Search NASA⌕ Search

SEARCH · Search NASA

Results for “Randomization”

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 703 records · Page 39

Characteristics of flow through randomly packed impermeable and permeable particles using pore resolved simulations

Pore resolved simulations are performed to study the mean flow characteristics in porous media formed by random distribution of impermeable and permeable particles. Here, permeability of the medium and reactive surface areas are analyzed for a range of solid fractions, packing patterns and flow rates. The results are compared with models presented in the literature and with simulations of flow through ordered packings. Permeability of a randomly packed porous medium is found to depend on the heterogeneous distribution of porosity which results from particle agglomeration and is typically ignored by simpler models. The reactive surface area is dependent on the flow field at pore scale in addition to geometry. Moreover, the wake formed downstream of the particles at moderately high flow rates also affects the reactive surface area. Similar analysis is also performed by assuming particles to be permeable to account for dual scale porosity relevant for various engineering applications.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Online randomized interpolative decomposition with a posteriori error estimator for temporal PDE data reduction

Traditional low-rank approximation is a powerful tool for compressing large data matrices that arise in simulations of partial differential equations (PDEs), but suffers from high computational cost and requires several passes over the PDE data. The compressed data may also lack interpretability thus making it difficult to identify feature patterns from the original data. Here, to address these issues, we present an online randomized algorithm to compute the interpolative decomposition (ID) of large-scale data matrices in situ. Compared to previous randomized IDs that used the QR decomposition to determine the column basis, we adopt a streaming ridge leverage score-based column subset selection algorithm that dynamically selects proper basis columns from the data and thus avoids an extra pass over the data to compute the coefficient matrix of the ID. In particular, we adopt a single-pass error estimator based on the non-adaptive Hutch++ algorithm to provide real-time error approximation for determining the best coefficients. As a result, our approach only needs a single pass over the original data and thus is suitable for large and high-dimensional matrices stored outside of core memory or generated in PDE simulations. A strategy to improve the accuracy of the reconstructed data gradient, when desired, within the ID framework is also presented. We provide numerical experiments on turbulent channel flow and ignition simulations, and on the NSTX Gas Puff Image dataset, comparing our algorithm with the offline ID algorithm to demonstrate its utility in real-world applications.

Column subset selection↗

Data-assisted combustion simulations with dynamic submodel assignment using random forests

This investigation outlines a data-assisted approach that employs random forest classifiers for local and dynamic submodel assignment in turbulent-combustion simulations. This method is demonstrated in simulations of a single-element GOX/GCH4 rocket combustor; a priori as well as a posteriori assessments are conducted to (i) evaluate the accuracy and adjustability of the classifier for targeting different quantities of interest (QoIs), and (ii) assess improvements, resulting from the data-assisted combustion model assignment, in predicting target QoIs during simulation runtime. Results from the a priori study show that random forests, trained with local flow properties as input variables and combustion model errors as training labels, assign three different combustion models – finite-rate chemistry (FRC), flamelet progress variable (FPV) model, and inert mixing (IM) – with reasonable classification performance even when targeting multiple QoIs. Applications in a posteriori studies demonstrate improved predictions from data-assisted simulations, in temperature and CO mass fraction, when compared with monolithic FPV calculations. An additional a posteriori data-assisted simulation of a modified configuration demonstrates that the present approach can be successfully applied to different configurations, as long as thermophysical behavior can be represented by the training data. Furthermore, these results demonstrate that this data-driven framework holds promise for dynamic combustion submodel assignments in reacting flow simulations.

42 ENGINEERING↗

A finite difference informed random walker (FDiRW) solver for strongly inhomogeneous diffusion problems

In nature, many complex multi-physics coupling problems exhibit strong diffusivity inhomogeneity. For instance, in the context of radionuclide absorption by porous wasteform materials within a flowing waste stream, the difference of species’ diffusivity in solid and liquid phases spans by 3~8 orders of magnitude. To solve the diffusion equations with strongly inhomogeneous diffusivity, traditional discretization-based methods, such as the Finite Difference Method (FDM), require infinitesimally small time steps (<10 -10 ) as high spatial resolutions are employed in most microstructure evolution processes, leading to prohibitively high computational costs. Here, this work developed an integrated numerical approach (FDiRW: Finite Difference informed Random Walk) to tackle this challenge. The idea is that utilizing the Random Walk concept, the fast diffusion is modeled as a superposition of point source’s solution for a concentration distribution while FDM is used to obtain the point source’s solution at each node. A mesh-coarsening algorithm is developed to generate an exclusive coarse mesh for FDiRW approach to maximize its efficiency. The effectiveness of the coarse mesh-based FDiRW approach is validated by benchmarking Finite Difference solutions. Numerical results demonstrated that FDiRW achieves a remarkable 1000x computational efficiency improvement over FDM while preserving desired accuracy for a medium-sized model of 192 × 192 × 192 grids. Finally, as models scale up, a floating-point operations (PLOPs) analysis of the FDiRW algorithm reveals that its computational complexity grows quadratically in terms of the number of nodes employed in computation.

36 MATERIALS SCIENCE↗

Incorporating space and time into random forest models for analyzing geospatial patterns of drug-related crime incidents in a major U.S. metropolitan area

The opioid crisis has hit American cities hard, and research on spatial and temporal patterns of drug-related activities including detecting and predicting clusters of crime incidents involving particular types of drugs is useful for distinguishing hot zones where drugs are present that in turn can further provide a basis for assessing and providing related treatment services. In this study, we investigated spatiotemporal patterns of more than 52,000 reported incidents of drug-related crime at block group granularity in Chicago, IL between 2016 and 2019. We applied a space-time analysis framework and machine learning approaches to build a model using training data that identified whether certain locations and built environment and sociodemographic factors were correlated with drug-related crime incident patterns, and establish the top contributing factors that underlaid the trends. Space and time, together with multiple driving factors, were incorporated into a random forest model to analyze these changing patterns. We accommodated both spatial and temporal autocorrelation in the model learning process to assist with capturing the changes over time and tested the capabilities of the space-time random forest model by predicting drug-related activity hot zones. Overall, we focused particularly on crime incidents that involved heroin and synthetic drugs as these have been key drug types that have highly impacted cities during the opioid crisis in the U.S.

97 MATHEMATICS AND COMPUTING↗

Randomized Trial of Hyperfractionation Versus Conventional Fractionation in T2 Squamous Cell Carcinoma of the Vocal Cord (RTOG 9512)

Purpose: To compare hyperfractionation versus standard fractionation for T2N0 vocal cord carcinoma in a randomized controlled trial. Methods and Materials: Patients with T2 vocal cord cancer were stratified by substage (T2a vs T2b) and randomly assigned to receive either hyperfractionation (HFX) to 79.2 Gy in 66 fractions of 1.2 Gy given twice a day, or standard fractionation (SFX) to 70 Gy in 35 fractions given once a day. The trial was designed to detect a 55% reduction in the local failure hazard rate with 80% statistical power. Results: Between April 1996 and July 2003, a total of 250 patients were enrolled. Of 239 patients analyzable for outcomes, 94% were male, 83% had a Karnofsky performance status of 90-100, and 62% had T2a tumor. Median follow-up for all surviving patients was 7.9 years (range, 0.6-13.1 years). The 5-year local control (LC) rate was 8 points higher but not statistically significant (P=.14 for HFX [78%] vs SFX [70%]), corresponding to a 30% hazard rate reduction. The 5-year disease-free survival (DFS) was 49% versus 40% (P=.13) and overall survival (OS) was 72% versus 63% (P=.29). HFX was associated with higher rates of acute skin, mucosal, and laryngeal toxicity. Grade 3-4 late effects were similar with a 5-year cumulative incidence of 8.5% (3.4%-13.6%) after SFX and 8.5% (3.4%-13.5%) after HFX. Conclusions: The 5-year local control was modestly higher with HFX compared to SFX for T2 glottic carcinoma, but the difference was not statistically significant. These results are consistent with prior studies of hyperfractionation showing a benefit in local control. Substaging by T2a versus T2b carries prognostic value for DFS and OS. For cost and convenience reasons other altered fractionation schedules have been adopted in routine practice.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Hyperfractionated Accelerated Radiation Therapy (HART) of 70.6 Gy With Concurrent 5-FU/Mitomycin C Is Superior to HART of 77.6 Gy Alone in Locally Advanced Head and Neck Cancer: Long-term Results of the ARO 95-06 Randomized Phase III Trial

Purpose: To report the long-term results of the ARO 95-06 randomized trial comparing hyperfractionated accelerated chemoradiation with mitomycin C/5-fluorouracil (C-HART) with hyperfractionated accelerated radiation therapy (HART) alone in locally advanced head and neck cancer. Patients and Methods: The primary endpoint was locoregional control (LRC). Three hundred eighty-four patients with stage III (6%) and IV (94%) oropharyngeal (59.4%), hypopharyngeal (32.3%), and oral cavity (8.3%) cancer were randomly assigned to 30 Gy/2 Gy daily followed by twice-daily 1.4 Gy to a total of 70.6 Gy concurrently with mitomycin C/5-FU (C-HART) or 16 Gy/2 Gy daily followed by twice-daily 1.4 Gy to a total dose of 77.6 Gy alone (HART). Statistical analyses were done with the log-rank test and univariate and multivariate Cox regression analyses. Results: The median follow-up time was 8.7 years (95% confidence interval [CI]: 7.8-9.7 years). At 10 years, the LRC rates were 38.0% (C-HART) versus 26.0% (HART, P=.002). The cancer-specific survival and overall survival rates were 39% and 10% (C-HART) versus 30.0% and 9% (HART, P=.042 and P=.049), respectively. According to multivariate Cox regression analysis, the combined treatment was associated with improved LRC (hazard ratio [HR]: 0.6 [95% CI: 0.5-0.8; P=.002]). The association between combined treatment arm and increased LRC appeared to be limited to oropharyngeal cancer (P=.003) as compared with hypopharyngeal or oral cavity cancer (P=.264). Conclusions: C-HART remains superior to HART in terms of LRC. However, this effect may be limited to oropharyngeal cancer patients.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Random walk particle tracking simulations of non-Fickian transport in heterogeneous media

Derivations of continuum nonlocal models of non-Fickian (anomalous) transport require assumptions that might limit their applicability. We present a particle-based algorithm, which obviates the need for many of these assumptions by allowing stochastic processes that represent spatial and temporal random increments to be correlated in space and time, be stationary or non-stationary, and to have arbitrary distributions. The approach treats a particle trajectory as a subordinated stochastic process that is described by a set of Langevin equations, which represent a continuous time random walk (CTRW). Convolution-based particle tracking (CBPT) is used to increase the computational efficiency and accuracy of these particle-based simulations. The combined CTRW-CBPT approach enables one to convert any particle tracking legacy code into a simulator capable of handling non-Fickian transport.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Note on coefficient matrices from stochastic Galerkin methods for random diffusion equations

In a recent work by Xiu and Shen [D. Xiu, J. Shen, Efficient stochastic Galerkin methods for random diffusion equations, J. Comput. Phys. 228 (2009) 266-281], the Galerkin methods are used to solve stochastic diffusion equations in random media, where some properties for the coefficient matrix of the resulting system are provided. They also posed an open question on the properties of the coefficient matrix. In this work, we will provide some results related to the open question.

97 MATHEMATICS AND COMPUTING↗

A partially reflecting random walk on spheres algorithm for electrical impedance tomography

In this work, we develop a probabilistic estimator for the voltage-to-current map arising in electrical impedance tomography. This novel so-called partially reflecting random walk on spheres estimator enables Monte Carlo methods to compute the voltage-to-current map in an embarrassingly parallel manner, which is an important issue with regard to the corresponding inverse problem. Our method uses the well-known random walk on spheres algorithm inside subdomains where the diffusion coefficient is constant and employs replacement techniques motivated by finite difference discretization to deal with both mixed boundary conditions and interface transmission conditions. We analyze the global bias and the variance of the new estimator both theoretically and experimentally. Subsequently, the variance of the new estimator is considerably reduced via a novel control variate conditional sampling technique which yields a highly efficient hybrid forward solver coupling probabilistic and deterministic algorithms.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

A Bloch decomposition-based stochastic Galerkin method for quantum dynamics with a random external potential

In this paper, we consider the numerical solution of the one-dimensional Schrödinger equation with a periodic lattice potential and a random external potential. This is an important model in solid state physics where the randomness results from complicated phenomena that are not exactly known. Here we generalize the Bloch decomposition-based time-splitting pseudospectral method to the stochastic setting using the generalized polynomial chaos with a Galerkin procedure so that the main effects of dispersion and periodic potential are still computed together. We prove that our method is unconditionally stable and numerical examples show that it has other nice properties and is more efficient than the traditional method. Finally, we give some numerical evidence for the well-known phenomenon of Anderson localization.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

RandONets: Shallow networks with random projections for learning linear and nonlinear operators

Deep neural networks have been extensively used for the solution of both the forward and the inverse problem for dynamical systems. However, their implementation necessitates optimizing a high-dimensional space of parameters and hyperparameters. This fact, along with the requirement of substantial computational resources, pose a barrier to achieving high numerical accuracy, but also interpretability. Here, to address the above challenges, we present Random Projection-based Operator Networks (RandONets): shallow networks with random projections and tailor-made numerical analysis methods that learn accurately and fast linear and nonlinear operators. Building on previous works, we prove that RandOnets are universal approximators of linear and nonlinear operators. Due to their simplicity, RandONets provide a one-step transformation of the input space, facilitating interpretability. For the evaluation of their performance, we focus on operators of PDEs. We show, that RandONets outperform by several orders of magnitude, both in terms of numerical approximation accuracy and computational cost, the “vanilla” DeepONets. Hence, we believe that our method will trigger further developments in the field of scientific machine learning, for the development of new ‘’light”schemes that will provide high accuracy while reducing dramatically the computational cost. A MATLAB toolbox for RandONets, including demos, is available on GitHub at https://github.com/GianlucaFabiani/RandONets.

Interpretable machine learning↗

Relationships between nonlinear normal modes and response to random inputs

The ability to model nonlinear structures subject to random excitation is of key importance in designing hypersonic aircraft and other advanced aerospace vehicles. When a structure is linear, superposition can be used to construct its response to a known spectrum in terms of its linear modes. Superposition does not hold for a nonlinear system, but several works have shown that a system's dynamics can still be understood qualitatively in terms of its nonlinear normal modes (NNMs). Here, this work investigates the connection between a structure's undamped nonlinear normal modes and the spectrum of its response to high amplitude random forcing. Two examples are investigated: a spring-mass system and a clamped-clamped beam modeled within a geometrically nonlinear finite element package. In both cases, an intimate connection is observed between the smeared peaks in the response spectrum and the frequency-energy dependence of the nonlinear normal modes. In order to understand the role of coupling between the underlying linear modes, reduced order models with and without modal coupling terms are used to separate the effect of each NNM's backbone from the nonlinear couplings that give rise to internal resonances. In the cases shown here, uncoupled, single-degree-of-freedom nonlinear models are found to predict major features in the response with reasonable accuracy; a highly inexpensive approximation such as this could be useful in design and optimization studies. More importantly, the results show that a reduced order model can be expected to give accurate results only if it is also capable of accurately predicting the frequency-energy dependence of the nonlinear modes that are excited.

42 ENGINEERING↗

Entanglement of random subspaces via the Hastings bound

Recently, Hastings ['A counterexample to additivity of minimum output entropy', Nat. Phys. 5, 255 (2009); e-print arXiv:0809.3972v3] proved the existence of random unitary channels, which violate the additivity conjecture. In this paper, we use Hastings' method to derive new bounds for the entanglement of random subspaces of bipartite systems. As an application we use these bounds to prove the existence of nonunital channels, which violate additivity of minimal output entropy.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Quantum correlations and dynamics from classical random fields valued in complex Hilbert spaces

One of the crucial differences between mathematical models of classical and quantum mechanics (QM) is the use of the tensor product of the state spaces of subsystems as the state space of the corresponding composite system. (To describe an ensemble of classical composite systems, one uses random variables taking values in the Cartesian product of the state spaces of subsystems.) We show that, nevertheless, it is possible to establish a natural correspondence between the classical and the quantum probabilistic descriptions of composite systems. Quantum averages for composite systems (including entangled) can be represented as averages with respect to classical random fields. It is essentially what Albert Einstein dreamed of. QM is represented as classical statistical mechanics with infinite-dimensional phase space. While the mathematical construction is completely rigorous, its physical interpretation is a complicated problem. We present the basic physical interpretation of prequantum classical statistical field theory in Sec. II. However, this is only the first step toward real physical theory.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Interpolation between Airy and Poisson statistics for unitary chiral non-Hermitian random matrix ensembles

We consider a family of chiral non-Hermitian Gaussian random matrices in the unitarily invariant symmetry class. The eigenvalue distribution in this model is expressed in terms of Laguerre polynomials in the complex plane. These are orthogonal with respect to a non-Gaussian weight including a modified Bessel function of the second kind, and we give an elementary proof for this. In the large n limit, the eigenvalue statistics at the spectral edge close to the real axis are described by the same family of kernels interpolating between Airy and Poisson that was recently found by one of the authors for the elliptic Ginibre ensemble. We conclude that this scaling limit is universal, appearing for two different non-Hermitian random matrix ensembles with unitary symmetry. As a second result we give an equivalent form for the interpolating Airy kernel in terms of a single real integral, similar to representations for the asymptotic kernel in the bulk and at the hard edge of the spectrum. This makes its structure as a one-parameter deformation of the Airy kernel more transparent.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Conductance statistics for the power-law banded random matrix model

We study numerically the conductance statistics of the one-dimensional (1D) Anderson model with random long-range hoppings described by the Power-law Banded Random Matrix (PBRM) model. Within a scattering approach to electronic transport, we consider two scattering setups in absence and presence of direct processes: 2M single-mode leads attached to one side and to opposite sides of 1D circular samples. For both setups we show that (i) the probability distribution of the logarithm of the conductance T behaves as w(lnT){proportional_to}T{sup M2/2}, for T<<<T{sub typ} = exp<lnT>, for both the critical and the non-critical samples; and (ii) at criticality there is a smooth crossover from localized-like to delocalized-like behavior in the transport properties of the PBRM model by decreasing the fractality of its eigenstates.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

In situ observation of nickel as an oxidizable electrode material for the solid-electrolyte-based resistive random access memory

In this letter, we dynamically investigate the resistive switching characteristics and physical mechanism of the Ni/ZrO{sub 2}/Pt device. The device shows stable bipolar resistive switching behaviors after forming process, which is similar to the Ag/ZrO{sub 2}/Pt and Cu/ZrO{sub 2}/Pt devices. Using in situ transmission electron microscopy, we observe in real time that several conductive filaments are formed across the ZrO{sub 2} layer between Ni and Pt electrodes after forming. Energy-dispersive X-ray spectroscopy results confirm that Ni is the main composition of the conductive filaments. The ON-state resistance increases with increasing temperature, exhibiting the feature of metallic conduction. In addition, the calculated resistance temperature coefficient is equal to that of the 10-30 nm diameter Ni nanowire, further indicating that the nanoscale Ni conductive bridge is the physical origin of the observed conductive filaments. The resistive switching characteristics and the conductive filament's component of Ni/ZrO{sub 2}/Pt device are consistent with the characteristics of the typical solid-electrolyte-based resistive random access memory. Therefore, aside from Cu and Ag, Ni can also be used as an oxidizable electrode material for resistive random access memory applications.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗