The role of randomness as a generator of long range correlations and ordinary statistical mechanics is investigated in this Dissertation. The difficulties about the derivation of thermodynamics from mechanics are pointed out and the connection between the ordinary fluctuation-dissipation process and possible anomalous properties of statistical systems is highlighted.

How can fluctuations in one-dimensional time series data be characterized and how can detected effects be decomposed into their dynamical origins or causes? In the context of these questions, a variety of problems are discussed and solutions are introduced. The first issue concerns the causes of anomalous diffusion. A previously proposed framework decomposes the Hurst exponent into the Joseph, Noah, and Moses effects. They represent violations of the three premises of the central limit theorem. Here the framework is applied to an intermittent deterministic system, which exhibits a rich combination of all three effects. Nevertheless, the results provide an intuitive interpretation of the dynamics. In addition, the framework is theoretically discussed and connected to a calculation that proves its validity for a large class of systems. Once the type of anomalous statistical behavior is classified, one might ask what the dynamical origin of the effects is. Especially the property of long range temporal correlations (the Joseph effect) is discussed in detail. In measurements, they might arise from different dynamical origins or can be explained as an emerging phenomenon. A collection of different routes to the observed behavior is established here. A popular tool for detecting long range correlations is detrended fluctuation analysis. Its advantages over traditional methods are stability and smoothness for timescales up to one fourth of the measurement time and the ability to neglect the slow dynamics and trends. Recently, a theory for an analytical understanding of this method was introduced. In this thesis, the method is further analyzed and developed. An approach is presented that enables scientists to use this method for short range correlated data, even if the dynamics is very complex. Fluctuations can be decomposed into a superposition of linear models that explain its features. Therefore, on the one hand, this thesis is about understanding the effects of anomalous diffusion. On the other hand, it is about widening the applicability of one of its detection methods such that it becomes useful for understanding normal or complex statistical behavior. A good example of a complex system, where the proposed stochastic methods are useful, is the atmosphere. Here it is shown how detrended fluctuation analysis can be used to uncover oscillatory modes and determine their periods. One of them is the El Ni\~no southern oscillation. A less well known and more challenging application is a 7--8 year mode in European temperature fluctuations. A power grid is a very different type of complex system. However, using the new method, it is possible to generate a data model that incorporates the important features of the grid frequency.

Submitted by (ana.araujo@ufrpe.br) on 2016-08-04T14:28:53Z No. of bitstreams: 1 Maira de Oliveira Santos.pdf: 1516572 bytes, checksum: ea3508c7d99ef42591a6b17f459901e0 (MD5) / Made available in DSpace on 2016-08-04T14:28:53Z (GMT). No. of bitstreams: 1 Maira de Oliveira Santos.pdf: 1516572 bytes, checksum: ea3508c7d99ef42591a6b17f459901e0 (MD5) Previous issue date: 2010-06-07 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / The study of climate has great economic end environmental importance, given that a single large and unexpected variation of a climatic element may devastate plantations or cities, and thus affect the economy of a region and life of the inhabitants. Climate can be influenced by diverse factors, such as latitude, altitude, air mass, proximity to sea, sea currents, terrain topology, vegetation, etc. The most important climate elements are temperature, humidity, atmospheric pressure, solar radiation, precipitation and wind. The wind is generated by atmospheric air mass movement, and may influence various phenomena such as soil erosion, pollutant dispersal, transport of pollen and seeds, propagation of diseases, as well as generation of eolic energy. Surface wind velocity is natural example of the phenomenon of turbulence, which represents a stochastic process characterized by temporal and spatial scale invariance. In this work we study long range correlations in temporal series of wind speed and direction registered at four meteorological stations in the cities of Arcoverde, Cabrobro, Garanhuns and Petrolina, in the state of Pernambuco, Brazil. To this end we apply Detrended Fluctuation Analysis (DFA) which was developed for quantification of long range correlations in non-stationary temporal series. We analyze the original wind speed series together with volatility (absolute value of increments) of the wind direction. All the analyzed series exhibit persistent long range correlations with the scale exponent above 0.5. In all cases the exponent values were found to be lower for wind direction then those for wind speed, indicating weaker persistence. No correlation was detected between the exponent values and the geographic parameters: longitutde, latitude and altitude of the station. The results of these analyses contribute to a better understanding of the nature of stochastic processes governing wind dynamics, necessary for development of more realistic theoretical and computational models as a base for modeling diverse phenomena influenced by climatic conditions. / O estudo do clima e dos seus elementos tem grande importância econômica e ambiental, visto que uma grande e inesperada variação em ao menos um dos elementos do clima pode devastar plantações, cidades, e assim mudar a economia de uma região e a vida das pessoas que ali habitam. O clima pode ser influenciado por diversos fatores, tais como latitude, altitude, massas do ar, continentalidade, maritmidade, correntes marítimas, relevo, vegetação, etc. Os elementos mais importantes do clima são temperatura, umidade, pressão atmosférica, radiação solar, precipitação e vento. O vento é gerado pelo movimento de massas do ar na atmosfera e pode influenciar vários fenômenos, como erosão do solo, dispersão de poluentes, transporte de pólen e sementes, propagação de doenças e geração da energia eólica. A velocidade do vento na superfície é um exemplo natural do fenômeno de turbulência, que representa um processo estocástico caracterizado pela invariância de escala temporal e espacial. Neste trabalho foram estudadas as correlações de longo alcance em séries temporais da velocidade e direção do vento registradas em quatro estações meteorológicas, nas cidades Arcoverde, Cabrobó, Garanhuns e Petrolina em Pernambuco. Foi utilizado o método Detrended fluctuation analysis (DFA), desenvolvido para quantificar as correlações de longo alcance em séries temporais não estacionárias. Foram analisadas as séries originais da velocidade do vento e as séries dos valores absolutos dos incrementos (volatilidade) da direção do vento. Todas as séries analisadas possuem as correlações de longo alcance persistentes, com expoente de escala acima de 0,5. Em todos os casos os valores dos expoentes são menores para a direção do que para a velocidade do vento, indicando que a persistência é mais fraca para a direção do vento. Não foi detectada a correlação entre os valores dos expoentes de escala e os parâmetros geográficos: longitude, latitude e altitude da estação. Os resultados destas análises vão ajudar a entender melhor a natureza dos processos estocásticos geradores da dinâmica do vento. Este entendimento é necessário para desenvolvimento dos modelos teóricos e computacionais mais precisos cujos resultados servirão como base para modelagem dos vários fenômenos influenciados pelas condições climáticas.

Submitted by (ana.araujo@ufrpe.br) on 2016-08-12T14:38:54Z No. of bitstreams: 1 Rosilda Benecio de Souza.pdf: 1979743 bytes, checksum: 6e2df03119bd441a84297bc73d2befeb (MD5) / Made available in DSpace on 2016-08-12T14:38:54Z (GMT). No. of bitstreams: 1 Rosilda Benecio de Souza.pdf: 1979743 bytes, checksum: 6e2df03119bd441a84297bc73d2befeb (MD5) Previous issue date: 2011-02-17 / Conselho Nacional de Pesquisa e Desenvolvimento Científico e Tecnológico - CNPq / Vegetation and forest fires affect millions of hectares of Brazilian land and have severe ecological, social and economic consequences, including emissions of green house gases, loss of biodiversity, soil erosion etc. To establish efficient methods for prevention and suppression of fires, which is crucial for preservation of environment, it is necessarily to know the spatial location and time of occurrence of fires, burned area, why they occur, and how they initiate and propagate. Several satellite systems are currently available for monitoring different fire characteristics: dry areas that are susceptible to fire, actively flaming fires, burned area and smoke, and trace gas emissions. Hot pixels are satellite image pixels with infrared intensity corresponding to burning vegetation. Depending of image resolution, a hot pixel may represent one fire, or a part of a larger fire. Together with other satellite data, the number of hot pixels can be used to estimate the burned area and predict environmental and economics consequences. In this work we study the dynamics of hot pixels detected in Brazil by satellite NOAA-12 during the period 1998-2007, using the method Multifractal Detrended Fluctuation Analysis, which serves to detect and quantify multifractal properties of non-stationary temporal series. We calculate the generalized Hurst exponent h(q), Renyi exponent (q) and singularity spectrum f( ). The results show the existence of power-law long-term correlations that are described by a hierarchy of scaling exponents, which is the consequence of an underlying multifractal stochastic process. Based on this empirical result we also show that the Multifractal Cascade Model can be used to produce synthetic data for hot pixels dynamics. The observed multifractal property of temporal series of hot pixels should be incorporated in theoretical models and computer simulations of the fire dynamics and related phenomena. / Queimadas e incêndios florestais atingem milhões de hectares de terras brasileiras, causando graves consequências ecológicas, sociais e econômicas, incluindo emissões de gases do efeito estufa, perda de biodiversidade, erosão do solo, etc. Para estabelecer métodos eficientes para prevenção e supressão do fogo, importantes para proteção do meio ambiente, é necessário conhecer onde, quando e porquê os incêndios ocorrem, a área queimada e como se iniciam e se propagam. Atualmente, vários satélites são disponibilizados para monitoramento das características do fogo: áreas de risco, incêndios atualmente ativos, área queimada, fumaça e emissão de gases. Focos de calor são pixels na imagem de satélite, com intensidade infravermelha correspondente a vegetação queimada. Dependendo da resolução, um foco pode representar uma queimada ou parte de um incêndio maior. O número de focos combinado a outras informações fornecidas pelos satélites pode ser usado para estimar a área queimada, e prever as consequências ecológicas e econômicas. Neste trabalho, foi estudada a dinâmica de focos de calor detectados no Brasil pelo satélite NOAA-12, durante o período 1998-2007, utilizando o método Multifractal Detrended Fluctuation Analysis, desenvolvido para detecção e quantificação das propriedades mutifractais das séries temporais não estacionárias. Foram calculados o expoente generalizado de Hurst h(q), o expoente Renyi (q) e o espectro de singularidade f( ). Os resultados mostraram a existência de correlações de longo alcance, caracterizadas por uma hierarquia dos expoentes de escala, conseqüência de um processo estocástico multifractal. Baseado nos resultados empíricos, também foram mostrados que o Multifractal Cascade Model pode ser usado para gerar séries artificiais dos focos de calor. A propriedade multifractal da dinâmica dos focos de calor poderá ser incorporada em modelos teóricos e simulações computacionais de dinâmica de incêndios e fenômenos relacionados.

Submitted by (ana.araujo@ufrpe.br) on 2016-08-12T15:41:19Z No. of bitstreams: 1 Silvio Fernando Alves Xavier Junior.pdf: 1969278 bytes, checksum: 716d98cbb2937cd01be0ff1272ed5033 (MD5) / Made available in DSpace on 2016-08-12T15:41:19Z (GMT). No. of bitstreams: 1 Silvio Fernando Alves Xavier Junior.pdf: 1969278 bytes, checksum: 716d98cbb2937cd01be0ff1272ed5033 (MD5) Previous issue date: 2011-06-29 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / Rainfall can be understood as an end product of a number of complex atmospheric processes, which vary in space and time, and it may be considered one of most important dominant factor of the meteorological-climatic features of an specified investigated area. In this study, we observed if the dynamics of rain in Piracicaba, São Paulo - Brazil is generated by a multifractal process and / or belongs to classes of Self-Organized Criticality systems. To detect long-term correlations and multifractal behavior, we apply MF-DFA method that systematically detect nonstationarities and overcome trends in the data at all timescales. We calculated the generalized Hurst exponent, h(q), and Renyi exponent, (q). The results showed the existence of power-law long-term correlations which are described by a hierarchy of scaling exponents, that is the consequence of an underlying multifractal stochastic process. For smaller scales of about 8 months, the dynamics of rain is generated by a multifractal process (the generalized Hurst exponent, h(q), decreases with the increase in order (q) meaning it can be modeled using the cascade models. For larger scales, the value of h(q) is between 0:35 �� 0:55 indicating a weaker multifractality. The hypothesis that rainfall may be a case of Self-Organized Criticality is assessed. We analyze two events: the daily amount of rain and drought events (days without rain), both are weather phenomena that are strongly linked to rainfall. It appears that the distribution of the daily amount of rain displays two different scaling regimes for small and large intensities. The value of the ratio of these exponents confirms the results that were obtained in regions with tropical and subtropical climates. However, for the distribution of drought events we find two distinct scaling exponents with values that are closer than those observed in the daily amount of rain. The multifractal properties and self-organized criticality should be incorporated into theoretical models and computer simulations of the dynamics of rainfall and related phenomena.Rainfall can be understood as an end product of a number of complex atmospheric processes, which vary in space and time, and it may be considered one of most important dominant factor of the meteorological-climatic features of an specified investigated area. In this study, we observed if the dynamics of rain in Piracicaba, São Paulo - Brazil is generated by a multifractal process and / or belongs to classes of Self-Organized Criticality systems. To detect long-term correlations and multifractal behavior, we apply MF-DFA method that systematically detect nonstationarities and overcome trends in the data at all timescales. We calculated the generalized Hurst exponent, h(q), and Renyi exponent, (q). The results showed the existence of power-law long-term correlations which are described by a hierarchy of scaling exponents, that is the consequence of an underlying multifractal stochastic process. For smaller scales of about 8 months, the dynamics of rain is generated by a multifractal process (the generalized Hurst exponent, h(q), decreases with the increase in order (q) meaning it can be modeled using the cascade models. For larger scales, the value of h(q) is between 0:35 �� 0:55 indicating a weaker multifractality. The hypothesis that rainfall may be a case of Self-Organized Criticality is assessed. We analyze two events: the daily amount of rain and drought events (days without rain), both are weather phenomena that are strongly linked to rainfall. It appears that the distribution of the daily amount of rain displays two different scaling regimes for small and large intensities. The value of the ratio of these exponents confirms the results that were obtained in regions with tropical and subtropical climates. However, for the distribution of drought events we find two distinct scaling exponents with values that are closer than those observed in the daily amount of rain. The multifractal properties and self-organized criticality should be incorporated into theoretical models and computer simulations of the dynamics of rainfall and related phenomena. / A precipitação pode ser entendida como um produto final de processos atmosféricos complexos, os quais variam no tempo e espaço, e pode ser considerada um dos mais importantes fatores dominante das características meteorológicas-climáticas de uma determinada área investigada. Neste trabalho, verificamos se a dinâmica da chuva em Piracicaba, São Paulo - Brasil é gerada por um processo multifractal e/ou pertence as classes dos sistemas com propriedade da criticalidade auto-organizada. Para detectar a correlação de longo alcance e o comportamento multifractal, aplicamos o método MF-DFA que sistematicamente detecta não-estacionariedades e tendências nos dados para todas escalas de tempo. Calculamos o expoente generalizado de Hurst, h(q), e o expoente de Renyi, (q). Os resultados mostraram a existência de correlações de longo alcance, caracterizadas por uma hierarquia dos expoentes de escala, consequência de um processo estocástico multifractal. Para as escalas menores, aproximadamente 8 meses, a dinâmica de chuva é gerada por um processo multifractal (o expoente de Hurst generalizado, h(q), diminui com o aumento de ordem q) significando que pode ser modelada utilizando os modelos de cascata. Para as escalas maiores, o valor de h(q) está entre 0,35-0,55 o que indica a multifractalidade mais fraca. A hipótese de que a precipitação pode ser um caso de Self- Organized Criticality é avaliada. Analisamos dois eventos: a quantidade diária de chuva e eventos de seca (dias sem chuva), ambos são fenômenos metereológicos os quais são fortemente ligados à precipitação. Verifica-se que a distribuição da quantidade diária de chuva exibe dois regimes de escala distintos para pequenas e grandes quantidades. O valor da razão desses expoentes encontrados confirmam os resultados que foram obtidos nas regiões com climas tropical e subtropical. No entanto, para a distribuição de eventos de seca encontramos dois expoentes de escala distintos com valores bem mais próximos comparados com os observados na quantidade diária de chuva. As propriedades multifractais e criticalidade auto-organizada deverão ser incorporados em modelos teóricos e simulações computacionais da dinâmica das chuvas e fenômenos relacionados.

La variabilité des mouvements humains est caractérisée par la présence de corrélations à long-terme, ou fluctuations fractales. Cette propriété est associée à des états sains et optimaux, tandis que les états non-optimaux sont associés avec une perte des corrélations à long-terme, devenant plus périodique ou plus aléatoire. Les métronomes isochrones sont largement utilisés pour guider le pas dans des protocoles de réhabilitation de la marche, mais leur utilisation modifie la dynamique des séries de pas qui ne présentent plus de corrélations à long-terme (persistantes) mais deviennent anti-persistante (i.e., corrélations négative). Des hypothèses récentes suggèrent que la synchronisation avec un environnement fractal pourrait induire un appariement de la structure temporelle de l'organisme avec la structure temporelle de l'environnement. L'objectif de cette thèse était de tester des stratégies de synchronisation alternatives préservant la nature fractale des séries temporelles. Différentes expérimentations ont été mises en places, impliquant des coordinations interpersonnelles, de la synchronisation avec des métronomes fractals et du « guidage humain ». De manière générale, nos résultats montrent que les séries comportementales des participants étaient corrélée à celle de l'environnement seulement si celui-ci présente des fluctuations fractales. Les résultats de nos modélisations suggèrent également que les métronomes isochrones et non-isochrones impliquent des réactions comportementales fondamentalement différentes. Nos résultats présentent des perspectives cliniques puisque l'élaboration de protocoles de réhabilitation de la marche utilisant des environnements fractals pourrait permettre de préserver les corrélations à long-terme, marqueurs d'adaptabilité du comportement. / Human movements variability is characterized by the presence of long-range (fractal) correlations. This feature is associated with optimal, healthy states while non-optimal states are associated with a loss of long-range correlations, toward more periodicity or more randomness. Isochronous pacing is widely used for gait rehabilitation, but changes the stride time dynamics from persistent long-range correlations to anti-persistent (negative) correlations. It has been recently argued that synchronization with fractal environment could induce a matching between the organism structure and the environmental structure. The aim of this thesis was to test alternatives pacing strategies preserving the fractal nature of stride time series. Different sets of experiments were run, involving interpersonal coordination, synchronization with non-isochronous metronomes and “human pacing”. Overall our results show that the time series produced by participants were correlated to those of the environment only if the environment presented fractal fluctuations. Our models suggest that isochronous and non-isochronous metronomes imply fundamental different behaviours. Our results have clinical perspectives because the use of fractal environment in rehabilitation protocols could help to preserve long-range correlations, a hallmark of behavioural adaptability.

Ce travail de thèse s'inscrit dans le cadre du développement de la méthode de mélange de configurations multiparticules-multitrous visant à décrire les propriétés de structure des noyaux atomiques. Basée sur un double principe variationnel, cette approche permet de déterminer simultanément les coefficients d'expansion de la fonction d'onde et les orbitales individuelles.Dans ce manuscrit, le formalisme complet méthode de mélange de configurations multiparticules-multitrous auto-cohérente est pour la première fois appliqué à la description de quelques noyaux des couches p et sd, avec l'interaction de Gogny D1S.Un première étude du 12C est effectuée afin de tester et comparer le double processus de convergence lorsque différents types de critères sont appliqués pour sélectionner les configurations à N-corps inclues dans la fonction d'onde du noyau. Une analyse détaillée de l'effet induit par l'optimisation des orbitales est conduite. En particulier, son impact sur la densité à un corps et sur la fragmentation de la fonction d'onde de l'état fondamental, est analysé.Une étude systématique de noyaux de la couche sd est ensuite conduite. Une analyse précise du contenu en corrélation de l'état fondamental est effectuée, et quelques quantités observables telles que les énergies de liaison et de séparation, ainsi que les rayons de charge, sont calculées et comparées à l'expérience. Les résultats obtenus sont satisfaisants. La spectroscopie de basse énergie est ensuite étudiée. Les énergies d'excitation théoriques sont en très bon accord avec les données expérimentales, et les caractéristiques dipolaires magnétiques sont également satisfaisantes. Les propriétés quadripolaires électriques, et en particulier les probabilités de transition B(E2), sont par contre largement sous-estimée par rapport aux valeurs expérimentales, et révèle un manque important de collectivité dans la fonction d'onde, dû à l'espace de valence restreint considéré. Si la renormalisation des orbitales induit une importante fragmentation de la fonction d'onde de l'état fondamental, seul un effet très faible est obtenu sur les probabilités de transition B(E2). Une tentative d'explication est donnée.Enfin, les informations de structure fournies par la méthode de mélange de configurations multiparticules-multitrous sont utilisées comme ingrédient de base pour des calculs de réactions telles que la diffusion inélastique de protons et d'électrons sur noyaux de la couche sd. Si les résultats révèlent aussi un manque de collectivité, les tendances expérimentales sont bien reproduites et sont améliorées par l'optimisation des orbitales. / This thesis project takes part in the development of the multiparticle-multihole configuration mixing method aiming to describe the structure of atomic nuclei. Based on a double variational principle, this approach allows to determine the expansion coefficients of the wave function and the single-particle states at the same time. In this work we apply for the first time the fully self-consistent formalism of the mp-mh method to the description of a few p- and sd-shell nuclei, using the D1S Gogny interaction.A first study of the 12C nucleus is performed in order to test the doubly iterative convergence procedure when different types of truncation criteria are applied to select the many-body configurations included in the wave-function. A detailed analysis of the effect caused by the orbital optimization is conducted. In particular, its impact on the one-body density and on the fragmentation of the ground state wave function is analyzed.A systematic study of sd-shell nuclei is then performed. A careful analysis of the correlation content of the ground state is first conducted and observables quantities such as binding and separation energies, as well as charge radii are calculated and compared to experimental data. Satisfactory results are found. Spectroscopic properties are also studied. Excitation energies of low-lying states are found in very good agreement with experiment, and the study of magnetic dipole features are also satisfactory. Calculation of electric quadrupole properties, and in particular transition probabilities B(E2), however reveal a clear lack of collectivity of the wave function, due to the reduced valence space used to select the many-body configurations. Although the renormalization of orbitals leads to an important fragmentation of the ground state wave function, only little effect is observed on B(E2) probabilities. A tentative explanation is given.Finally, the structure description of nuclei provided by the multiparticle-multihole configuration mixing method is utilized to study reaction mechanisms such as electron and proton inelastic scattering on sd-shell nuclei. Although the results also suffer from the lack of collectivity, the experimental trends are well reproduced and improved by the orbital optimization.

Submitted by (ana.araujo@ufrpe.br) on 2016-07-05T15:38:42Z No. of bitstreams: 1 Erinaldo Leite Batista Almeida.pdf: 3620819 bytes, checksum: b2532ef7524f47d5417d01445fec797b (MD5) / Made available in DSpace on 2016-07-05T15:38:42Z (GMT). No. of bitstreams: 1 Erinaldo Leite Batista Almeida.pdf: 3620819 bytes, checksum: b2532ef7524f47d5417d01445fec797b (MD5) Previous issue date: 2009-08-10 / Financial markets are complex systems that contain large numbers of interacting units, including interactions among various units in the same market and interactions between units in different markets. Various methods of economics, statistics and econophysics have been developed to analyze financial temporal series (such as price returns, share volume, number of transactions), and serve to establish theoretical models for underlying stochastic processes. The availability of financial data on the internet and increasing computational power have enabled researchers to conduct a large number of empirical studies on financial markets. These studies have shown some universal properties: the risk function of price returns is scale invariant, with power-law behavior and similar value of exponent for different markets; the absolute values of returns (volatility) exhibit long-range power-law correlations. In this work, we use methods if econophysics to study the statistical properties of Brazilian financial markets. We analyze and compare scale properties of risk functions and correlations in temporal series of price returns of agricultural commodities and stocks of various companies traded at Bovespa. We analyze the daily prices of five commodities and twenty stocks traded in the period 2000-2008. For both commodities and stocks, the risk function of daily price returns shows powerlaw behavior with the exponent outside the Levy stable region. The values of exponents are higher for stocks than for commodities. We use Detrended Fluctuation Analysis (DFA) to study correlations in daily time series of absolute values of returns (volatility). This method was developed to quantify long range correlations in non-stationary temporal series.All analyzed series show persistent behavior, meaning that large (small) values are more likely to be followed with large (small) values. The value of the DFA exponent is higher for commodities than for stocks. We also use Detrended Cross Correlation Analysis (DCCA) to study cross-correlations between two series. The values of DCCA exponents are above 0.5 for all series, indicating the existence of long range cross-correlations. This means that each stock or commodity has long memory of its own previous values and of previous values of other stocks or commodities studied. These results are in agreement with results obtained for American financial markets. / Mercados financeiros são caracterizados por um grande número de unidades e interações complexas, incluindo as interações internas (entre diferentes elementos de um mercado) e fatores externos (influência de outros mercados). Vários métodos de economia, estatística e recentemente econofísica foram desenvolvidos para analisar as séries temporais de variáveis financeiras (retorno de preços de ações, mercadorias e taxas de cambio, índice de mercado, volume de negociação, etc.), com objetivo de estabelecer os modelos teóricos para processos estocásticos que estão em base desses fenômenos. A disponibilidade de dados financeiros de vários mercados e crescente poder computacional resultaram em um grande número de estudos empíricos cujos resultados mostraram algumas propriedades universais: a função risco de retornos de preços segue uma lei de potência com o valor de expoente similar para os vários mercados; os valores absolutos de retornos possuem correlações de longo alcance. Neste trabalho foram usados os métodos de econofísica para estudar as propriedades estatísticas do mercado financeiro brasileiro. Foram analisadas e comparadas as propriedades de escala de função risco e de correlações em séries temporais de retornos de preços de mercadorias agrícolas e preços de ações de várias empresas negociadas na Bolsa de Valores de São Paulo (BOVESPA). Foram analisados os preços diários de cinco mercadorias: açúcar, algodão, café, soja e boi, registrados em período 2000-2008. Para ações, analisamos as características seguintes: preços de abertura, fechamento, valores máximo e mínimo, volume e montante. Todas as séries são diárias, registradas no período de 2000-2008. São estudadas 20 empresas divididas em 4 grupos: bancos, energia, telecomunicações e siderurgia (5 empresas de cada grupo). Para todas as séries estudadas a função risco de retornos de preços segue uma lei de potência com os valores de expoente maiores para ações do que para mercadorias. As correlações são analisadas para os valores absolutos de retornos de preços (volatilidade). Foi usado o método Detrended Fluctuation Analysis (DFA), desenvolvido para quantificar as correlações de longo alcance em séries temporais não estacionárias. Todas as séries mostraram um comportamento persistente, significando que os valores grandes (pequenos) tem maior probabilidade de serem seguidos por valores grandes (pequenos). Os valores de expoente DFA são maiores para mercadorias do que para as ações. Foi utilizada uma generalização de DFA, Detrended Cross Correlation Analysis (DCCA) para analisar as correlações cruzadas entre duas séries. Os valores de expoente DCCA para todas as séries estudadas indicam a existência de correlações cruzadas de longo alcance significando que os valores de cada série possuem memória de longo alcance de seus valores anteriores e também de valores anteriores de outras série. Os resultados estão em acordo com os resultados obtidos para mercado americano.

Submitted by (ana.araujo@ufrpe.br) on 2016-08-02T15:32:53Z No. of bitstreams: 1 Luciano Rodrigues da Silva.pdf: 1477739 bytes, checksum: e1ea61981eacbff2c9319865f5504f91 (MD5) / Made available in DSpace on 2016-08-02T15:32:53Z (GMT). No. of bitstreams: 1 Luciano Rodrigues da Silva.pdf: 1477739 bytes, checksum: e1ea61981eacbff2c9319865f5504f91 (MD5) Previous issue date: 2009-10-20 / Vegetation fires represent a natural hazard with severe ecological, social, health and economic consequences. Every year fires burn millions of hectares of forest worldwide and their number have been increasing, principally because of the increase in population and combustion material. The preservation of the environment depends on global and regional policies and methods of prevention and suppression of fires. To establish these methods it is necessarily to know the profile of fires: spatial location, time of occurrence, burned area, why they occur, and how they initiate and propagate. Recently, various methods of Statistical Physics (including data analysis and computational models) have been applied to provide additional information about spatial and temporal distribution of fire sequences, which is crucial for assessing various consequences of burning, such as emissions of gasses and particulates to the atmosphere, loss of biodiversity, loss of wildlife habitat, soil erosion etc. Several satellite systems (with different capabilities in terms of spatial resolution, sensitivity, spectral bands, and times and frequency of overpasses) are currently available for monitoring different fire characteristics: dry areas that are susceptible to wild fire outbreak, actively flaming fires, burned area and smoke, and trace gas emissions. Hotspots are satellite image pixels with infrared intensity corresponding to burning vegetation. A hotspot may represent one fire, or be one of several hotspots representing a larger fire. Together with other satellite data, thenumber of hot-spots can be used to estimate the burned area. In this work we study the dynamics of hotspots using the Detrended Fluctuation Analysis (DFA) method, which serves to quantify correlations in non stationary time series. We analyze daily hotspot temporal series detected in Brazil by various satellites during the period 1998-2008. The results show the existence of power-law long-range correlations that represent an important property of the underlying stochastic process. This property, also found in climatic phenomena, should be incorporated in theoretical models and computer simulations of the fire dynamics. / Incêndios em vegetação é um tipo de desastre natural com conseqüências ambientais, sociais econômicas, etc. Todos os anos incêndios destroem milhões de hectares das florestas e aumentam em número como conseqüência de vários fatores, principalmente de crescimento populacional e acúmulo de material combustível. A preservação de meio ambiente depende das políticas protecionistas globais e regionais adequadas às características de cada região. Para estabelecer essas políticas de controle e prevenção é necessário conhecer o perfil dos incêndios florestais: onde, quando e porque ocorrem. Além das estatísticas de ocorrências de incêndios os métodos emergentes da Física Estatística incluindo análise de dados e modelos computacionais, providenciam as informações adicionais sobre a distribuição e agrupamento espaço-temporal dos incêndios, que são cruciais para o estudo de várias conseqüências de fogo, como emissão de gases e partículas em atmosfera, perda de biodiversidade, erosão de solo, etc. Vários satélites (com características diferentes em termos de resolução espacial, bandas espectrais, tempo e freqüência de escaneamento) são disponíveis para monitoramento das varias características de fogos: áreas de risco, incêndios atualmente ativos, área queimada, fumaça, emissão de poluentes etc. Focos de calor são pixels na imagem de satélite com intensidade infravermelha correspondente a vegetação queimada. Um foco pode representar uma queimada, parte de um incêndio maior ou outras fontes de calor como, por exemplo, a reflexão de luz da superfície de um lago. O número de focos junto com outras informações providenciadas pelos satélites podem ser usados para estimar a área queimada, para detecção e monitoramento dos incêndios florestais, estimação de risco de fogo, e para avaliação da influencia de outros fatores ambientais. Neste trabalho estudamos a dinâmica de focos de calor no Brasil usando o método Detrended Fluctuation Analysis (DFA), desenvolvido para quantificar as correlações em séries temporais não estacionárias. Analisamos séries temporais diárias de focos de calor detectados no Brasil pelo vários satélites, durante o período 1998-2008. Os resultados mostram a existência de correlações de longo alcance persistentes, que representa uma propriedade importante dos processos estocásticos geradores desse fenômeno. Esta propriedade, também presente em fenômenos climáticos deveria ser incorporada em modelos teóricos e simulações computacionais de dinâmica de incêndios.

A commonly used tool in data analysis is to compute a sample mean. Assuming a uni-modal distribution, its mean provides valuable information about which value is typically found in an observation. Also, it is one of the simplest and therefore very robust statistics to compute and suffers much less from sampling effects of tails of the distribution than estimates of higher moments. In the context of a time series, the sample mean is a time average. Due to correla- tions among successive data points, the information stored in a time series might be much less than the information stored in a sample of independently drawn data points of equal size, since correlation always implies redundancy. Hence, the issue of how close the sample estimate of a time average is to the true mean value of the process depends on correlations in data. In this thesis, we will study the proba- bility that a single time average deviates by more than some threshold value from the true process mean. This will be called the Large Deviation Probability (LDP), and it will be a function of the time interval over which the average is taken: The longer the time interval, the smaller will this probability be. However, it is the precise functional form of this decay which will be in the focus of this thesis. The LDP is proven to decay exponentially for identically independently distributed data. On the other hand we will see in this thesis that this result does not apply to long-range correlated data. The LDP is found to decay slower than exponential for such data. It will be shown that for intermittent series this exponential decay breaks down severely and the LDP is a power law. These findings are outlined in the methodological explanations in chapter 3, after an overview of the theoretical background in chapter 2. In chapter 4, the theoretical and numerical results for the studied models in chapter 3 are compared to two types of empirical data sets which are both known to be long- range correlated in the literature. The earth surface temperature of two stations of two climatic zones are modelled and the error bars for the finite time averages are estimated. Knowing that the data is long-range correlated by estimating the scaling exponent of the so called fluctuation function, the LDP estimation leads to noticeably enlarged error bars of time averages, based on the results in chapter 3. The same analysis is applied on heart inter-beat data in chapter 5. The contra- diction to the classical large deviation principle is even more severe in this case, induced by the long-range correlations and additional inherent non-stationarity. It will be shown that the inter-beat intervals can be well modeled by bounded fractional Brownian motion. The theoretical and numerical LDP, both for the model and the data, surprisingly indicates no clear decay of LDP for the time scales under study.