No description available.

Title: Generalized stable distributions and their applications Author: Mgr. Lenka Slámová, MSc. Department: Department of probability and mathematical statistics Supervisor: Prof. Lev Klebanov, DrSc. Abstract: This thesis deals with different generalizations of the strict stability property with a particular focus on discrete distributions possessing some form of stability property. Three possible definitions of discrete stability are introduced, followed by a study of some particular cases of discrete stable distributions and their properties. The random normalization used in the definition of discrete stability is applicable for continuous random variables as well. A new concept of casual stability is introduced by replacing classical normalization in the definition of stability by random normalization. Examples of casual stable distributions, both discrete and continuous, are given. Discrete stable distributions can be applied in discrete models that exhibit heavy tails. Applications of discrete stable distributions on rating of scientific work and financial time series modelling are presented. A method of parameter estimation for discrete stable family is also introduced. Keywords: discrete stable distribution, casual stability, discrete approximation of stable distribution

Many financial time series show heavy tail behavior. Such tail characteristic is important for risk management. In this research, we focus on the calculation of Value-at-Risk (VaR) for portfolios of financial assets. We consider nonlinear autoregressive models with heavy tail innovations to model the return. Predictive distribution of the return are used to compute the VaR of the portfolios of financial assets. Examples are also given to compare the VaR computed by our approach with those by other methods.

Stabilieji skirstiniai yra plati tikimybinių skirstinių klasė. Atsitiktiniai dydžiai, pasiskirstę pagal stabiliuosius skirstinius, pasižymi savybe – jų suma taip pat yra stabili. Šie pasižymi sunkiomis uodegomis ir, kai kuriais atvejais, asimetriškumu. Taigi jie gerai aprašo duomenis. Pagrindinis šių skirstinių trūkumas yra tas, kad nežinomos tikslios pasiskirstymo ir tankio funkcijų išraiškos (išskyrus kelis atvejus: normalusis, Koši ir Levi skirstiniai). Darbo pradžioje pateikta stabiliųjų skirtinių apžvalga bei jų pritaikymas finansų rinkose. Aprašytos pagrindinės stabiliųjų skirstinių savybės, įverčių skaičiavimo algoritmai bei optimalaus portfelio sudarymas ir jo vertės pokyčio rizikos mato (VaR) skaičiavimas. Antroje darbo dalyje nagrinėjamas optimaliojo investicinio portfelio „normalioje“ ir „stabilioje“ rinkoje sudarymas. Rizikos matu laikomas sklaidos parametras (stabiliuoju atveju) arba standartinis nuokrypis, padalintas iš kvaratinės šaknies iš 2, (normaliuoju atveju). Palyginami portfeliai, sudaryti iš septyniolikos lietuviškų akcijų, gauti pagal skirtingas tikimybines prielaidas. Parodyta, kad optimalieji portfeliai skiriasi, kuomet duomenys yra pasiskirstę pagal stabilųjį ir normalųjį skirstinius. / Stable distributions are a rich class of probability distributions that allow skewness and heavy tails. The lack of closed formulas for densities and distribution functions for all distributions (except Gaussian, Cauchy and Levy distributions) is the major drawback. There is an overview of the stable distributions and their applications in finance markets at the beginning of this paper. There are described basic properties of stable distributions, estimation algorithms and optimal asset allocation and stable computation of Value at Risk in the first part of the work. We analyze an investment allocation problems in this work. We consider as the risk measure the estimate of scale parameter (in the stable case) or the expected value of absolute deviation divided by square root of 2 (in Gaussian case). We examine the optimal allocation between seventeen risky assets with normal or stable distributed returns and then we compare the allocation obtained under the Gaussian and stable distributional assumptions. We show that there are differences in the allocation when the data follow the stable non-Gaussian and the normal distribution.

The Gaussian distribution is typically used to model the additive noise affecting communication systems. However, in many cases the noise cannot be modeled by a Gaussian distribution. In this thesis, we investigate the performance of different communication systems perturbed by non-Gaussian noise. Three families of noise are considered in this work, namely the generalized Gaussian noise, the Laplace noise/interference, and the impulsive noise that is modeled by an α-stable distribution. More specifically, in the first part of this thesis, the impact of an additive generalized Gaussian noise is studied by computing the average symbol error rate (SER) of one dimensional and two dimensional constellations in fading environment. We begin by the simple case of two symbols, i.e. binary phase shift keying (BPSK) constellation. From the results of this constellation, we extended the work to the average SER of an M pulse amplitude modulation (PAM). The first 2 − D constellation is the M quadrature amplitude modulation (QAM) (studied for two geometric shapes, namely square and rectangular), which is the combination of two orthogonal PAM signals (in-phase and quadrature phase PAM). In the second part, the system performance of a circular constellation, namely M phase shift keying (MPSK) is studied in conjunction with a Laplace noise with independent noise components. A closed form and an asymptotic expansion of the SER are derived for two detectors, maximum likelihood and minimum distance detectors. Next, we look at the intra cell interference of a full duplex cellular network which is shown to follow a Laplacian distribution with dependent, but uncorrelated, complex components. The densities of that interference are expressed in a closed form in order to obtain the SER of several communication systems (BPSK, PAM, QAM, and MPSK). Finally, we study the statistics of the α-stable distribution. Those statistics are expressed in closed form in terms of the Fox H function and used to get the SER of BPSK, PAM, and QAM constellations. An approximation and an asymptotic expansion for high signal to noise ratio are presented also and their efficiency is proved using Monte Carlo simulations. It is worth mentioning that all the error rates presented in this work are averaged over a generalized flat fading, namely the extended generalized K, which has the ability to capture most of the known fading distribution. Many special cases are treated and simpler closed form expressions of the probability of error are derived and compared to some previous reported results.

京都大学 / 新制・課程博士 / 博士(情報学) / 甲第24740号 / 情博第828号 / 新制||情||139(附属図書館) / 京都大学大学院情報学研究科数理工学専攻 / (主査)教授 梅野 健, 教授 山下 信雄, 准教授 加嶋 健司 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM

No description available.

No description available.

La plupart des réseaux WSN utilisent une bande passante industrielle, scientifique et médicale (ISM) sans licence, qui crée un phénomène probable d'interférence sur un canal donné. Le débit du système est influencé par les interférences car il peut être encombré, causant des pertes de paquets, des retransmissions, une instabilité de liaison et un comportement de protocole incohérent. Dans la recherche sur les réseaux de capteurs sans fil, la simulation est l’une des approches essentielles pour évaluer un protocole de système ou de performance. La précision des résultats estimés dépend des paramètres de simulation sélectionnés. Dans les analyses existantes sur WSN, des modèles d'interférence simples sont utilisés dans les simulations. Cependant, ces modèles d'interférence ne sont pas assez précis pour l'analyse pratique d'applications de réseau de capteurs sans fil. De plus, la croissance rapide dans le domaine des réseaux WSN implique la nécessité de créer de nouveaux simulateurs dotés de capacités plus spécifiques pour lutter contre les effets de propagation par brouillage et par trajets multiples. La recherche a pour objectif principal de rechercher un environnement de simulation approprié permettant aux chercheurs de vérifier de nouvelles idées et de comparer les solutions futures. / Most of WSNs use unlicensed Industrial, Scientific and Medical (ISM) frequency band that makes interference probable phenomenon on a given channel.System throughput gets influenced by the interference as it can congest wireless medium, cause packet drops, re-transmissions, link instability, and inconsistent protocol behavior. In wireless sensor network research, simulation is one of the essential approaches to asses and evaluate system or protocol performance. The accuracy of estimated results depends on selected simulation parameters. In existing analysis on WSN, simple interference models are used in simulations. However, these interference models are not accurate enough for practical wireless sensor network applications analysis.Moreover, the rapid growth in the field of WSNs entails the need of creating new simulators that have more specific capabilities to tackle interference and multipath propagation effects.Finding a suitable simulation environment that allows researchers to verify new ideas and compare proposed future solutions is main task of this research.

Introdução: A curva de crescimento é um modelo empírico da evolução de uma quantidade ao longo do tempo. As curvas de crescimento são utilizadas em muitas disciplinas , em particular no domínio da estatística, onde há uma grande literatura sobre o assunto relacionado a modelos não lineares. Método:No desenvolvimento dessa dissertação de mestrado, foi realizado um estudo baseado em dados de crescimento nas áreas biológica e médica para comparar os dois tipos de inferência (Clássica e Bayesiana), na busca de melhores estimativas e resultados para modelos de regressão não lineares, especialmente considerando alguns modelos de crescimento introduzidos na literatura. No método Bayesiano para a modelagem não linear assumimos erros normais uma suposição usual e também distribuições estáveis para a variável resposta. Estudamos também alguns aspectos de robustez dos modelos de regressão não linear para a presença de outliers ou observações discordantes considerando o uso de distribuições estáveis para a resposta no lugar da suposição de normalidade habitual. Resultados e Conclusões: Análise dos dois exemplos pode-se observar melhores ajustes quando utilizada o método Bayesiano de ajustes de modelos não lineares de curvas de crescimento. É bem sabido que, em geral, não há nenhuma forma fechada para a função densidade de probabilidade de distribuições estáveis. No entanto, sob uma abordagem Bayesiana, a utilização de uma variável aleatória latente ou auxiliar proporciona uma simplificação para obter qualquer distribuição a posteriori quando relacionado com distribuições estáveis. Esses resultados poderiam ser de grande interesse para pesquisadores e profissionais, ao lidar com dados não Gauss. Para demonstrar a utilidade dos aspectos computacionais, a metodologia é aplicada a um exemplo relacionado com as curvas de crescimento intra-uterino para prematuros. Resumos a posteriori de interesse são obtidos utilizando métodos MCMC (Markov Chain Monte Carlo) e o software OpenBugs. / Introduction: The growth curve is an empirical model of the evolution of a quantity over time. Growth curves are used in many disciplines, particularly in the field of statistics, where there is a large literature on the subject related to nonlinear models. Method: In the development of this dissertation, a study based on data growth in biological areas and medical was conducted to compare two types of inferences (Classical and Bayesian), in search of better estimates and results for nonlinear regression models, especially considering some growth models introduced in the literature. The Bayesian method for nonlinear modeling assume normal errors an usual assumption and also stable distributions for the response variable. We also study some aspects of robustness of nonlinear regression models for the presence of outliers or discordant observations regarding the use of stable distributions to the response in place of the usual assumption of normality. Results and Conclusions: In the analysis of two examples it can be seen best results using Bayesian methodology for non linear models of growth curves. It is well known that, in general, there is no closed form for the probability density function of stable distributions. However, under a Bayesian approach, the use of a latent random variable or auxiliary variable provides a simplification to get every conditional posterior related to stable distributions. These results could be of great interest to researchers and practitioners when dealing with non-Gaussian data. To demonstrate the utility of the computational aspects, the methodology is also applied to an example related to intrauterine growth curves for premature infants. Posterior summaries of interest are obtained using MCMC methods (MCMC) and the OpenBugs software.