In advanced democracies, many citizens abstain from participating in the political process. Does low and unequal voter turnout influence partisan election results or public policies? If so, how can participation be increased and how can the electorate become more representative of the greater population? / Government

Computer models or simulators are becoming increasingly common in many fields in science and engineering, powered by the phenomenal growth in computer hardware over the past decades. Many of these simulators implement a particular mathematical model as a deterministic computer code, meaning that running the simulator again with the same input gives the same output. Often running the code involves some computationally expensive tasks, such as solving complex systems of partial differential equations numerically. When simulator runs become too long, it may limit their usefulness. In order to overcome time or budget constraints by making the most out of limited computational resources, a statistical methodology has been proposed, known as the "Design and Analysis of Computer Experiments". The main idea is to run the expensive simulator only at a relatively few, carefully chosen design points in the input space, and based on the outputs construct an emulator (statistical model) that can emulate (predict) the output at new, untried locations at a fraction of the cost. This approach is useful provided that we can measure how much the predictions of the cheap emulator deviate from the real response surface of the original computer model. One way to quantify emulator error is to construct pointwise prediction bands designed to envelope the response surface and make assertions that the true response (simulator output) is enclosed by these envelopes with a certain probability. Of course, to be able to make such probabilistic statements, one needs to introduce some kind of randomness. A common strategy that we use here is to model the computer code as a random function, also known as a Gaussian stochastic process. We concern ourselves with smooth response surfaces and use the Gaussian covariance function that is ideal in cases when the response function is infinitely differentiable. In this thesis, we propose Fast Bayesian Inference (FBI) that is both computationally efficient and can be implemented as a black box. Simulation results show that it can achieve remarkably accurate prediction uncertainty assessments in terms of matching coverage probabilities of the prediction bands and the associated reparameterizations can also help parameter uncertainty assessments. / Science, Faculty of / Statistics, Department of / Graduate

The Dirichlet process is perhaps the most popular prior used in the nonparametric Bayesian inference. This prior which is placed on the space of probability distributions has conjugacy property and asymptotic consistency. In this thesis, our concentration is on applying this nonparametric Bayesian inference on the Tukey depth and Tukey median. Due to the complexity of the distribution of Tukey median, we use this nonparametric Bayesian inference, namely the Lo’s bootstrap, to approximate the distribution of the Tukey median. We also compare our results with the Efron’s bootstrap and Rubin’s bootstrap. Furthermore, the existing asymptotic theory for the Tukey median is reviewed. Based on these existing results, we conjecture that the bootstrap sample Tukey median converges to the same asymp- totic distribution and our simulation supports the conjecture that the asymptotic consistency holds.

This study evaluated the comprehension of inference statements by language-learning disabled (LLD) children and children with normal language development (NL) under two conditions: uncontextualized and contextualized. The contextualized condition was designed to encourage retrieval of information from the subject's general knowledge — a procedure proposed to encourage elaborative inference-making. Two text passages were analyzed according to a model developed by Trabasso and presented by Trabasso, van den Broek & Suh (1989), which yielded a set of bridging causal connections across clause units. Sets of three true and three false causal inference statements were developed to represent bridging inferences for each story. In addition, three true and three premise statements were obtained directly from each story, yielding a total of twelve statements for each text. Subjects were ten language-learning disabled students (7 boys, 3 girls) and ten children with normal language development (5 boys, 5 girls) aged 9 to 11 years old. Mean age for children in both groups was 10 years, 4 months. Children were selected for the LLD group on the basis of the following criteria: (1) enrollment in a learning assistance or learning resource program for learning-disabled students, preferably for remediation of Language Arts; (2) history of speech-language intervention in preschool or early primary grades; (3) normal nonverbal cognitive skills; (4) lexical and syntactic comprehension within normal abilities (as determined by standardized language tests for the LLD group); (5) native English speaker and (6) normal hearing ability. Every subject received both stories and conditions. Story presentation and condition were counterbalanced across 8 of the 10 subjects in each group; condition only was counterbalanced across the remaining two subjects in each group. Inference and premise statements were randomized; each random set was randomly presented to each subject. Items were scored correct or incorrect. Subjects were also asked to answer open-ended wh-questions. Responses were compared and analyzed using a nonparametric statistical method appropriate for small sample sizes. Results indicated significant differences between the LLD and the NL groups on the number of correct responses to inference and premise items. Both groups scored significantly worse on inference than premise items. Analysis did not indicate that the LLD group scored significantly worse on inference items than the NL group did. Results also suggested that a contextualization effect operated for both groups, which affected the retention of premise items but acted to improve scores on inference items. This effect was seen most notably for the LLD group. / Medicine, Faculty of / Audiology and Speech Sciences, School of / Graduate

Orientador: Laura Leticia Ramos Rifo / Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Matemática Estatística e Computação Científica / Made available in DSpace on 2018-08-26T15:05:12Z (GMT). No. of bitstreams: 1 Tasca_GustavoHenrique_M.pdf: 979052 bytes, checksum: bb1371bb1b8626882cebcf01550bb823 (MD5) Previous issue date: 2015 / Resumo: Neste trabalho, estudamos métodos de inferência bayesiana para distribuições de cauda longa, que não envolvam o cálculo da função de verossimilhança. Inicialmente, apresentamos uma análise das propriedades de distribuições de cauda pesada e seus casos particulares, como as famílias de distribuições de cauda longa, subexponenciais e de variação regular. Apresentamos algumas estatísticas e seus comportamentos amostrais, a fim de desenvolvermos medidas de diagnóstico. Para obtenção de inferências a posteriori, discutimos o método ABC de mínima entropia e outros algoritmos para verificação e seleção de modelos, que não utilizam o cálculo da função de verossimilhança. Introduzimos um novo algoritmo para seleção de modelos baseado na distribuição preditiva a posteriori, cujos resultados são validados através de simulações e análises de dados reais relacionados à hidrologia / Abstract: In this work, we study Bayesian inference methods for long-tailed distributions that don't involve the evaluation of the likelihood function. Initially, we present an analysis of the properties of heavy-tailed distributions and particular cases, as long-tailed, subexponencial and regular variation families. Some statistics are presented and their sampling behavior studied, in order to develop diagnostic measures. For obtaining posterior inferences, we discuss the minimum entropy ABC and others likelihood-free algorithms, aiming model checking and model selection. We introduce a new model selection algorithm based on the posterior predictive distribution, the results of which are validated through simulations and real data related to river flow / Mestrado / Estatistica / Mestre em Estatística

Orientador: Helio Fiori de Castro / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Mecânica / Made available in DSpace on 2018-08-28T11:56:54Z (GMT). No. of bitstreams: 1 Tyminski_NataliaCezaro_M.pdf: 6550021 bytes, checksum: 3fb888c14fc5d24d29f7f155d0e4f2b0 (MD5) Previous issue date: 2015 / Resumo: As unidades geradoras de energia das usinas são formadas por turbinas e turbo-geradores, que são exemplos típicos de máquinas rotativas. Essas maquinas são componentes críticos, pois são essenciais à geração de energia. Sabendo que a análise dinâmica de máquinas rotativas é uma tarefa complexa envolvendo diversos parâmetros a serem analisados, sua realização não deve considerar apenas o rotor, pois seu comportamento dinâmico é influenciado pela interação com os demais componentes do mesmo sistema. O comportamento dinâmico de uma máquina rotativa é, geralmente, representado por um modelo determinístico. Entretanto, sistemas rotativos reais possuem características estocásticas, visto que os inúmeros parâmetros de projetos possuem incertezas inerentes à fabricação e, principalmente, às condições de operações. Desta forma, modelos estocásticos são uma opção importante para representação de sistemas rotativos na fase de projeto, onde se podem prever os efeitos da variação dos parâmetros de projeto. O tema em foco nesta dissertação de mestrado é a aplicação de Inferência Bayesiana para ajustar um modelo de sistema rotativo. Neste trabalho foram analisadas as incertezas nos parâmetros de projeto de um sistema rotativo, e a partir das incertezas obtidas foi possível obter a resposta estocástica do sistema. A primeira analise considera as incertezas dos parâmetros relacionados ao eixo; como o modulo de elasticidade, a massa especifica do material e o coeficiente de proporcionalidade a matriz de rigidez. Na segunda análise, os parâmetros escolhidos foram os parâmetros de desbalanceamento; ângulo de fase, momento de desbalanceamento e posição axial. Em uma terceira abordagem, foi analisado parâmetros dos mancais hidrodinâmicos, folga radial do mancal e temperatura do óleo lubrificante. A partir das incertezas dos referidos parâmetros, foi possível analisar a propagação de incertezas desses parâmetros no cálculo da posição do eixo no mancal e dos coeficientes dinâmicos dos mancais hidrodinâmicos / Abstract: Energy generation plants rely on units such as turbines and turbo-generators, which are common examples of rotating machines. These machines are critical components in these units, once they are essential to the energy generation. The dynamic analysis of rotating machines is a complex task including several parameters to be considered. This analysis requires taking the rotor into account but also the other components, which affect the dynamic behavior of the system. The dynamic behavior of rotating machines is usually represented by a deterministic model. Although, real rotating system have stochastic characteristics once that the parameters on project have uncertainties. In this way, stochastic models are an important option for the representation of these systems, and it's possible predict the variation's parameters. This study aims the application of Bayesian Inference for model updating on rotating systems. The uncertainties of rotating machines parameters were analyzed, and the system stochastic response was obtained. The first analyzed considers the uncertainties of the beam parameters, as the Young¿s modulus, and the proportionality coefficient to the stiffness matrix. In the second analysis, the selected parameters were the unbalance parameters; phase angle, unbalance moment and axial position. In a third approach, it was analyzed parameters of journal bearings, clearance radial and lubricating oil temperature. From the uncertainties of these parameters, it was possible to analyze the propagation of uncertainties of these parameters, to calculate the center line position in the bearing, and the dynamic coefficients of journal bearings / Mestrado / Mecanica dos Sólidos e Projeto Mecanico / Mestra em Engenharia Mecânica

No description available.

No description available.

Reliable forecasting of wind power generation is crucial to optimal control of costs in generation of electricity with respect to the electricity demand. Here, we propose and analyze stochastic wind power forecast models described by parametrized stochastic differential equations, which introduce appropriate fluctuations in numerical forecast outputs. We use an approximate maximum likelihood method to infer the model parameters taking into account the time correlated sets of data. Furthermore, we study the validity and sensitivity of the parameters for each model. We applied our models to Uruguayan wind power production as determined by historical data and corresponding numerical forecasts for the period of March 1 to May 31, 2016.

Python is dynamically typed interpreted programming language. Thanks to its dynamic type system, it is difficult to compile it into statically typed source code. The kind of source code, where it is exactly specified what types exist and what their structure is. Multiple approaches exist how to achieve this and one of the primary ones is type inference. This approach is attempting to infer the type structure from the source code. In case of Python language, this approach is difficult, because resulting type system is quite complex and language itself is not designed for type inference. In this work, I have focused on identifying subset of this language, so that type inference is possible while keeping the natural way the language is used. Then I implemented a compiler, which will compile this subset into statically typed language, which can be translated into native code.