SEARCH · Engineering Papers

Results for “stochastic gradient”

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.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 55 records · Page 3

All sources NASA NTRS DOE OSTI

NASA NTRS1987

Stochastic and hybrid-stress plate/shell finite elements for hot-section components

The research effort in the Center for the Advancement of Computational Mechanics at Georgia Tech has two main thrusts. The first of these is the development of special approaches for the numerical stress analysis of solids and structures whose material and geometric properties are uncertain. The second seeks to develop and implement high-efficiency plate and shell elements. The stochastic element method, currently being implemented, will be able to more accurately portray the probabilistic nature of stress, strain, and displacement in actual structures. Current research has provided a hybrid-stress shell element whose behavior is acceptable for aspect ratios as high as 30 to 1. Thus, substantially more complex analyses will be practicable as soon as this element is fully implemented. An additional advantage of the hybrid approach is that it permits more accurate stress-recovery at the upper and lower surfaces of the shell, an important consideration in high thickness-gradient applications. The software associated with the above research is being implemented in the form of extensions to the Nessus code. The hybrid shell element has been successfully tested in several small-deformation elastic analyses. The theoretical formulation of the stochastic elements is essentially complete; its implementation is just beginning.

Atluri, S. N.↗

NASA NTRS2019

A Communication Channel Density Estimating Generative Adversarial Network

Autoencoder-based communication systems use neural network channel models to backwardly propagate message reconstruction error gradients across an approximation of the physical communication channel. In this work, we develop and test a new generative adversarial network (GAN) architecture for the purpose of training a stochastic channel approximating neural network. In previous research, investigators have focused on additive white Gaussian noise (AWGN) channels and/or simplified Rayleigh fading channels, both of which are linear and have well defined analytic solutions. Given that training a neural network is computationally expensive, channel approximation networks— and more generally the autoencoder systems—should be evaluated in communication environments that are traditionally difficult. To that end, our investigation focuses on channels that contain a combination of non-linear amplifier distortion, pulse shape filtering, intersymbol interference, frequency-dependent group delay, multipath, and non-Gaussian statistics. Each of our models are trained without any prior knowledge of the channel. We show that the trained models have learned to generalize over an arbitrary amplifier drive level and constellation alphabet. We demonstrate the versatility of our GAN architecture by comparing the marginal probability density function of several channel simulations with that of their corresponding neural network approximations

Smith, Aaron↗

NASA NTRS1974

Parameter estimation for terrain modeling from gradient data

This paper developes a method for mathematically modeling terrain surfaces for use on an unmanned Martian vehicle. The data collected by the vehicle consists of terrain height and two directional slopes at each data point. The parameters for the mathematical terrain model are stochastically determined by using least square approximations.-

Shen, C. N.↗

NASA NTRS1974

Cosmic ray scintillations. II - General theory of interplanetary scintillations

The motion of charged particles in a stochastic magnetic field with nonzero mean is considered via a generalized quasi-linear expansion of Liouville's equation. The general result is an equation relating cosmic ray scintillations to magnetic fluctuations and to cosmic ray gradients. The resonant interaction between particles and the random magnetic field is considered in detail, and the effect of nonlinear terms in the equations is considered. The nonlinear terms are important in damping out initial conditions and in determining conditions near cyclotron resonances. The application of the theory to the propagation of cosmic rays during quiet times in interplanetary space is considered. It is concluded that cosmic ray scintillations in interplanetary space may provide useful information about interplanetary particles and fields and also about nonlinear plasma interactions.

Owens, A. J.↗

NASA NTRS1976

A kinetic-theory approach to turbulent chemically reacting flows

The paper examines the mathematical and physical foundations for the kinetic theory of reactive turbulent flows, discussing the differences and relation between the kinetic and averaged equations, and comparing some solutions of the kinetic equations obtained by the Green's function method with those obtained by the approximate bimodal method. The kinetic method described consists essentially in constructing the probability density functions of the chemical species on the basis of solutions of the Langevin stochastic equation for the influence of eddies on the behavior of fluid elements. When the kinetic equations are solved for the structure of the diffusion flame established in a shear layer by the bimodal method, discontinuities in gradients of the mean concentrations at the two flame edges appear. This is a consequence of the bimodal approximation of all distribution functions by two dissimilar half-Maxwellian functions, which is a very crude approximation. These discontinuities do not appear when the solutions are constructed by the Green's function method described here.

Chung, P. M.↗

NASA NTRS2012

Upper-Ocean Heat Balance Processes and the Walker Circulation in CMIP5 Model Projections

Considerable uncertainty remains as to the importance of mechanisms governing decadal and longer variability of the Walker Circulation, its connection to the tropical climate system, and prospects for tropical climate change in the face of anthropogenic forcing. Most contemporary climate models suggest that in response to elevated CO2 and a warmer but more stratified atmosphere, the required upward mass flux in tropical convection will diminish along with the Walker component of the tropical mean circulation as well. Alternatively, there is also evidence to suggest that the shoaling and increased vertical stratification of the thermocline in the eastern Pacific will enable a muted SST increase there-- preserving or even enhancing some of the dynamical forcing for the Walker cell flow. Over the past decade there have been observational indications of an acceleration in near-surface easterlies, a strengthened Pacific zonal SST gradient, and globally-teleconnected dislocations in precipitation. But is this evidence in support of an ocean dynamical thermostat process posited to accompany anthropogenic forcing, or just residual decadal fluctuations associated with variations in warm and cold ENSO events and other stochastic forcing? From a modeling perspective we try to make headway on this question by examining zonal variations in surface energy fluxes and dynamics governing tropical upper ocean heat content evolution in the WCRP CMIP5 model projections. There is some diversity among model simulations; for example, the CCSM4 indicates net ocean warming over the IndoPacific region while the CSIRO model concentrates separate warming responses over the central Pacific and Indian Ocean regions. The models, as with observations, demonstrate strong local coupling between variations in column water vapor, downward surface longwave radiation and SST; but the spatial patterns of changes in the sign of this relationship differ among models and, for models as a whole, with observations. Our analysis focuses initially on probing the inter-model differences in energy fluxes / transports and Walker Circulation response to forcing. We then attempt to identify statistically the El Nino- / La Nina-related ocean heat content variability unique to each model and regress out the associated energy flux, ocean heat transport and Walker response on these shorter time scales for comparison to that of the anthropogenic signals.

Robertson, F. R.↗

NASA NTRS1975

Computation of output feedback gains for linear stochastic systems using the Zangnill-Powell Method

Because conventional optimal linear regulator theory results in a controller which requires the capability of measuring and/or estimating the entire state vector, it is of interest to consider procedures for computing controls which are restricted to be linear feedback functions of a lower dimensional output vector and which take into account the presence of measurement noise and process uncertainty. To this effect a stochastic linear model has been developed that accounts for process parameter and initial uncertainty, measurement noise, and a restricted number of measurable outputs. Optimization with respect to the corresponding output feedback gains was then performed for both finite and infinite time performance indices without gradient computation by using Zangwill's modification of a procedure originally proposed by Powell. Results using a seventh order process show the proposed procedures to be very effective.

Kaufman, H.↗

NASA NTRS1977

Computation of output feedback gains for linear stochastic systems using the Zangwill-Powell method

Kaufman, H.↗

NASA NTRS1976

Reduced state feedback gain computation

Because application of conventional optimal linear regulator theory to flight controller design requires the capability of measuring and/or estimating the entire state vector, it is of interest to consider procedures for computing controls which are restricted to be linear feedback functions of a lower dimensional output vector and which take into account the presence of measurement noise and process uncertainty. Therefore, a stochastic linear model that was developed is presented which accounts for aircraft parameter and initial uncertainty, measurement noise, turbulence, pilot command and a restricted number of measurable outputs. Optimization with respect to the corresponding output feedback gains was performed for both finite and infinite time performance indices without gradient computation by using Zangwill's modification of a procedure originally proposed by Powell. Results using a seventh order process show the proposed procedures to be very effective.

Kaufman, H.↗

NASA NTRSDate not supplied

Uncertainty Quantification using Deep Ensembles for Decision Making in Cyber-Physical-Human Systems

In this paper and its companion, Differential Equation Approximation Using Gradient-Boosted Quantile Regression, Robison et al., we examine an approach to quantifying model uncertainty with the aim of increasing the trustworthiness of computational models in human-machine interactions. In Differential Equation Approximation Using Gradient-Boosted Quantile Regression, we focus on gradient-boosted decision trees, while in this one, we give more details about deep ensembles. Uncertainty quantification is crucial for building trustworthy autonomous decision-making agents in human-machine teams. There are two types of uncertainties: aleatoric and epistemic. The former is related to the inherent stochasticity (noise) of the process, whereas the latter is associated with the lack of knowledge or representation capability of models, such as neural networks. By lack of knowledge, we mean the model’s inability to accurately predict outputs for all possible inputs. The aleatory uncertainty can be estimated fairly easily with, for example, filters, whereas epistemic uncertainty is challenging to compute. This paper uses deep ensembles to quantify both aleatory and epistemic uncertainty. It can act as an uncertainty-aware surrogate transition model for decision-making frameworks. "Uncertainty-aware" means that the surrogate transition model should make predictions along with confidence in those predictions. In the context of decision-making, the transition models are ordinary differential equations (ODEs). Since ODEs can be simulated to make one-step or multi-step predictions, a good surrogate model for them should perform reasonably well in both modes. In a multi-step approach, the trajectory sampling method TS∞ was used to propagate uncertainty over multiple steps. The cartpole dynamical system was selected to demonstrate the ability of deep ensembles as good surrogate transition models for decision-making frameworks. The deep ensembles modeled the dynamics of cartpole ODEs and made uncertainty-aware predictions in single-step and multi-step transition modes.

CPH systems↗

NASA NTRS1992

Stochastic modeling of turbulent reacting flows

Direct numerical simulations of a single-step irreversible chemical reaction with non-premixed reactants in forced isotropic turbulence at R(sub lambda) = 63, Da = 4.0, and Sc = 0.7 were made using 128 Fourier modes to obtain joint probability density functions (pdfs) and other statistical information to parameterize and test a Fokker-Planck turbulent mixing model. Preliminary results indicate that the modeled gradient stretching term for an inert scalar is independent of the initial conditions of the scalar field. The conditional pdf of scalar gradient magnitudes is found to be a function of the scalar until the reaction is largely completed. Alignment of concentration gradients with local strain rate and other features of the flow were also investigated.

Fox, R. O.↗

NASA NTRS1993

Feedback control for unsteady flow and its application to the stochastic Burgers equation

The study applies mathematical methods of control theory to the problem of control of fluid flow with the long-range objective of developing effective methods for the control of turbulent flows. Model problems are employed through the formalism and language of control theory to present the procedure of how to cast the problem of controlling turbulence into a problem in optimal control theory. Methods of calculus of variations through the adjoint state and gradient algorithms are used to present a suboptimal control and feedback procedure for stationary and time-dependent problems. Two types of controls are investigated: distributed and boundary controls. Several cases of both controls are numerically simulated to investigate the performances of the control algorithm. Most cases considered show significant reductions of the costs to be minimized. The dependence of the control algorithm on the time-descretization method is discussed.

Choi, Haecheon↗

NASA NTRS2007

Scope of Gradient and Genetic Algorithms in Multivariable Function Optimization

Global optimization of a multivariable function - constrained by bounds specified on each variable and also unconstrained - is an important problem with several real world applications. Deterministic methods such as the gradient algorithms as well as the randomized methods such as the genetic algorithms may be employed to solve these problems. In fact, there are optimization problems where a genetic algorithm/an evolutionary approach is preferable at least from the quality (accuracy) of the results point of view. From cost (complexity) point of view, both gradient and genetic approaches are usually polynomial-time; there are no serious differences in this regard, i.e., the computational complexity point of view. However, for certain types of problems, such as those with unacceptably erroneous numerical partial derivatives and those with physically amplified analytical partial derivatives whose numerical evaluation involves undesirable errors and/or is messy, a genetic (stochastic) approach should be a better choice. We have presented here the pros and cons of both the approaches so that the concerned reader/user can decide which approach is most suited for the problem at hand. Also for the function which is known in a tabular form, instead of an analytical form, as is often the case in an experimental environment, we attempt to provide an insight into the approaches focusing our attention toward accuracy. Such an insight will help one to decide which method, out of several available methods, should be employed to obtain the best (least error) output. *

Shaykhian, Gholam Ali↗

NASA NTRS1982

Random element method for numerical modeling of diffusional processes

The random element method is a generalization of the random vortex method that was developed for the numerical modeling of momentum transport processes as expressed in terms of the Navier-Stokes equations. The method is based on the concept that random walk, as exemplified by Brownian motion, is the stochastic manifestation of diffusional processes. The algorithm based on this method is grid-free and does not require the diffusion equation to be discritized over a mesh, it is thus devoid of numerical diffusion associated with finite difference methods. Moreover, the algorithm is self-adaptive in space and explicit in time, resulting in an improved numerical resolution of gradients as well as a simple and efficient computational procedure. The method is applied here to an assortment of problems of diffusion of momentum and energy in one-dimension as well as heat conduction in two-dimensions in order to assess its validity and accuracy. The numerical solutions obtained are found to be in good agreement with exact solution except for a statistical error introduced by using a finite number of elements, the error can be reduced by increasing the number of elements or by using ensemble averaging over a number of solutions.

Ghoniem, A. F.↗

NASA NTRS1985

Structure of ducted particle-laden turbulent jets

New and existing measurements have been used to develop and evaluate models of particle-laden jets and sprays, yielding encouraging results for a stochastic separated flow (SSF) model which treats effects of interphase slip and turbulence on the basis of random walk computations of particle or drop motion. An evaluation of the SSF model using the measurements of Modarress et al. (1982, 1983) yielded results comparable to earlier work, suggesting that data from coflowing jets in ducts should be accompanied by static pressure measurements, since even small pressure gradients appreciably influence jet properties.

Zhang, Q.-F.↗

NASA NTRS1986

Density waves in the solar nebula - Differential Lindblad torque

The differential torque exerted by Lindblad resonances on a perturbing object embedded in a two-dimensional nonself-gravitating disk with density, pressure and sound speed gradients is quantified. First-order corrections are made to account for Keplerian rotation and the presence of the gradients. The total torque is calculated by summing over all resonances in the absence of local wave damping. When applied to the primordial solar nebula disk, the calculations show that disks that cool with increasing heliocentric distance will cause decay of the orbit of the perturbing object. Conditions in which the perturber will escape orbit delay are also described. The characteristic drift time will be no greater than the stochastic accretion time scales. Implications of the calculations for planetary formation are discussed.

Ward, W. R.↗

NASA NTRS1981

On the polarity of cyclostrophic flow in planetary atmospheres

Fluids which are completely inviscid in the mathematical sense do not exist. Thus, the concepts of gradient flow and cyclostrophic balance are interpreted as approximate solutions of a boundary value problem for small but finite viscosity. Large scale phenomena such as the superrotation of Venus and cyclones are effectively bounded by the rigidly rotating planetary surface. This polarizes the circulation and excludes so-called anomalous motions from the flow regime. With scale phenomena such as dust devils, both directions are observed which is attributed to the stochastic nature of wind systems surrounding the disturbance.

Mayr, H. G.↗

NASA NTRS1993

Ion mixing in the plasma sheet boundary layer by drift instabilities

The linear stability properties of collisionless drift instabilities are analyzed in a Harris equilibrium model of the plasma sheet boundary layer (PSBL). The strearmng ions with drift-type instabilities driven in the PSBL are considered. The fluid approximation leads to growth but predicts that the mode width approaches the gyroradius of the energetic ions. Thus an integral equation theory for the modes is developed taking into account that in the PSBL the curvature drift is weak compared with the grad-B drift. The exact wave particle resonance is kept in the nonlocal response functions. Plasma density, temperature, and magnetic gradient drift motions are taken into account. The drift modes produce an anomalous cross-field momentum transport mixing the PSBL ions on the time scale of tens of seconds. A nonlinear simulation is performed which shows the coalescence of the small scale, fast growing modes into large-scale vortices. The relation between these collective modes and plasma sheet transport phenomena is discussed including the comparison with the competing plasma mixing from single-particle stochasticity.

Horton, W.↗