Search NASA⌕ Search

SEARCH · Search NASA

Results for “Conditional Distribution”

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 19 records

Distributionally Robust Decision Making Leveraging Conditional Distributions

Distributionally robust optimization (DRO) is a powerful tool for decision making under uncertainty. It is particularly appealing because of its ability to leverage existing data. However, many practical problems call for decision-making with some auxiliary information, and DRO in the context of conditional distributions is not straightforward. We propose a conditional kernel distributionally robust optimization (CKDRO) method that enables robust decision making under conditional distributions through kernel DRO and the conditional mean operator in the reproducing kernel Hilbert space (RKHS). In particular, we consider problems where there is a correlation between the unknown variable y and an auxiliary observable variable x. Given past data of the two variables and a queried auxiliary variable, CKDRO represents the conditional distribution P(y|x) as the conditional mean operator in the RKHS space and quantifies the ambiguity set in the RKHS as well, which depends on the size of the dataset as well as the query point. To justify the use of RKHS, we demonstrate that the ambiguity set defined in RKHS can be viewed as a ball under a metric that is similar to the Wasserstein metric. The DRO is then dualized and solved via a finite dimensional convex program. The proposed CKDRO approach is applied to a generation scheduling problem and shows that the result of CKDRO is superior to common benchmarks in terms of quality and robustness.

Chen, Yuxiao↗

Adversarial sampling of unknown and high-dimensional conditional distributions

Many engineering problems require the prediction of realization-to-realization variability or a refined description of modeled quantities. In that case, it is necessary to sample elements from unknown high-dimensional spaces with possibly millions of degrees of freedom. While there exist methods able to sample elements from probability density functions (PDF) with known shapes, several approximations need to be made when the distribution is unknown. In this paper the sampling method, as well as the inference of the underlying distribution, are both handled with a data-driven method known as generative adversarial networks (GAN), which trains two competing neural networks to produce a network that can effectively generate samples from the training set distribution. In practice, it is often necessary to draw samples from conditional distributions. When the conditional variables are continuous, only one (if any) data point corresponding to a particular value of a conditioning variable may be available, which is not sufficient to estimate the conditional distribution. This work handles this problem using an a priori estimation of the conditional moments of a PDF. Herein, two approaches, stochastic estimation, and an external neural network are compared for computing these moments; however, any preferred method can be used. The algorithm is demonstrated in the case of the deconvolution of a filtered turbulent flow field. It is shown that all the versions of the proposed algorithm effectively sample the target conditional distribution with minimal impact on the quality of the samples compared to state-of-the-art methods. Additionally, the procedure can be used as a metric for the diversity of samples generated by a conditional GAN (cGAN) conditioned with continuous variables.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Distributed Conditions for Small-signal Stability of Power Grids and Local Control Design

Operating modern power grids with stability guarantees is markedly important. Typical methods for analyzing and certifying power grid stability are largely centralized relying on the ability of the system operator to gather network-wide information and accurately compute the system's eigenvalues. These methods are oftentimes not privacy-preserving and computationally burdensome. They are therefore, not well-suited to modern power grids where small-signal stability has to be evaluated timely, efficiently and in a privacy-preserving fashion. Herein, we introduce a distributed methodology for certifying small-signal stability of power grids and designing the local controllers. First, we analytically derive distributed conditions for network-wide stability that bus agents can inspect using local information. By leveraging these conditions, we then introduce a distributed control design algorithm (DCDA) that can guide the local control design so that stability of the interconnected system is guaranteed. The agents that adopt the proposed distributed algorithm are responsible for tuning their local controllers, producing their local control commands and ensuring that their local stability condition is met. The system operator is only responsible for verifying network-wide stability upon receiving affirmative responses from all agents and, announcing, that the overall system is stable. The proposed DCDA algorithm is numerically validated via simulations using the IEEE 39-bus system.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Conditional distribution estimation of building characteristics with diffusion models for urban energy modeling

Understanding current energy consumption behavior in communities is critical for informing future energy use decisions and enabling efficient energy management. Urban energy models, which are used to simulate these energy use patterns, require large datasets with detailed building characteristics for accurate outcomes. However, such detailed characteristics at the individual building level are often unknown and costly to acquire, or unavailable. Through this work, we propose using a generative modeling approach to generate realistic building attributes to fill in the data gaps and finally provide complete characteristics as inputs to energy models. Our model learns complex, building-level patterns from training on a large-scale residential building stock model containing 2.2 million buildings. We employ a tabular diffusion-based framework that is designed to handle heterogeneous (discrete and continuous) features in tabular building data, such as occupancy, floor area, heating, cooling, and other equipment details. We develop a capability for conditional diffusion, enabling the imputation of missing building characteristics conditioned on known attributes. We conduct a comprehensive validation of our conditional diffusion model, firstly by comparing the generated conditional distributions against the underlying data distribution, and secondly, by performing a case study for a Baltimore residential region, showing the practical utility of our approach. Our work is one of the first to demonstrate the potential of generative modeling to accelerate building energy modeling workflows.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Distributed Online Conditions Database of the BaBar Experiment

This paper presents CDB - The distributed Conditions Database of the BaBar Experiment. CDB is the second major iteration of the database deployed in BaBar for production use as of October 2002. It replaced the original version of the database used through the first three years of the data taking. The new design and its implementation aims to improve the performance and scalability of the original database and addresses the emerging challenges of the distributed data production and analysis system of the Experiment.

99 GENERAL AND MISCELLANEOUS↗

Classification with asymmetric label noise: Consistency and maximal denoising

In many real-world classification problems, the labels of training examples are randomly corrupted. Most previous theoretical work on classification with label noise assumes that the two classes are separable, that the label noise is independent of the true class label, or that the noise proportions for each class are known. In this work, we give conditions that are necessary and sufficient for the true class-conditional distributions to be identifiable. These conditions are weaker than those analyzed previously, and allow for the classes to be nonseparable and the noise levels to be asymmetric and unknown. The conditions essentially state that a majority of the observed labels are correct and that the true class-conditional distributions are “mutually irreducible,” a concept we introduce that limits the similarity of the two distributions. For any label noise problem, there is a unique pair of true class-conditional distributions satisfying the proposed conditions, and we argue that this pair corresponds in a certain sense to maximal denoising of the observed distributions. Our results are facilitated by a connection to “mixture proportion estimation,” which is the problem of estimating the maximal proportion of one distribution that is present in another. We establish a novel rate of convergence result for mixture proportion estimation, and apply this to obtain consistency of a discrimination rule based on surrogate loss minimization. Experimental results on benchmark data and a nuclear particle classification problem demonstrate the efficacy of our approach. MSC 2010 subject classifications: Primary 62H30; secondary 68T10. Keywords and phrases: Classification, label noise, mixture proportion estimation, surrogate loss, consistency.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Modeling highway travel time distribution with conditional probability models

ABSTRACT Under the sponsorship of the Federal Highway Administration's Office of Freight Management and Operations, the American Transportation Research Institute (ATRI) has developed performance measures through the Freight Performance Measures (FPM) initiative. Under this program, travel speed information is derived from data collected using wireless based global positioning systems. These telemetric data systems are subscribed and used by trucking industry as an operations management tool. More than one telemetric operator submits their data dumps to ATRI on a regular basis. Each data transmission contains truck location, its travel time, and a clock time/date stamp. Data from the FPM program provides a unique opportunity for studying the upstream-downstream speed distributions at different locations, as well as different time of the day and day of the week. This research is focused on the stochastic nature of successive link travel speed data on the continental United States Interstates network. Specifically, a method to estimate route probability distributions of travel time is proposed. This method uses the concepts of convolution of probability distributions and bivariate, link-to-link, conditional probability to estimate the expected distributions for the route travel time. Major contribution of this study is the consideration of speed correlation between upstream and downstream contiguous Interstate segments through conditional probability. The established conditional probability distributions, between successive segments, can be used to provide travel time reliability measures. This study also suggests an adaptive method for calculating and updating route travel time distribution as new data or information is added. This methodology can be useful to estimate performance measures as required by the recent Moving Ahead for Progress in the 21st Century Act (MAP 21).

Oliveira Neto, Francisco Moraes↗

A multistage distributionally robust optimization approach to water allocation under climate uncertainty

This paper investigates a Multistage Distributionally Robust Optimization (MDRO) approach to water allocation under climate uncertainty. The MDRO is formed by creating sets of conditional distributions (called conditional ambiguity sets) on a finite scenario tree. The distributions in the conditional ambiguity sets remain close to a nominal conditional distribution according a ø-divergence (e.g., Kullback-Leibler divergence, Hellinger distance, Burg entropy, etc.). Here, the paper discusses a decomposition algorithm to solve the resulting MDRO with ø-divergences, which uses the dual formulation and solves only linear subproblems instead of convex ones. Some properties of the algorithm such as generating feasible policies and valid upper/lower bounds are established. The paper then applies the modeling and solution techniques to allocate water in a rapidly-developing area of Tucson, Arizona. Tucson, like many arid and semi-arid regions around the world, faces considerable uncertainty in its ability to provide water for its citizens in the future. The primary sources of uncertainty in the Tucson region include (1) unpredictable population growth, (2) the availability of water from the Colorado River, and (3) the effects of climate variability on water consumption. This paper integrates forecasts for all these sources of uncertainty into a single optimization model for robust and sustainable water allocation. Then, it uses this model to analyze the value of constructing additional treatment facilities to reduce future water shortages. The results indicate that the MDRO approach can be very valuable for water managers by providing insights to minimize their risks and help them plan for the future.

54 ENVIRONMENTAL SCIENCES↗

Ford Thermoelectric HVAC Project (Final Report)

Current light-duty vehicles provide passenger thermal comfort primarily through the use of a centralized HVAC (Heating, Ventilation, and Air Conditioning) unit that distributes conditioned air to vent locations throughout the vehicle. A substantial portion of the air energy content is spent conditioning various elements of the vehicle's interior structures. This is not an overwhelming issue in today’s vehicles, but as pressure builds to improve powertrain efficiency, the energy available to provide occupant comfort will decrease. Thermoelectric devices were identified by the US Department of Energy as one method to potentially reduce HVAC-related energy consumption in vehicles. The objective of the Ford Thermoelectric HVAC Project was to identify a technical and business approach to accelerate research and development of light-duty automotive thermoelectric HVAC technology. A zonal HVAC system was developed to provide thermal comfort to each occupant, with the objective of reducing the overall power consumed by the HVAC system compared with a baseline system. The system was designed to provide a basis for a down-sized central HVAC unit, while providing equivalent comfort and equivalent heating/cooling time-to-comfort at startup, and reduce the energy consumed to provide these functions. Project success was evaluated using occupant comfort and energy consumption metrics, based on a novel thermal comfort-modeling approach, and empirical testing. The ability to produce and market the thermoelectric devices developed in this program was assessed by a business analysis that outlined key steps required for thermoelectric HVAC devices to come to market. Key hardware and controls attribute requirements were identified to determine how they contribute to the commercialization potential of the technology. Final testing and analysis showed that use of thermal comfort and zonal design criteria reduced the energy consumption of the climate control system in both heating and cooling tests, and were significantly better in low vehicle occupancy cases where zonal strategies could be employed. Thermoelectric devices were utilized in the vehicle design to support achieving the technical performance goals of the project. It was found that these devices need additional development in order to address commercial issues, due to their cost and performance limitations.

99 GENERAL AND MISCELLANEOUS↗

A new method for wind speed forecasting based on copula theory

Highlights: • A copula method is developed to determine the representative wind speed in wind farm. • The method proposed can determine the joint distribution without assuming the variables to be independent. • The representative wind speed obtained by the conditional distribution is more reasonably. How to determine representative wind speed is crucial in wind resource assessment. Accurate wind resource assessments are important to wind farms development. Linear regressions are usually used to obtain the representative wind speed. However, terrain flexibility of wind farm and long distance between wind speed sites often lead to low correlation. In this study, copula method is used to determine the representative year's wind speed in wind farm by interpreting the interaction of the local wind farm and the meteorological station. The result shows that the method proposed here can not only determine the relationship between the local anemometric tower and nearby meteorological station through Kendall's tau, but also determine the joint distribution without assuming the variables to be independent. Moreover, the representative wind data can be obtained by the conditional distribution much more reasonably. We hope this study could provide scientific reference for accurate wind resource assessments.

54 ENVIRONMENTAL SCIENCES↗

A Recursive Multiscale Correlation-Averaging Algorithm for an Automated Distributed Road Condition Monitoring System

A signal processing approach is proposed to jointly filter and fuse spatially indexed measurements captured from many vehicles. It is assumed that these measurements are influenced by both sensor noise and measurement indexing uncertainties. Measurements from low-cost vehicle-mounted sensors (e.g., accelerometers and Global Positioning System (GPS) receivers) are properly combined to produce higher quality road roughness data for cost-effective road surface condition monitoring. The proposed algorithms are recursively implemented and thus require only moderate computational power and memory space. These algorithms are important for future road management systems, which will use on-road vehicles as a distributed network of sensing probes gathering spatially indexed measurements for condition monitoring, in addition to other applications, such as environmental sensing and/or traffic monitoring. Our method and the related signal processing algorithms have been successfully tested using field data.

47 OTHER INSTRUMENTATION↗

A probabilistic graphical model based stochastic input model construction

Model reduction techniques have been widely used in modeling of high-dimensional stochastic input in uncertainty quantification tasks. However, the probabilistic modeling of random variables projected into reduced-order spaces presents a number of computational challenges. Due to the curse of dimensionality, the underlying dependence relationships between these random variables are difficult to capture. In this work, a probabilistic graphical model based approach is employed to learn the dependence by running a number of conditional independence tests using observation data. Thus a probabilistic model of the joint PDF is obtained and the PDF is factorized into a set of conditional distributions based on the dependence structure of the variables. The estimation of the joint PDF from data is then transformed to estimating conditional distributions under reduced dimensions. To improve the computational efficiency, a polynomial chaos expansion is further applied to represent the random field in terms of a set of standard random variables. This technique is combined with both linear and nonlinear model reduction methods. Numerical examples are presented to demonstrate the accuracy and efficiency of the probabilistic graphical model based stochastic input models. - Highlights: • Data-driven stochastic input models without the assumption of independence of the reduced random variables. • The problem is transformed to a Bayesian network structure learning problem. • Examples are given in flows in random media.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

SILICON CARBIDE GRAIN BOUNDARY DISTRIBUTIONS, IRRADIATION CONDITIONS, AND SILVER RETENTION IN IRRADIATED AGR-1 TRISO FUEL PARTICLES

Precession electron diffraction in the transmission electron microscope was used to map grain orientation and ultimately determine grain boundary misorientation angle distributions, relative fractions of grain boundary types (random high angle, low angle or coincident site lattice (CSL)-related boundaries) and the distributions of CSL-related grain boundaries in the SiC layer of irradiated TRISO-coated fuel particles. Two particles from the AGR-1 experiment exhibiting high Ag-110m retention (>80%) were compared to a particle exhibiting low Ag-110m retention (<19%). Irradiated particles with high Ag-110m retention exhibited a lower fraction of random, high angle grain boundaries compared to the low Ag-110m retention particle. An inverse relationship between the random, high angle grain boundary fraction and Ag-110m retention is found and is consistent with grain boundary percolation theory. Also, comparison of the grain boundary distributions with previously reported unirradiated grain boundary distributions, based on SEM-based EBSD for similarly fabricated particles, showed only small differences, i.e. a greater low angle grain boundary fraction in unirradiated SiC. It was, thus, concluded that SiC layers with grain boundary distributions susceptible to Ag-110m release were present prior to irradiation. Finally, irradiation parameters were found to have little effect on the association of fission product precipitates with specific grain boundary types.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

Local structure graph models with higher-order dependence

Local structure graph models (LSGMs) describe random graphs and networks as a Markov random field (MRF)—each graph edge has a specified conditional distribution dependent on explicit neighbourhoods of other graph edges. Centered parameterizations of LSGMs allow for direct control and interpretation of parameters for large- and small-scale structures (e.g., marginal means vs. dependence). Here, we extend this parameterization to account for triples of dependent edges and illustrate the importance of centered parameterizations for incorporating covariates and interpreting parameters. Using a MRF framework, common exponential random graph models are also shown to induce conditional distributions without centered parameterizations and thereby have undesirable features. This work attempts to advance graph models through conditional model specifications with modern parameterizations, covariates and higher-order dependencies.

97 MATHEMATICS AND COMPUTING↗

Prediction of interest rate using CKLS model with stochastic parameters

The Chan, Karolyi, Longstaff and Sanders (CKLS) model is a popular one-factor model for describing the spot interest rates. In this paper, the four parameters in the CKLS model are regarded as stochastic. The parameter vector φ{sup (j)} of four parameters at the (J+n)-th time point is estimated by the j-th window which is defined as the set consisting of the observed interest rates at the j′-th time point where j≤j′≤j+n. To model the variation of φ{sup (j)}, we assume that φ{sup (j)} depends on φ{sup (j−m)}, φ{sup (j−m+1)},…, φ{sup (j−1)} and the interest rate r{sub j+n} at the (j+n)-th time point via a four-dimensional conditional distribution which is derived from a [4(m+1)+1]-dimensional power-normal distribution. Treating the (j+n)-th time point as the present time point, we find a prediction interval for the future value r{sub j+n+1} of the interest rate at the next time point when the value r{sub j+n} of the interest rate is given. From the above four-dimensional conditional distribution, we also find a prediction interval for the future interest rate r{sub j+n+d} at the next d-th (d≥2) time point. The prediction intervals based on the CKLS model with stochastic parameters are found to have better ability of covering the observed future interest rates when compared with those based on the model with fixed parameters.

97 MATHEMATICS AND COMPUTING↗