Made available in DSpace on 2014-12-17T15:14:49Z (GMT). No. of bitstreams: 1 AlzeyGF.pdf: 1374839 bytes, checksum: 04a6f9d10ed4cb9e18af438a89ba7f26 (MD5) Previous issue date: 2008-10-02 / Systems whose spectra are fractals or multifractals have received a lot of attention in recent years. The complete understanding of the behavior of many physical properties of these systems is still far from being complete because of the complexity of such systems. Thus, new applications and new methods of study of their spectra have been proposed and consequently a light has been thrown on their properties, enabling a better understanding of these systems. We present in this work initially the basic and necessary theoretical framework regarding the calculation of energy spectrum of elementary excitations in some systems, especially in quasiperiodic ones. Later we show, by using the Schr?odinger equation in tight-binding approximation, the results for the specific heat of electrons within the statistical mechanics of Boltzmann-Gibbs for one-dimensional quasiperiodic systems, growth by following the Fibonacci and Double Period rules. Structures of this type have already been exploited enough, however the use of non-extensive statistical mechanics proposed by Constantino Tsallis is well suited to systems that have a fractal profile, and therefore our main objective was to apply it to the calculation of thermodynamical quantities, by extending a little more the understanding of the properties of these systems. Accordingly, we calculate, analytical and numerically, the generalized specific heat of electrons in one-dimensional quasiperiodic systems (quasicrystals) generated by the Fibonacci and Double Period sequences. The electronic spectra were obtained by solving the Schr?odinger equation in the tight-binding approach. Numerical results are presented for the two types of systems with different values of the parameter of nonextensivity q / Sistemas cujos espectros s?o fractais ou multifractais t?m sido bastante estudados nos ?ltimos anos. O entendimento completo do comportamento de muitas propriedades f?sicas destes sistemas ainda est? longe de ser completamente efetivado devido ? complexidade dos pr?prios sistemas. Desta maneira, novas aplica??es e novos m?todos de estudo dos seus espectros t?m sido feitos, possibilitando uma melhor compreens?o acerca desses sistemas. Apresentamos neste trabalho de disserta??o inicialmente todo o arcabou?o te?rico b?sico e necess?rio no tocante ? obten??o dos espectros de energia de excita??es elementares em alguns sistemas, mais especificamente nos sistemas quasiperi?dicos. Posteriormente mostramos, usando a equa??o de Schrodinger na aproxima??o de liga??o forte, os resultados para o calor espec?fico de el?trons com a mec?nica estat?stica de Boltzmann-Gibbs para sistemas quasiperi?dicos unidimensionais tipo Fibonacci e Per?odo Duplo. Estruturas desse tipo j? foram bastante exploradas, no entanto o uso da mec?nica estat?stica n?o-extensiva proposta por Constantino Tsallis ? bem adequado para sistemas que apresentam de alguma forma um perfil fractal, e portanto nosso principal objetivo foi aplic?-la para o c?lculo de grandezas termodin?micas ampliando um pouco mais a compreens?o das propriedades desses sistemas. Neste sentido, calculamos anal?tica e numericamente o calor espec?fico generalizado de el?etrons em sistemas quasiperi?dicos unidimensionais (quasicristais) gerados pelas sequ?ncias de Fibonacci e Per?odo Duplo. Os espectros eletr?nicos foram obtidos fazendo-se uso tamb?m da equa??o de Schrodinger na aproxima??o de liga??o forte. Resultados num?ricos s?o apresentados para os dois tipos de sistemas com diferentes valores do par?metro de n?o-extensividade q

Ensemble learning is a multiple-classi&amp / #64257 / er machine learning approach which combines, produces collections and ensembles statistical classi&amp / #64257 / ers to build up more accurate classi&amp / #64257 / er than the individual classi&amp / #64257 / ers. Bagging, boosting and voting methods are the basic examples of ensemble learning. In this thesis, a novel boosting technique targeting to solve partial problems of AdaBoost, a well-known boosting algorithm, is proposed. The proposed systems &amp / #64257 / nd an elegant way of boosting a bunch of classi&amp / #64257 / ers successively to form a better classi&amp / #64257 / er than each ensembled classi&amp / #64257 / er. AdaBoost algorithm employs a greedy search over hypothesis space to &amp / #64257 / nd a good suboptimal solution. On the other hand, this work proposes an evolutionary search with genetic algorithms instead of greedy search. Empirical results show that classi&amp / #64257 / cation with boosted evolutionary computing outperforms AdaBoost in equivalent experimental environments.

CoordenaÃÃo de AperfeiÃoamento de NÃvel Superior / Modelar objetos e personagens virtuais pode ser uma tarefa tediosa e difÃcil em vÃrias situaÃÃes. Um sistema de modelagem ideal deve ser capaz de representar qualquer mudanÃa ou caracterÃstica desejada pelo usuÃrio da forma mais rÃpida e transparente possÃvel, abstraindo toda a complexidade topolÃgica e matemÃtica, permitindo ao modelador concentrar-se exclusivamente no ato de criar, em oposiÃÃo a ter de especificar a posiÃÃo de cada elemento de forma manual. Tal sistema, se por um lado pode oferecer menor controle na determinaÃÃo da posiÃÃo de cada vÃrtice, alÃm de poder ser mais caro computacionalmente em algumas situaÃÃes, por outro simplifica enormemente o trabalho do artista, tornando a modelagem acessÃvel a qualquer usuÃrio. O presente trabalho abordou o problema das tÃcnicas de sculpting tradicionais de adicionar detalhes em malhas triangulares, com resoluÃÃo adaptativa e de forma interativa, em regiÃes com baixa densidade de vÃrtices. O algoritmo proposto busca subdividir faces e arestas na regiÃo de interesse do artista de modo a garantir uma quantidade de vÃrtices suficiente para representar o detalhe pretendido, sem necessidade de qualquer informaÃÃo adicional por parte do usuÃrio. O sistema resultante à capaz de encurtar a distÃncia entre a modelagem tradicional e a virtual, superando alguns dos principais mÃtodos propostos no meio acadÃmico no que tange à localidade das subdivisÃes e a manutenÃÃo de uma malha de boa topologia e de boa qualidade geomÃtrica, com poucos vÃrtices de alta valÃncia, ainda que o modelo original seja fracamente refinado, tornando possÃvel a incorporaÃÃo de novos detalhes. / Modeling of virtual objects and characters can be tedious and difficult in many situations. A modeling system should be capable of representing any change or feature desired by the user as quickly and transparently as possible, allowing the modeler to focus on the act of creating, as opposed to having to specify the position of each vertex manually. Such a system, if on the one hand offers less control in determining the position of each vertex, being computationally more expensive in some situations, on the other hand greatly simplifies the work of the artist, allowing character modeling to be accessible to any user. This work addressed the problem of traditional sculpting techniques to add details on triangular meshes with adaptive resolution, in regions with low vertex density. The proposed algorithm splits faces and edges in the region of interest of the artist to ensure the existence of a sufficient number of vertices to represent the desired detail, without any additional input requested. The resulting system is capable of bridging the gap between traditional and virtual modeling, improving some of the main methods proposed in the literature regarding the location of the subdivisions. Moreover, it also helps to maintain a good mesh topology and geometric quality, with few vertices of high valence, even if the original mesh is poorly refined, allowing the incorporation of new details.

En el capítulo I, se examina el rendimiento de los activos financieros que representan las "mejores ideas" de los gestores de los fondos de inversión. Las inversiones para las que un gestor activo augura un buen rendimiento obtienen mejor retorno de mercado, asi como el resto de inversiones en sus carteras. En el capítulo II, se muestra explicitamente que los gestores que concentran sus carteras en un número reducido de activos, superan reiteradamente sus benchmarks y otros fondos más diversificados. Esta diferencia de rendimiento se puede explicar gracias a las diferencias en la exposición a factores de riesgo valorados por el mercado y al mayor talento de los gestores que se centran en invertir en activos de alta incertidumbre. En el capítulo III, se estudia la información contenida en las transacciones de activos y se muestra que las decisiones recientes de los gestores predicen el rendimiento futuro de las inversiones. Mientras que las compras llevadas a cabo por gestores con una habilidad superior se asocian a un rendimiento futuro anormalmente positivo, los gestores poco hábiles cometen errores de forma sistemática en la selección y en las transacciones de activos. / In Chapter I, we examine the performance of stocks that represent mutual fund managers' "best ideas". The stock that active managers display the most conviction towards ex-ante, significantly outperforms the market, as well as the other stocks in those managers' portfolios. In Chapter II, I explicitly show that managers, who concentrate their portfolios into a small number of stocks, consistently beat their benchmarks and their more diversified peers. This performance gap can be explained by differing portfolio exposures towards priced risk factors as well as stronger abilities of concentrated managers when investing in stocks with high uncertainty of information. In Chapter III, I study the information content of portfolio rebalances by mutual fund managers and show that their recent trading decisions predict future stock returns. While purchases by skilled managers are associated with positive future abnormal performance, unskilled managers systematically commit errors in the selection and trading of stocks.

Made available in DSpace on 2014-12-17T15:15:04Z (GMT). No. of bitstreams: 1 MarioAM.pdf: 1150858 bytes, checksum: 72544ace95c9eefd2f93d90922e6d6ec (MD5) Previous issue date: 2007-03-21 / Conselho Nacional de Desenvolvimento Cient?fico e Tecnol?gico / The magnetic order of bylayers composed by a ferromagnetic film (F) coupled with an antiferromagnetic film (AF) is studied. Piles of coupled monolayers describe the films and the interfilm coupling is described by an exchange interaction between the magnetic moments at the interface. The F has a cubic anisotropy while the AF has a uniaxial anisotropy. We analyze the effects of an external do magnetic field applied parallel to the interface. We consider the intralayer coupling is strong enough to keep parallel all moments of the monolayer an then they are described by one vector proportional to the magnetization of the layer. The interlayer coupling is represented by an exchange interaction between these vectors. The magnetic energy of the system is the sum of the exchange. Anisotropy and Zeeman energies and the equilibrium configuration is one that gives the absolute minimum of the total energy. The magnetization of the system is calculated and the influence of the external do field combined with the interfilm coupling and the unidirectional anisotropy is studied. Special attention is given to the region near of the transition fields. The torque equation is used to study dynamical behavior of these systems. We consider small oscillations around the equilibrium position and we negleet nonlinear terms to obtain the natural frequencies of the system. The dependence of the frequencies with the external do field and their behavior in the phase transition region is analized / O ordenamento de bicamadas magn?ticas compostas por um filme ferromagn?tico (F) fino acoplado com um filme antiferromagn?tico (AF) ? estudado, considerando que os filmes podem ser descritos por uma pilha de camadas monoat?micas (infinitas) acopladas. O acoplamento entre os filmes ? representado pela intera??o entre os spins das monocamadas da interface. Consideramos que o filme F tem uma anitropia c?bica e os momentos do filme AF sentem a presen?a de uma anisotropia uniaxial. Adicionalmente, os spins que comp?em as camadas da interface sentem a presen?a de uma anisotropia unidirecional. Os efeitos de um capo magn?tico est?tico aplicado paralelo ao plano da interface s?o analisados. Consideramos que a intera??o de troca entre os momentos magn?ticos de uma mesma monocamada ? suficientemente forte para mant?-los alinhados em uma mesma dire??o. Cada camada monoat?mica ? representada por um ?nico vetor (vetor magnetiza??o) proporcional ao momento magn?tico total da monocamada, e a intera??o entre estes vetores ? proporcional a intera??o efetiva entre as monocamadas. A energia magn?tica do sistema ? escrita como a soma das energias de troca, anisotropia e Zeeman, e as configura??es de equil?brio s?o obtidas atrav?s da minimiza??o da energia. A magnetiza??o do sistema ? calculada em fun??o do campo magn?tico dc aplicado e o efeito do meesmo, combinando com a intera??o interfilmes e anisotropia unidirecional, ? investrigado. A aten??o especial ? dada ?s regi?es de campo pr?ximas as transi??es de fase. Usamos a equa??o do torque para estudae o comportamento din?mico desses sistemas. Considerando que os vetores magnetiza??o oscilam em torno da posi??o de equil?brio, com amplitudes suficientemente pequenas para que os efeitos n?o lineares possam ser desprezados, calculamos as "freq??ncias naturais" do sistema dando especial aten??o as duas de menos valor. A depend?ncia das freq??ncias com a intensidade do campo ? determinada e o comportamento nas regi?es de transi??o de fase ? analisado

Made available in DSpace on 2014-12-17T15:47:49Z (GMT). No. of bitstreams: 1 NatalHC.pdf: 2764921 bytes, checksum: 846252ab05f6de9b65d1cf7bbe1aba09 (MD5) Previous issue date: 2008-08-04 / In this work we propose a technique that uses uncontrolled small format aerial images, or SFAI, and stereohotogrammetry techniques to construct georeferenced mosaics. Images are obtained using a simple digital camera coupled with a radio controlled (RC) helicopter. Techniques for removing common distortions are applied and the relative orientation of the models are recovered using projective geometry. Ground truth points are used to get absolute orientation, plus a definition of scale and a coordinate system which relates image measures to the ground. The mosaic is read into a GIS system, providing useful information to different types of users, such as researchers, governmental agencies, employees, fishermen and tourism enterprises. Results are reported, illustrating the applicability of the system. The main contribution is the generation of georeferenced mosaics using SFAIs, which have not yet broadly explored in cartography projects. The proposed architecture presents a viable and much less expensive solution, when compared to systems using controlled pictures / No presente trabalho, descrevemos uma metodologia para a constru??o de mosaicos georreferenciados a partir de imagens a?reas de pequeno formato, ou SFAI, consideradas n?o controladas. T?cnicas de Est?reo-Fotogrametria e de Vis?o Computacional s?o aplicadas para auxiliarem em resultados de qualidade e confiabilidade. As imagens s?o obtidas atrav?s de uma c?mera digital comum acoplada a um helic?ptero aeromodelo. T?cnicas para eliminar distor??es comuns s?o aplicadas e a orienta??o relativa dos modelos ? realizada baseada na geometria projetiva. Pontos reais do terreno s?o usados para executar a orienta??o absoluta, al?m de uma defini??o de escala e de um sistema de coordenadas que relaciona a imagem medida no terreno. O mosaico ser? manipulado em um SIG, fornecendo informa??o ?til a diferentes tipos de usu?rios, tais como pesquisadores, institui??es governamentais, pescadores e empresas de turismo. Resultados s?o gerados, ilustrando a aplicabilidade do sistema. A principal contribui??o ? a gera??o de mosaicos georreferenciados usando imagens a?reas de pequeno formato, o que tem sido pouqu?ssimo explorado em projetos de cartografia. A arquitetura proposta apresenta uma solu??o vi?vel e bem barata quando comparada com sistemas que utilizam imagens controladas