Search NASA⌕ Search

SEARCH · Search NASA

Results for “stochastic approximation”

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 919 records · Page 51

Stochastic models for software project management

A method for determining the number and characteristics of milestones to be achieved during a development project is presented in order that effective monitors of progress can be provided. Projections of progress data lead to estimates of the completion with determinable accuracy, but accuracy imposes a requirement that the number of milestones be inversely proportional to the estimate-error variance, and that the milestones themselves be defined in such a way that each represents approximately the same level of effort to complete.

Tausworthe, R. C.↗

Proactive Operations and Investment Planning via Stochastic Optimization to Enhance Power Systems’ Extreme Weather Resilience

We present scalable stochastic optimization approaches for improving power systems’ resilience to extreme weather events. We consider both proactive redispatch and transmission line hardening as alternatives for mitigating expected load shed due to extreme weather, resulting in large-scale stochastic linear programs (LPs) and mixed-integer linear programs (MILPs). We solve these stochastic optimization problems with progressive hedging (PH), a parallel, scenario-based decomposition algorithm. Our computational experiments indicate that our proposed method for enhancing power system resilience can provide high-quality solutions efficiently. With up to 128 scenarios on a 2,000-bus network, the operations (redispatch) and investment (hardening) resilience problems can be solved in approximately 6 min and 2 h of wall-clock time, respectively. Additionally, we solve the investment problems with up to 512 scenarios, demonstrating that the approach scales very well with the number of scenarios. Moreover, the method produces high quality solutions that result in statistically significant reductions in expected load shed. Our proposed approach can be augmented to incorporate a variety of other operational and investment resilience strategies, or a combination of such strategies.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Projective Integral Updates for High-Dimensional Variational Inference

Variational inference is an approximation framework for Bayesian inference that seeks to improve quantified uncertainty in predictions by optimizing a simplified distribution over parameters to stand in for the full posterior. Capturing model variations that remain consistent with training data enables more robust predictions by reducing parameter sensitivity. This work introduces a fixed-point optimization for variational inference that is applicable when every feasible log density can be expressed as a linear combination of functions from a given basis. In such cases, the optimizer becomes a fixed-point of projective integral updates. When the basis spans univariate quadratics in each parameter, the feasible distributions are Gaussian mean-fields and the projective integral updates yield quasi-Newton variational Bayes (QNVB). Other bases and updates are also possible. Since these updates require high-dimensional integration, this work begins by proposing an efficient quasirandom sequence of quadratures for mean-field distributions. Each iterate of the sequence contains two evaluation points that combine to correctly integrate all univariate quadratic functions and, if the mean-field factors are symmetric, all univariate cubics. More importantly, averaging results over short subsequences achieves periodic exactness on a much larger space of multivariate polynomials of quadratic total degree. The corresponding variational updates require four loss evaluations with standard (not second-order) backpropagation to eliminate error terms from over half of all multivariate quadratic basis functions. Furthermore, this integration technique is motivated by first proposing stochastic blocked mean-field quadratures, which may be useful in other contexts. A PyTorch implementation of QNVB allows for better control over model uncertainty during training than competing methods. Experiments demonstrate superior generalizability for multiple learning problems and architectures.

Gaussian mean-field↗

Emulation of radiation transport in 3D stochastic media using 1D planar Monte Carlo stochastic media radiation transport algorithms

A subset of stochastic media radiation transport problems involves those in which radiation is incident on a thin slab of stochastic material. Particle tracking in 3D for such problems is expensive, and 1D planar models lack accuracy because they only allow the material to change in one dimension. Therefore, we propose dimensional emulation, which through a slight modification allows existing 1D planar geometry stochastic media radiation transport models to reproduce results from the equivalent 3D models by allowing the material to change in all three dimensions, reproducing the fidelity of the 3D model for the low computational cost of the 1D planar model. In this work, we apply dimensional emulation to three Monte Carlo stochastic media radiation transport models: Chord Length Sampling (CLS), the Local Realization Preserving method (LRP), and a variant of Conditional Point Sampling (CoPS). For a common Markovian benchmark set, the 3D emulation variants of these algorithms are numerically verified to reproduce the results of the 3D variants within statistics while running 1.3 to 2 times faster in the implementation within Sandia National Laboratories open-source research code PlaybookMC. The 3D emulation variants are also shown to yield a 72%–92% reduction in error for the thin slab problems in comparison to the 1D benchmark. As a result, the 3D emulation variant of CLS and CoPS-1 are shown to reproduce 3D CLS results that were used to approximate results for a 3D spherical inclusion geometry benchmark set.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Reliable algorithms for power system analysis in the presence of data uncertainties

A robust and reliable power flow analysis represents an essential requirement for many power systems applications as far as network optimization, voltage control, state estimation, and service restoration are concerned. The most common power flow approach, referred to here as a deterministic power flow (PLF), requires precise or 'crisp' values chosen by the analyst for each input variable. The solution provides precise network voltages and flows through each line. The specified values rest upon assumptions about the operating condition derived from historical measurements or predictions about future conditions and thus, cannot be considered accurate. Even in the case where the inputs are based on measurements, inaccuracies arise from time-skew problems, three-phase unbalance, static modeling approximations of dynamic components (e.g., transformer tap changers), variations in line parameters, and so on. The advent of deregulation and competitive power markets will only exacerbate this problem as well-known generation patterns change, loading becomes less predictable and the transmission paths grow more diverse. Conventional methodologies proposed in literature address tolerance analysis of power flow solution by means of detailed probabilistic methods, accounting for the variability and stochastic nature of the input data, and sampling based approaches. In particular uncertainty propagation using sampling based methods, such as the Monte Carlo, requires several model runs that sample various combinations of input values. Since the number of model runs can sometimes be very large, the required computer resources can sometimes be prohibitively expensive resulting in substantial computational demands. As far as probabilistic methods are concerned, they represent a useful tool, especially for planning studies, but, as evidenced by the many discussions reported in literature, they could reveal some shortcomings principally arising from: (1) the non-normal distribution and the statistical dependence of the input data; and (2) the difficulty arising in accurately identifying probability distributions for some input data, such as the power generated by wind or photovoltaic generators. All these could result in time consuming computations with several limitations in practical applications especially in power flow analysis of complex power networks. In order to try and overcome some of these limitations, obtaining thereby comprehensive power flow solution tolerance analysis at adequate computational costs, self validated computation could play a crucial role. Armed with such a vision, this chapter will analyze two advanced techniques for power flow analysis in the presence of data uncertainty namely the boundary power flow and the affine arithmetic power flow.

24 POWER TRANSMISSION AND DISTRIBUTION↗

THE IMPACT OF THE IONOSPHERE ON GROUND-BASED DETECTION OF THE GLOBAL EPOCH OF REIONIZATION SIGNAL

The redshifted 21 cm line of neutral hydrogen (H i), potentially observable at low radio frequencies (∼50–200 MHz), is a promising probe of the physical conditions of the intergalactic medium during Cosmic Dawn and the Epoch of Reionization (EoR). The sky-averaged H i signal is expected to be extremely weak (∼100 mK) in comparison to the Galactic foreground emission (∼10{sup 4} K). Moreover, the sky-averaged spectra measured by ground-based instruments are affected by chromatic propagation effects (∼tens of kelvin) originating in the ionosphere. We analyze data collected with the upgraded Broadband Instrument for Global Hydrogen Reionization Signal system deployed at the Murchison Radio-astronomy Observatory to assess the significance of ionospheric effects on the detection of the global EoR signal. The ionospheric effects identified in these data are, particularly during nighttime, dominated by absorption and emission. We measure some properties of the ionosphere, such as the electron temperature (T{sub e} ≈ 470 K at nighttime), magnitude, and variability of optical depth (τ{sub 100} {sub MHz} ≈ 0.01 and δτ ≈ 0.005 at nighttime). According to the results of a statistical test applied on a large data sample, very long integrations (∼100 hr collected over approximately 2 months) lead to increased signal-to-noise ratio even in the presence of ionospheric variability. This is further supported by the structure of the power spectrum of the sky temperature fluctuations, which has flicker noise characteristics at frequencies ≳10{sup −5} Hz, but becomes flat below ≈10{sup −5} Hz. Hence, we conclude that the stochastic error introduced by the chromatic ionospheric effects tends to zero in an average. Therefore, the ionospheric effects and fluctuations are not fundamental impediments preventing ground-based instruments from integrating down to the precision required by global EoR experiments, provided that the ionospheric contribution is properly accounted for in the data analysis.

79 ASTRONOMY AND ASTROPHYSICS↗

Exploring redshift-space distortions in large-scale structure

We explore and compare different ways large-scale structure observables in redshift-space and real space can be connected. These include direct computation in Lagrangian space, moment expansions and two formulations of the streaming model. Here we derive for the first time a Fourier space version of the streaming model, which yields an algebraic relation between the real- and redshift-space power spectra which can be compared to earlier, phenomenological models. By considering the redshift-space 2-point function in both configuration and Fourier space, we show how to generalize the Gaussian streaming model to higher orders in a systematic and computationally tractable way. We present a closed-form solution to the Zeldovich power spectrum in redshift space and use this as a framework for exploring convergence properties of different expansion approaches. While we use the Zeldovich approximation to illustrate these results, much of the formalism and many of the relations we derive hold beyond perturbation theory, and could be used with ingredients measured from N-body simulations or in other areas requiring decomposition of Cartesian tensors times plane waves. Lastly, we finish with a discussion of the redshift-space bispectrum, bias and stochasticity and terms in Lagrangian perturbation theory up to 1-loop order.

79 ASTRONOMY AND ASTROPHYSICS↗

Wavefront shaping with a Hadamard basis for scattering soil imaging

Here, soil is a scattering medium that inhibits imaging of plant-microbial-mineral interactions that are essential to plant health and soil carbon sequestration. However, optical imaging in the complex medium of soil has been stymied by the seemingly intractable problems of scattering and contrast. Here, we develop a wavefront shaping method based on adaptive stochastic parallel gradient descent optimization with a Hadamard basis to focus light through soil mineral samples. Our approach allows a sparse representation of the wavefront with reduced dimensionality for the optimization. We further divide the used Hadamard basis set into subsets and optimize a certain subset at once. Simulation and experimental optimization results demonstrate our method has an approximately seven times higher convergence rate and overall better performance compared to that with optimizing all pixels at once. The proposed method can benefit other high-dimensional optimization problems in adaptive optics and wavefront shaping.

47 OTHER INSTRUMENTATION↗

Single particle momentum and angular distributions in hadron-hadron collisions at ultrahigh energies

The forward-backward charged multiplicity distribution (P n sub F, n sub B) of events in the 540 GeV antiproton-proton collider has been extensively studied by the UA5 Collaboration. It was pointed out that the distribution with respect to n = n sub F + n sub B satisfies approximate KNO scaling and that with respect to Z = n sub F - n sub B is binomial. The geometrical model of hadron-hadron collision interprets the large multiplicity fluctuation as due to the widely different nature of collisions at different impact parameters b. For a single impact parameter b, the collision in the geometrical model should exhibit stochastic behavior. This separation of the stochastic and nonstochastic (KNO) aspects of multiparticle production processes gives conceptually a lucid and attractive picture of such collisions, leading to the concept of partition temperature T sub p and the single particle momentum spectrum to be discussed in detail.

Chou, T. T.↗

Levermore-Pomraning Model Results for an Interior Source Binary Stochastic Medium Benchmark Problem

The accuracy of the Levermore-Pomraning model for particle transport through a binary stochastic medium is investigated using an interior source benchmark problem. As in previous comparisons of the model for incident angular flux benchmark problems, the model accurately computes the leakage and the scalar flux distributions for optically thin slabs. The model is less accurate for more optically thick slabs but has a maximum relative error in the leakage of approximately 10% for the problems examined. The maximum root-mean-squared relative errors for the total and material scalar flux distributions approach 65% for the more optically thick slabs. Consistent with previous benchmark comparisons, the results of these interior source benchmark comparisons demonstrate that the Levermore-Pomraning model produces qualitatively correct and semi-quantitatively correct results for both leakage values and scalar flux distributions.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Data driven analytic continuation for one particle spectral functions

In this proceeding, an idea is outlined suggesting a generic treatment on any type of input data for a numerical analytic continuation problem, which is needed when dynamical information is to be extracted from a calculationally convenient one particle imaginary time Green function. The quality of the resulting spectral function will rely only on the data to be treated, viz, data-driven. This is different from the Maximum Entropy or the Stochastic method which relies on an entropy term to guide convergence of the resulting spectral function.

36 MATERIALS SCIENCE↗

Modeling microbial dynamics in heterogeneous environments: Growth on soil carbon sources

We have developed a new hybrid model to study how microbial dynamics are affected by the heterogeneity in the physical structure of the environment. The modeling framework can represent porous media such as soil. The individual based biological model can explicitly simulate microbial diversity, and cell metabolism is regulated via optimal allocation of cellular resources to enzyme synthesis, control of growth rate by protein synthesis capacity, and shifts to dormancy. This model was developed to study how microbial community functioning is influenced by local environmental conditions and by the functional attributes of individual microbes. Different strategies for acquisition of carbon from polymeric cellulose were investigated. Bacteria that express membrane-associated hydrolase had different growth and survival dynamics in soil pores than bacteria that release extracellular hydrolases. The kinetic differences may suggest different functional roles for these two classes of microbes in cellulose utilization. Our model predicted an emergent behavior in which co-existence led to higher cellulose utilization efficiency and reduced stochasticity. Microbial community dynamics were simulated at two spatial scales: micro-pores that resemble 6-20 {micro}m size portions of the soil physical structure and in 111 {micro}m size soil aggregates with a random pore structure. Trends in dynamic properties were very similar at these two scales, implying that micro-scale studies can be useful approximations to aggregate scale studies when local effects on microbial dynamics are studied.

54 ENVIRONMENTAL SCIENCES↗

Prediction of the phase equilibria of methane hydrates using the direct phase coexistence methodology

The direct phase coexistence method is used for the determination of the three-phase coexistence line of sI methane hydrates. Molecular dynamics (MD) simulations are carried out in the isothermal–isobaric ensemble in order to determine the coexistence temperature (T{sub 3}) at four different pressures, namely, 40, 100, 400, and 600 bar. Methane bubble formation that results in supersaturation of water with methane is generally avoided. The observed stochasticity of the hydrate growth and dissociation processes, which can be misleading in the determination of T{sub 3}, is treated with long simulations in the range of 1000–4000 ns and a relatively large number of independent runs. Statistical averaging of 25 runs per pressure results in T{sub 3} predictions that are found to deviate systematically by approximately 3.5 K from the experimental values. This is in good agreement with the deviation of 3.15 K between the prediction of TIP4P/Ice water force field used and the experimental melting temperature of ice Ih. The current results offer the most consistent and accurate predictions from MD simulation for the determination of T{sub 3} of methane hydrates. Methane solubility values are also calculated at the predicted equilibrium conditions and are found in good agreement with continuum-scale models.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Capacity Value of Wind Power

Power systems are planned such that they have adequate generation capacity to meet the load, according to a defined reliability target. The increase in the penetration of wind generation in recent years has led to a number of challenges for the planning and operation of power systems. A key metric for system adequacy is the capacity value of generation. The capacity value of a generator is the contribution that a given generator makes to overall system adequacy. The variable and stochastic nature of wind sets it apart from conventional energy sources. As a result, the modeling of wind generation in the same manner as conventional generation for capacity value calculations is inappropriate. In this paper a preferred method for calculation of the capacity value of wind is described and a discussion of the pertinent issues surrounding it is given. Approximate methods for the calculation are also described with their limitations highlighted. The outcome of recent wind capacity value analyses in Europe and North America are highlighted with a description of open research questions also given.

17 WIND ENERGY↗

DAKOTA Design Analysis Kit for Optimization and Terascale

The DAKOTA (Design Analysis Kit for Optimization and Terascale Applications) toolkit provides a flexible and extensible interface between simulation codes (computational models) and iterative analysis methods. By employing object-oriented design to implement abstractions of the key components required for iterative systems analyses, the DAKOTA toolkit provides a flexible and extensible problem-solving environment for design and analysis of computational models on high performance computers.A user provides a set of DAKOTA commands in an input file and launches DAKOTA. DAKOTA invokes instances of the computational models, collects their results, and performs systems analyses. DAKOTA contains algorithms for optimization with gradient and nongradient-based methods; uncertainty quantification with sampling, reliability, polynomial chaos, stochastic collocation, and epistemic methods; parameter estimation with nonlinear least squares methods; and sensitivity/variance analysis with design of experiments and parameter study methods. These capabilities may be used on their own or as components within advanced strategies such as hybrid optimization, surrogate-based optimization, mixed integer nonlinear programming, or optimization under uncertainty. Services for parallel computing, simulation interfacing, approximation modeling, fault tolerance, restart, and graphics are also included.

SciDAC↗

Set-point control of nonlinear systems

A method of nonlinear stochastic control is applied to the problem of providing set-point control. The procedure is illustrated for a nonlinear second-order plant. The controller design uses a variation of the 'multiple-model adaptive control' approach. Specific problems related to the design are investigated, including learning time and set-point accuracy dependence on the number of approximating models, chord versus tangent approximations, and the a priori Markov switching parameter values. Simulations are provided for different situations under various levels of measurement noise.

Vanlandingham, H. F.↗

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↗