Muitos sistemas complexos naturais ou construídos pelos seres humanos podem ser representados por redes complexas, uma teoria que une o estudo de grafos com a mecânica estatística. Esse tipo de representação, porém, pode ser comprometido pela maneira como os dados são obtidos. Em geral, os dados utilizados para representar tais sistemas nem sempre são precisos ou completos e correspondem a apenas amostras pequenas de redes maiores, como é o caso da teia mundial (WWW). Dessa forma, mesmo que as amostras sejam grandes, as suas propriedades são diretamente afetadas pela maneira como elas são obtidas e podem não corresponder com as de suas respectivas redes originais. Por exemplo, a amostragem mais utilizada para captura de roteadores da Internet, se empregada em redes aleatórias, tende a obter redes sem escala como resultado. Em contrapartida, amostras de redes sem escala não têm garantia de preservar essa estrutura. Por causa desses e outros problemas que possam ocorrer na amostragem das redes, é muito importante avaliar a variação das propriedades das redes a ruídos (para saber quais variam menos, sendo, portanto, mais adequadas para caracterizar redes com problemas de amostragem) e os efeitos da amostragem na caracterização, classificação e análise de redes complexas (pois redes amostradas podem não corresponder ao sistemas dos quais foram obtidas, tornando os resultados incorretos). Neste trabalho, foi investigada a influência de três tipos de perturbação (ruído): adição, remoção e troca aleatória de conexões nas propriedades de redes complexas, e as mais apropriadas para caracterizar redes amostradas foram identificadas. Além disso, foram definidas duas novas estruturas em redes complexas: árvores de borda e cadeias de vértices. A ocorrência dessas estruturas em redes mal amostradas tende a ser alta, indicando que existe uma relação com redes parcialmente amostradas. Para verificar tal hipótese, foi investigada a presença de cadeias de vértices em redes gradativamente amostradas por caminhadas aleatórias. / Several natural or human made complex systems can be represented by complex networks a theory which integrates the study of graphs with statistical mechanics. This kind of representation, however, can be biased by the way in which the data is obtained. In general, the data used to represent such systems is not always accurate, as in the case of theWorldWideWeb (WWW). Therefore, even if the sampled networks are large, their properties are directly affected by the way in which they were obtained and may not correspond to those of their respective original networks. For instance, the most used sampling methodology for capturing routers of the Internet, if performed on random networks, tends to obtain scale-free networks as results. On the other hand, sampled scale-free networks are not guaranteed to have this property. Because of these and other problems which may occur during the network sampling, it is very important to evaluate the variation of the network properties with respect to noise (in order to know which of them have less variation, being therefore more suitable for the characterization of networks with sampling problems) and the effect of sampling in the characterization, classification, and analysis of complex networks. In this work, we investigated the effect of three types of perturbations (noise), namely, edge addition, removal, and rewiring on the respectively estimated complex network properties, and the most suitable properties to characterize sampled networks were identified. Furthermore, two novel structures in complex networks were defined, namely, border trees and chains of vertices, which are possibly related to sampling. The occurrence of these structures in poorly-sampled networks was found to be high, implying a relation with partially sampled networks. In order to investigate such a hypothesis, the presence of chains of vertices was investigated in networks which were gradually sampled by random walks.

Markov decision processes (MDPs) are a general framework used by Artificial Intelligence (AI) researchers to model decision theoretic planning problems. Solving real world MDPs has been a major and challenging research topic in the AI literature. This paper discusses two main groups of approaches in solving MDPs. The first group of approaches combines the strategies of heuristic search and dynamic programming to expedite the convergence process. The second makes use of graphical structures in MDPs to decrease the effort of classic dynamic programming algorithms. Two new algorithms proposed by the author, MBLAO* and TVI, are described here.

Muitos sistemas complexos naturais ou construídos pelos seres humanos podem ser representados por redes complexas, uma teoria que une o estudo de grafos com a mecânica estatística. Esse tipo de representação, porém, pode ser comprometido pela maneira como os dados são obtidos. Em geral, os dados utilizados para representar tais sistemas nem sempre são precisos ou completos e correspondem a apenas amostras pequenas de redes maiores, como é o caso da teia mundial (WWW). Dessa forma, mesmo que as amostras sejam grandes, as suas propriedades são diretamente afetadas pela maneira como elas são obtidas e podem não corresponder com as de suas respectivas redes originais. Por exemplo, a amostragem mais utilizada para captura de roteadores da Internet, se empregada em redes aleatórias, tende a obter redes sem escala como resultado. Em contrapartida, amostras de redes sem escala não têm garantia de preservar essa estrutura. Por causa desses e outros problemas que possam ocorrer na amostragem das redes, é muito importante avaliar a variação das propriedades das redes a ruídos (para saber quais variam menos, sendo, portanto, mais adequadas para caracterizar redes com problemas de amostragem) e os efeitos da amostragem na caracterização, classificação e análise de redes complexas (pois redes amostradas podem não corresponder ao sistemas dos quais foram obtidas, tornando os resultados incorretos). Neste trabalho, foi investigada a influência de três tipos de perturbação (ruído): adição, remoção e troca aleatória de conexões nas propriedades de redes complexas, e as mais apropriadas para caracterizar redes amostradas foram identificadas. Além disso, foram definidas duas novas estruturas em redes complexas: árvores de borda e cadeias de vértices. A ocorrência dessas estruturas em redes mal amostradas tende a ser alta, indicando que existe uma relação com redes parcialmente amostradas. Para verificar tal hipótese, foi investigada a presença de cadeias de vértices em redes gradativamente amostradas por caminhadas aleatórias. / Several natural or human made complex systems can be represented by complex networks a theory which integrates the study of graphs with statistical mechanics. This kind of representation, however, can be biased by the way in which the data is obtained. In general, the data used to represent such systems is not always accurate, as in the case of theWorldWideWeb (WWW). Therefore, even if the sampled networks are large, their properties are directly affected by the way in which they were obtained and may not correspond to those of their respective original networks. For instance, the most used sampling methodology for capturing routers of the Internet, if performed on random networks, tends to obtain scale-free networks as results. On the other hand, sampled scale-free networks are not guaranteed to have this property. Because of these and other problems which may occur during the network sampling, it is very important to evaluate the variation of the network properties with respect to noise (in order to know which of them have less variation, being therefore more suitable for the characterization of networks with sampling problems) and the effect of sampling in the characterization, classification, and analysis of complex networks. In this work, we investigated the effect of three types of perturbations (noise), namely, edge addition, removal, and rewiring on the respectively estimated complex network properties, and the most suitable properties to characterize sampled networks were identified. Furthermore, two novel structures in complex networks were defined, namely, border trees and chains of vertices, which are possibly related to sampling. The occurrence of these structures in poorly-sampled networks was found to be high, implying a relation with partially sampled networks. In order to investigate such a hypothesis, the presence of chains of vertices was investigated in networks which were gradually sampled by random walks.

The aim of this thesis is to construct exact, axially symmetric solutions of Einstein- Maxwell(-dilaton) equations, which possess a discrete translational symmetry along an axis. We present two possible approaches to their construction. The first one is to solve Einstein-Maxwell equations, the second one relies on a dimensional reduction from a higher dimension. We examine the geometry of the solutions, their horizons and singu- larities, motions of charged test particles and compare them. 1

No / This paper presents a new partitioning strategy with the objective of increasing scalability by reducing computational effort of routing in networks. The original network is partitioned into blocks (subnetworks) so that there is a bi-directional link between any two blocks. When there is a connection request between a pair of nodes, if the nodes are in the same block, we only use the small single block to derive routings. Otherwise we combine the two blocks where the two nodes locate and in this way the whole network will never be used. The strategy is generic in that it can be used in any underlying routing algorithms in the network layer and can be applied to any networks with fixed topology such as fixed wired subnetworks of the Internet. The performance of this strategy has been investigated by building a simulator in Java and a comparison with existing stochastic partitioning techniques is shown to give superior performance in terms of trade-off in blocking probability (the probability of failure to find a path between source and destination satisfying QoS constraints) and reduction of computational effort.