique Gibbs measure for a hierarchical two-body interaction.
The origin of chaotic behavior and the relation of chaos to randomness are explained. Two mathematical results are described: (1) a representation theorem guarantees the existence of a specific time-domain model for chaos and addresses the relation between chaotic, random, and strictly deterministic processes; (2) a theorem assures that information on the behavior of a physical system in its complete state space can be extracted from time-series data on a single observable. Focus is placed on an important connection between the dynamical state space and an observable time series. These two results lead to a practical deconvolution technique combining standard random process modeling methods with new embedded techniques.
Scalable electronic predictions are critical for soft materials design. Recently, the Electronic Coarse-Graining (ECG) method was introduced to renormalize all-atom quantum chemical (QC) predictions to coarse-grained (CG) resolutions using deep neural networks (DNNs). While DNNs can learn complex representations that prove challenging for kernel-based methods, they are susceptible to overfitting and the overconfidence of uncertainty estimations. Here, we develop ECG within a GPU-accelerated Deep Kernel Learning (DKL) framework to enable CG QC predictions using range-separated hybrid density functional theory (DFT), obtaining a 107 speedup relative to naive all-atom QC. By treating the predicted electronic properties as random Gaussian Processes, DKL incorporates CG mapping degeneracy by learning the distribution of electronic energies as a function of CG configuration. DKL-ECG accurately reproduces molecular orbital energies from range-separated DFT while facilitating efficient training via active learning using the uncertainties provided by DKL. Further, we show that while active learning algorithms enable efficient sampling of a more diverse configurational space relative to random sampling, all explored query methods exhibit comparable performance for the examined system. We attribute this result to the significant overlap of the feature space and output property distributions across multiple temperatures.
Analogues of 1-shuffle elements for complex reflection groups of type G(m,1,n) are introduced. A geometric interpretation for G(m,1,n) in terms of rotational permutations of polygonal cards is given. We compute the eigenvalues, and their multiplicities, of the 1-shuffle element in the algebra of the group G(m,1,n). Considering shuffling as a random walk on the group G(m,1,n), we estimate the rate of convergence to randomness of the corresponding Markov chain. We report on the spectrum of the 1-shuffle analogue in the cyclotomic Hecke algebra H(m,1,n) for m=2 and small n.
The relative abundances are treated as a consequence of processes in cosmic ray transport occurring during passage of the radiation through interstellar material at high velocity. Some of the subjects mentioned are nuclear fragmentation and the production of secondary nuclei, nuclear reactions, energy loss and nuclear decay, ionization, the range-energy relation and propagation variables, capture and loss of electrons, the propagation of nuclei, the transport equation, equilibrium solutions, energy-dependent path length distribution, exponential path length distributions, discrete spectra, sources, supernovae, and the origin of the abundances. The connection between the space-time features of the sources, the material traversed, and the effects of magnetic fields is established by describing the particle-field interaction as a diffusive or random-walk process.
A family of science enabling radiation hard Application Specific Integrated Circuits (ASICs), Front End Electronics (FEEs) and Event Processing Systems, with flight heritage on many NASA missions, is presented. These technologies play an important role in the miniaturization of instruments -and spacecraft systems- at the same time increasing performance and reducing power. The technologies target time of flight, position sensing, and energy measurements as well as standard housekeeping and telemetry functions for particle and fields instruments, but find applications in other instrument categories too. More specifically the technologies include: the TOF chip, 1D and 2D Delay Lines with MCP detectors, for high precision fast and low power time of flight and position sensing; the Energy chip for multichannel SSD readout with time over threshold and standard voltage read out for TDC and ADC digitization; Fast multi channel read out chip with commandable thresholds; the TRIO chip for multiplexed ADC and housekeeping etc. It should be mentioned that the ASICs include basic trigger capabilities to enable random event processing in a heavy background of penetrators and UV foreground. Typical instruments include time of flight versus energy and look angle particle analyzers such as: plasma composition, energetic particle, neutral atom imaging as well as fast plasma and deltaE/E ion/electron telescopes. Flight missions include: Cassini/LEMMS, IMAGE/HENA, MESSENGER/EPPS/MLA/X-ray/MLA, STEREO, PLUTO-NH/PEPSSI/LORI, IBEX-Lo, JUNO/JEDI, RBSP/RBSPICE, MMS/HPCA/EPD, SO/SIS. Given the proven capability on heavy radiation missions such as JUNO, MMS and RBSB, as well diverse long duration missions such as MESSENGER, PLUTO and Cassini, it is expected that these technologies will play an important role in the particle and fields (at least) instruments on the upcoming JUICE and JEO missions.
The paper continues the development of the rigorous supersymmetric transfer matrix approach to the random band matrices started in (J Stat Phys 164:1233–1260, 2016; Commun Math Phys 351:1009–1044, 2017). We consider random Hermitian block band matrices consisting of W×W random Gaussian blocks (parametrized by j,k∈Λ=[1,n]{sup d}∩ℤ{sup d}) with a fixed entry’s variance J{sub jk}=δ{sub j,k}W{sup −1}+βΔ{sub j,k}W{sup −2}, β>0 in each block. Taking the limit W→∞ with fixed n and β, we derive the sigma-model approximation of the second correlation function similar to Efetov’s one. Then, considering the limit β,n→∞, we prove that in the dimension d=1 the behaviour of the sigma-model approximation in the bulk of the spectrum, as β≫n, is determined by the classical Wigner–Dyson statistics.
Artificial intelligence has transformed the seismic community with deep learning models (DLMs) that are trained to complete specific tasks within workflows. However, there is still a lack of robust evaluation frameworks for evaluating and comparing DLMs. Here, we address this gap by designing an evaluation framework that jointly incorporates two crucial aspects: performance uncertainty and learning efficiency. To target these aspects, we meticulously construct the training, validation, and test splits using a clustering method tailored to seismic data and enact an expansive training design to segregate performance uncertainty arising from stochastic training processes and random data sampling. The framework’s ability to guard against misleading declarations of model superiority is demonstrated through the evaluation of PhaseNet (Zhu and Beroza, 2018), a popular seismic phase picking DLM, under three training approaches. Our framework helps practitioners choose the best model for their problem and set performance expectations by explicitly analyzing model performance with uncertainty at varying budgets of training data.
Neuromorphic computing provides a means for achieving faster and more energy efficient computations than conventional digital computers for artificial intelligence (AI). However, its current accuracy is generally less than the dominant software-based AI. The key to improving accuracy is to reduce the intrinsic randomness of memristive devices, emulating synapses in the brain for neuromorphic computing. Here using a planar device as a model system, the controlled formation of conduction channels is achieved with high oxygen vacancy concentrations through the design of sharp protrusions in the electrode gap, as observed by X-ray multimodal imaging of both oxygen stoichiometry and crystallinity. Classical molecular dynamics simulations confirm that the controlled formation of conduction channels arises from confinement of the electric field, yielding a reproducible spatial distribution of oxygen vacancies across switching cycles. Furthermore, this work demonstrates an effective route to control the otherwise random electroforming process by electrode design, facilitating the development of more accurate memristive devices for neuromorphic computing.
In classical chaotic systems the entropy, averaged over initial phase space distributions, follows a universal behavior. While approaching thermal equilibrium it passes through a stage where it grows linearly, while the growth rate, the Kolmogorov-Sinai entropy (rate), is given by the sum over all positive Lyapunov exponents. A natural question is whether a similar relation is valid for quantum systems. We argue that the Maldacena-Shenker-Stanford bound on quantum Lyapunov exponents implies that the upper bound on the growth rate of the entropy, averaged over states in Hilbert space that evolve towards a thermal state with temperature T, should be given by πT times the thermal state’s von Neumann entropy. Strongly coupled, large N theories with black hole duals should saturate the bound. To test this we study a large number of isotropization processes of random, spatially homogeneous, far from equilibrium initial states in large N, $\mathcal{N}$ = 4 Super Yang Mills theory at strong coupling and compute the ensemble averaged growth rate of the dual black hole’s apparent horizon area. We find both an analogous behavior as in classical chaotic systems and numerical evidence that the conjectured bound on averaged entropy growth is saturated granted that the Lyapunov exponents are degenerate and given by λ i = ±2πT. This fits to the behavior of classical systems with plus/minus symmetric Lyapunov spectra, a symmetry which implies the validity of Liouville’s theorem.
Traditional probabilistic methods for the simulation of advection-diffusion equations (ADEs) often overlook the entropic contribution of the discretization, e.g., the number of particles, within associated numerical methods. Many times, the gain in accuracy of a highly discretized numerical model is outweighed by its associated computational costs or the noise within the data. Herein, we address the question of how many particles are needed in a simulation to best approximate and estimate parameters in one-dimensional advective-diffusive transport. To do so, we use the well-known Akaike Information Criterion (AIC) and a recently-developed correction called the Computational Information Criterion (COMIC) to guide the model selection process. Random-walk and mass-transfer particle tracking methods are employed to solve the model equations at various levels of discretization. Numerical results demonstrate that the COMIC provides an optimal number of particles that can describe a more efficient model in terms of parameter estimation and model prediction compared to the model selected by the AIC even when the data is sparse or noisy, the sampling volume is not uniform throughout the physical domain, or the error distribution of the data is non-IID Gaussian.
The traditional method to study application resilience to errors in HPC applications uses fault injection (FI), a time-consuming approach. Furthermore, while analytical models have been built to overcome the inefficiencies of FI, they lack accuracy. In this paper, we present PARIS, a machine-learning method to predict application resilience that avoids the time-consuming process of random FI and provides higher prediction accuracy than analytical models. PARIS captures the implicit relationship between application characteristics and application resilience, which is difficult to capture using most analytical models. We overcome many technical challenges for feature construction, extraction, and selection to use machine learning in our prediction approach. Our evaluation on 16 HPC benchmarks shows that PARIS achieves high prediction accuracy. PARIS is up to 450x faster than random FI (49x on average). Compared to the state-of-the-art analytical model, PARIS is at least 63% better in terms of accuracy and has comparable execution time on average.
We use artificial intelligence (AI) to learn and infer the physics of higher order gravitational wave modes of quasi-circular, spinning, non precessing binary black hole mergers. We trained AI models using 14 million waveforms, produced with the surrogate model NRHybSur3dq8, that include modes up to $\ell$ ≤ 4 and (5,5), except for (4,0) and (4,1), that describe binaries with mass-ratios $\textit{q}$ ≤ 8, individual spins $s^z_{\{1,2\}} \in$[–0.8,0.8], and inclination angle $θ \in$ [0,π]. Our probabilistic AI surrogates can accurately constrain the mass-ratio, individual spins, effective spin, and inclination angle of numerical relativity waveforms that describe such signal manifold. We compared the predictions of our AI models with Gaussian process regression, random forest, k-nearest neighbors, and linear regression, and with traditional Bayesian inference methods through the PyCBC Inference toolkit, finding that AI outperforms all these approaches in terms of accuracy, and are between three to four orders of magnitude faster than traditional Bayesian inference methods. Our AI surrogates were trained within 3.4 hours using distributed training on 1,536 NVIDIA V100 GPUs in the Summit supercomputer.
Although advances in synthesizing hierarchical semiconductor materials, there have been few studies on the fundamental nucleation mechanisms to explain the origins of such complex structures. Resolving these nucleation and growth pathways is very difficult and time consuming, but is critical for developing predictive synthetic capabilities for the synthesis and applications of new materials. In this paper, we use state-of-the-art in-situ liquid phase scanning electron microscopy (SEM) and high-resolution transmission electron microscopy in combination of classical Density Functional Theory (cDFT) to study the nucleation of highly-branched wurtzite ZnO nanostructures via a facile, room temperature aqueous synthesis route. Using a range of precursor concentrations, we systematically vary their hierarchical organization. In situ liquid phase SEM demonstrates that all branches form through secondary nucleation and grow by classical processes. Neither random aggregation nor oriented attachment is observed. cDFT results imply the morphological evolution with increasing [Zn 2+ ] arises from an interplay between rising thermodynamic driving force, which promotes branch number and variability of orientation, and increasing barriers to interfacial transport due to ion correlation forces that alter the anisotropic kinetics of growth. Finally, these findings provide a quantitative picture of branching that sets to rest past controversies and advances efforts to decipher growth mechanisms of hierarchical structures in real solution environments.
Evapotranspiration (ET) is a key component of the atmospheric and terrestrial water and energy budgets. Satellite-based vegetation index approaches have used remotely-sensed vegetation and reanalysis meteorological properties with surface energy balance models to estimate global ET (MOD16 ET). We reconstructed satellite retrievals using in situ meteorology (Argonne-ET) and evaluated them using a dense network of surface turbulent heat flux measurements. Argonne-ET resolves spatial heterogeneity of ET across the U.S. Southern Great Plains that is not well characterized by MOD16; MOD16 ET exhibits spatial autocorrelation (Moran's I (MI) significantly > 0 in May – Aug.), whereas in situ and Argonne-ET exhibit characteristics of a random spatial process (MI not significantly different from 0). The skill in resolving ET temporal variability is not significantly different between MOD16 and Argonne-ET (correlation coefficient = 0.75 and 0.72, respectively). However, the root mean square errors were significantly lower for Argonne-ET (36 W m -2 ) than MOD16 (43 W m -2 ), and MOD16 exhibits substantial bias in annual ET relative to in situ measurements (-38%). This is attributed to overestimation in the dry canopy surface resistance (rs) parameterization. Using rs constrained to the range of typical measured values, Argonne-ET substantially reduces the bias in the annual ET (+1%). The improved ET estimates are critical for regional water budget analyses. The methodology presented herein also demonstrates the ability to retrieve high temporally resolved (30-min; cf. 8-day MOD16) ET that can be used for development of processed-based diagnostics of model biases, and to elucidate avenues to improve ET model parameterizations.