Les files d'attente fluides sont des processus markoviens à deux dimensions, où la première composante, appelée le niveau, représente le contenu d'un réservoir et prend des valeurs continues, et la deuxième composante, appelée la phase, est l'état d'un processus markovien dont l'évolution contrôle celle du niveau. Le niveau de la file fluide varie linéairement avec un taux qui dépend de la phase et qui peut prendre n'importe quelle valeur réelle. Dans cette thèse, nous explorons le lien entre les files fluides et les processus QBD, et nous appliquons des arguments utilisés en théorie des processus de renouvellement pour obtenir la distribution stationnaire de plusieurs modèles fluides. Nous commençons par l'étude d'une file fluide avec un réservoir de taille infinie; nous déterminons sa distribution stationnaire, et nous présentons un algorithme permettant de calculer cette distribution de manière très efficace. Nous observons que la distribution stationnaire de la file fluide de capacité infinie est très semblable à celle d'un processus QBD avec une infinité de niveaux. Nous poursuivons la recherche des similarités entre les files fluides et les processus QBD, et nous étudions ensuite la distribution stationnaire d'une file fluide de capacité finie. Nous montrons que l'algorithme valable pour le cas du réservoir infini permet de calculer toutes les quantités importantes du modèle avec un réservoir fini. Nous considérons ensuite des modèles fluides plus complexes, de capacité finie ou infinie, où le comportement du processus markovien des phases peut changer lorsque le niveau du réservoir atteint certaines valeurs seuils. Nous montrons que les méthodes développées pour des modèles classiques s'étendent de manière naturelle à ces modèles plus complexes. Pour terminer, nous étudions les conditions nécessaires et suffisantes qui mènent à l'indépendance du niveau et de la phase d'une file fluide de capacité infinie en régime stationnaire. Ces résultats s'appuient sur des résultats semblables concernant des processus QBD. Markov modulated fluid queues are two-dimensional Markov processes, of which the first component, called the level, represents the content of a buffer or reservoir and takes real values; the second component, called the phase, is the state of a Markov process which controls the evolution of the level in the following manner: the level varies linearly at a rate which depends on the phase and which can take any real value. In this thesis, we explore the link between fluid queues and Quasi Birth-and-Death (QBD) processes, and we apply Markov renewal techniques in order to derive the stationary distribution of various fluid models. To begin with, we study a fluid queue with an infinite capacity buffer; we determine its stationary distribution and we present an algorithm which performs very efficiently in the determination of this distribution. We observe that the equilibrium distribution of the fluid queue is very similar to that of a QBD process with infinitely many levels. We further exploit the similarity between the two processes, and we determine the stationary distribution of a finite capacity fluid queue. We show that the algorithm available in the infinite case allows for the computation of all the important quantities entering in the expression of this distribution. We then consider more complex models, of either finite or infinite capacities, in which the behaviour ff the phase process may change whenever the buffer is empty or full, or when it reaches certain thresholds. We show that the techniques that we develop for the simpler models can be extended quite naturally in this context. Finally, we study the necessary and sufficient conditions that lead to the independence between the level and the phase of an infinite capacity fluid queue in the stationary regime. These results are based on similar developments for QBD processes.

O projeto de arquitetura é uma atividade que frequentemente demanda respostas claras a problemas não bem definidos. Isto conduz a diferentes tipos de abordagem para a elaboração de propostas de projetos, muitas das quais baseadas em tomadas de decisão intuitivas. Sistema de Apoio à Decisão (SAD) é um conceito que se aplica a qualquer ferramenta capaz de organizar informações com o intuito de apoiar a tomada de decisões. Na prática, se traduz em softwares que conduzem e organizam informações, para que a tomada de decisões possua maior embasamento, maior grau de racionalização e melhor eficiência. O uso de softwares na arquitetura data da década de 1960, com a introdução dos sistemas de Desenho Assistido por Computador (CAD), dedicado quase inteiramente restrito à atividade representacional, possuindo pouco impacto sobre o processo de concepção do projeto. O desenho algorítmico, ou paramétrico, entretanto, apresenta crescente utilização no mercado da arquitetura, com o intuito de fornecer arcabouços lógicos, típicos da programação de computadores, orientados à concepção de projetos. Estudos demonstram que o desenho algorítmico pode ser de grande valia na compreensão de espaços urbanos, e em espacial, de assentamentos informais. Os assentamentos informais geralmente possuem conformação complexa, dada a sua natureza transgressora, alheia à lógica e aos códigos da cidade formal. Por outro lado, os projetos de habitação social, comumente direcionado a população residente nestes assentamentos, pouco correspondem a esta complexidade ou se apoiam em leituras locais. As experiências nacionais de programas de oferta habitacional, como o Banco Nacional de Habitação (BNH) e o Programa Minha Casa Minha Vida (PMCMV) priorizaram, de uma maneira geral, quantidade sobre a qualidade, resultando em massificação de unidades habitacionais repetitivas e pouco conectadas ao perfil socioambiental local. Dados locais que possam auxiliar no desenvolvimento de propostas habitacionais de melhor qualidade são escassos quando relativas aos assentamentos informais. Este estudo objetivou a elaboração de um framework para a condução de informação de assentamentos informais, extraída por meio de processamento de imagens de satélite e dados censitários, na elaboração de estudos primários volumétricos desenvolvidos com uso do desenho algorítmico. Para isso, uma estrutura de trabalho de 6 etapas foi desenvolvida e explorada num estudo de caso hipotético, tendo por base uma comunidade informal real, o bairro da Ilhinha, localizado em São Luís - MA. Neste estudo de caso, foi possível verificar que o framework forneceu amparo para a tomada de decisões no desenvolvimento de estudos de implantação de blocos, quantificação de unidades e distribuição destas nos blocos, parâmetros conectados direta e dinamicamente com os dados extraídos nas etapas iniciais. / Architectural design is an activity that often demands clear answers for not well-defined problems. That incurs in different approaches to design process. Many of these approaches are based in intuitive decision making. Decision Support Systems (DSS) is a concept applied to any tool capable of organizing information that aims to support decision making process. In practice, this concept is applied to software oriented to drive and organize information so that the decision-making process can be enhanced upon theoretical basis, rationalization and efficiency. In architecture, software usage dates from the 1960 decade, when Computer Aided Design (CAD) systems were introduced. That usage remains almost entirely restricted to the representational activity, having small impact upon conceptive design stages. Algorithmic (or parametric design) on the other hand, gains increasing usage and acceptance within architecture market, aiming to deliver logic frameworks (common in computer programming) oriented to conceptive design. Some studies point out that algorithmic design can be useful to understand urban spaces, especially unformal settlements. The settlements usually have complex conformation, given is transgressive nature, aside from logic and codes of formal cities. Housing projects, often directed to these settlements resident population, weakly corresponds to this complexity and rarely is supported by local readings. Brazilian experiences with housing programs as Banco Nacional de Habitação (BNH) and Minha Casa Minha Vida program (PMCMV) prioritized, generally speaking, quantity over quality, resulting in massification of repetitive habitational units, weakly or not connected to local socioenvironmental characteristics. Local data availability, which can assist the development of better quality housing design, is rare when related to unformal settlements. This study aimed to develop a framework to drive information from unformal settlements, extracted by satellite imagery processing and census data, to develop basic volumetric studies with algorithmic modeling. To achieve this, a 6-stage workflow was developed and explored in a hypothetic case study, based in a real unformal settlement - Ilhinha neighborhood, in São Luís, northeast Brazil. It was verified that the framework enabled a better supported decision mass study making process of implantation of housing blocks and quantification and distribution of habitational unities, parameters directly and dynamically linked to information gathered from the framework first stages.

A tecnologia de virtualização de redes é um novo paradigma de redes que permite a múltiplas redes virtuais (VNs) compartilharem de uma forma eficiente e eficaz a mesma rede de infraestrutura denominada rede de substrato (SN). A implementação e o desenvolvimento de novos protocolos, testes de novas soluções e arquiteturas para a Internet atual e do futuro podem ser tratadas por meio da virtualização de redes. Com a virtualização de redes surge um desafio denominado problema VNE. O problema de virtualização de redes embutidas (VNE) consiste em realizar o mapeamento dos nós virtuais e o mapeamento dos enlaces virtuais sobre uma rede de substrato (SN). O problema é conhecido como NP-Hard e a sua solução é realizada por meio de algoritmos heurísticos e aproximados que realizam o mapeamento de nós e enlaces virtuais em dois estágios de forma independente ou coordenada. A presente tese tem o objetivo de resolver o mapeamento dos enlaces virtuais do problema VNE com a utilização da álgebra de caminhos. A solução apresentada fornece o melhor desempenho quando comparada com as demais soluções de virtualização de redes encontradas na literatura. Os resultados obtidos nas simulações para o problema VNE foram avaliados e analisados com a utilização do algoritmo desenvolvido nesta tese denominado Path Algebra for Virtual Link Mapping (PAViLiM), que utiliza a álgebra de caminhos para realizar o mapeamento de enlaces virtuais para caminhos na rede de substrato. A álgebra de caminhos é poderosa e flexível. Tal flexibilidade permite que ocorra uma exploração detalhada do espaço de soluções e a identificação do melhor critério e política que devem ser utilizados para a virtualização de redes. / The network virtualization technology is a new paradigm of network that allows multiple virtual networks (VNs) share in an efficient and effective way the same network infrastructure called substrate network (SN). The implementation and the development of new protocols, testing of new solutions and architectures for current and future Internet can be addressed through network virtualization. With the network virtualization arises a challenge called VNE problem. The problem of virtual network embedded (VNE) is to conduct the mapping of the virtual nodes and mapping of the virtual links over a substrate network (SN).The problem is known as NP-Hard and its solution is accomplished by means of approximate and heuristic algorithms that perform the mapping of virtual nodes and links in two stages independently or coordinated. This thesis aims to solve the mapping of virtual links for VNE problem using the paths algebra. The solution presented provides the best performance when compared with other networks virtualization solutions from the literature. The results of simulation for the VNE problem were evaluated and analyzed using the algorithm developed in this thesis called Path Algebra for Virtual Link Mapping (PAViLiM), which uses the paths algebra to perform the mapping of virtual links to paths in substrate network. The paths algebra is powerful and flexible. This flexibility allows the occurrence of a detailed exploration for identifying the best solutions and political criteria to be used for network virtualization.

O ensino de projeto arquitetônico é, via de regra, o tronco da estrutura curricular dos cursos de Arquitetura no Brasil. Não obstante, com o passar do tempo, o ensino de projeto se mostra à prova de definições estáticas a respeito de metodologias de ensino para sua prática e aprendizado. São inúmeras as metodologias propostas e em desenvolvimento para o ensino de projeto, pois a elaboração de um projeto envolve a inter-relação entre diversas categorias de conhecimentos e habilidades, que interagem de modo não-linear entre si. Com a revolução digital do exercício profissional de arquitetura, um novo paradigma se apresenta como um particular desafio para as universidades brasileiras: introduzir no currículo do curso metodologias de ensino de projeto arquitetônico estruturadas, de modo articulado com meios e plataformas digitais que exploram as ferramentas computacionais para além do limiar da simples representação gráfica. Analisando o panorama geral, é possível notar que esta prática ainda está em estágio inicial no País (ROMCY, 2017). Segundo Orciuoli, (2009) a arquitetura digital ainda é vista com certa resistência pelas próprias faculdades de arquitetura. Neste sentido, esta dissertação se propõe a pesquisar a implantação de estratégias para o desenvolvimento de projeto utilizando o ambiente digital algorítmico e paramétrico em uma disciplina de projeto arquitetônico, assim como a instrumentalização necessária para a utilização destas ferramentas computacionais Foi desenvolvida uma pesquisa-ação em disciplinas do 4ª semestre da faculdade de arquitetura onde se propuseram exercícios relacionando conteúdos instrumentais às disciplinas de projeto para conectar o conhecimento computacional às estratégias projetuais. Na sequência, foi realizada uma análise sobre a integração entre softwares de desenho em plataforma BIM (Building Information Modelling) e softwares de modelagem algorítmica/paramétrica. Com isso, pretendeu-se fornecer subsídios instrumentais para que os alunos possam explorar estratégias projetuais digitais que utilizassem formas complexas, diminuindo assim os percalços habituais relativos à falta de destreza técnica na ferramenta digital tanto para o raciocínio projetual quanto para a posterior representação gráfica de suas propostas. Ademais foram analisados comparativamente os dados fornecidos pelos alunos antes e depois da implementação da abordagem de projeto digital, a fim de detectar problemas, potencialidades e efetividade da estratégia de ensino proposta. / The teaching of architectural design is commonly the main curricula structure in architectural graduation in Brazil. Even though, over time, project teaching has been difficult to define by static methodologies to teach its practice and learning. In fact, there are lots of methodologies proposed and in development to teach the activity of design. That´s because the design activity involves inter relationships between several knowledge domains and skills that´s interact in a nonlinear way. With the digital revolution of architectural practice, a new paradigm reveals itself as a particular challenge to Brazilians architecture schools: how to introduce, in the course curricula, structured methodologies to architectural design teaching in an articulated manner with digital media that explores computational tools beyond graphic representations boundary. Analyzing the general panorama, it is possible to note that this practice is still in its initial stages in Brazil (ROMCY, 2017). According to Orciuoli, (2009) digital architecture is still seen with some resistance by the architecture faculties themselves. In this sense, this work proposes to study strategies to develop projects using digital algorithmic and parametric media in a discipline of architectural design as well as the knowledge necessary to carry on these computational tools An action research was developed in disciplines of the 4th semester of the architecture faculty where they were proposed exercises relating instrumental disciplines with design disciplines to connect computational knowledge to design strategies. Next was analyzed the integration between a BIM software with a algorithmic/parametric modeling software. To this end, instrumental grants were able to assist people in the development of digital capacities, using the complex forms reducing in this way the usual gaps related to a lack of technical skills on reasoning or drawing with digital tool. Data developed by the students before and after the implementation of the digital design approach was analyzed comparatively in order to detect problems, potentialities and effectiveness of the proposed teaching strategy.

Dans le premier chapitre de cette thèse on rappelle l'énoncé ainsi que des équivalents de la conjecture de Leopoldt puis l'on donne un algorithme permettant de vérifier cette conjecture pour un corps de nombre et premier donnés. Pour la suite on suppose cette conjecture vraie pour le premier p fixé Et on étudie la torsion du groupe de Galois de l'extension abélienne maximale p-ramifiée. On présente une méthode qui détermine effectivement les facteurs invariants de ce groupe fini. Dans le troisième chapitre on donne des résultats numériques que l'on interpréte via des heuristiques à la Cohen-Lenstra. Dans le quatrième chapitre, à l'aide de l'algorithme qui permet le calcul de ce module, on donne des exemples de corps et de premiers vérifiant la conjecture de Greenberg. / In the first chapter of this thesis we explain Leopoldt's conjecture and some equivalent formulations. Then we give an algorithm that checks this conjecture for a given prime p and a number field. Next we assume that this conjecture is true, and we study the torsion part of the Galois group of the maximal abelian p-ramified p-extension of a given number field. We present a method to compute the invariant factors of this finite group. In the third chapter we give an interpretation of our numrical result by heuristics “à la” Cohen-Lenstra. In the fourth and last chapter, using our algorithm which computes this torsion submodule, we give new examples of numbers fields which satisfy Greenberg's conjecture.

A tecnologia de virtualização de redes é um novo paradigma de redes que permite a múltiplas redes virtuais (VNs) compartilharem de uma forma eficiente e eficaz a mesma rede de infraestrutura denominada rede de substrato (SN). A implementação e o desenvolvimento de novos protocolos, testes de novas soluções e arquiteturas para a Internet atual e do futuro podem ser tratadas por meio da virtualização de redes. Com a virtualização de redes surge um desafio denominado problema VNE. O problema de virtualização de redes embutidas (VNE) consiste em realizar o mapeamento dos nós virtuais e o mapeamento dos enlaces virtuais sobre uma rede de substrato (SN). O problema é conhecido como NP-Hard e a sua solução é realizada por meio de algoritmos heurísticos e aproximados que realizam o mapeamento de nós e enlaces virtuais em dois estágios de forma independente ou coordenada. A presente tese tem o objetivo de resolver o mapeamento dos enlaces virtuais do problema VNE com a utilização da álgebra de caminhos. A solução apresentada fornece o melhor desempenho quando comparada com as demais soluções de virtualização de redes encontradas na literatura. Os resultados obtidos nas simulações para o problema VNE foram avaliados e analisados com a utilização do algoritmo desenvolvido nesta tese denominado Path Algebra for Virtual Link Mapping (PAViLiM), que utiliza a álgebra de caminhos para realizar o mapeamento de enlaces virtuais para caminhos na rede de substrato. A álgebra de caminhos é poderosa e flexível. Tal flexibilidade permite que ocorra uma exploração detalhada do espaço de soluções e a identificação do melhor critério e política que devem ser utilizados para a virtualização de redes. / The network virtualization technology is a new paradigm of network that allows multiple virtual networks (VNs) share in an efficient and effective way the same network infrastructure called substrate network (SN). The implementation and the development of new protocols, testing of new solutions and architectures for current and future Internet can be addressed through network virtualization. With the network virtualization arises a challenge called VNE problem. The problem of virtual network embedded (VNE) is to conduct the mapping of the virtual nodes and mapping of the virtual links over a substrate network (SN).The problem is known as NP-Hard and its solution is accomplished by means of approximate and heuristic algorithms that perform the mapping of virtual nodes and links in two stages independently or coordinated. This thesis aims to solve the mapping of virtual links for VNE problem using the paths algebra. The solution presented provides the best performance when compared with other networks virtualization solutions from the literature. The results of simulation for the VNE problem were evaluated and analyzed using the algorithm developed in this thesis called Path Algebra for Virtual Link Mapping (PAViLiM), which uses the paths algebra to perform the mapping of virtual links to paths in substrate network. The paths algebra is powerful and flexible. This flexibility allows the occurrence of a detailed exploration for identifying the best solutions and political criteria to be used for network virtualization.

De nombreuses recherches montrent que les adultes résolvent de simples problèmes arithmétiques plus ou moins exclusivement par récupération de la réponse dans un réseau d’associations stockées en mémoire à long terme (Ashcraft, 1992 ; 1995 ; Campbell, 1995). Il est admis que les performances arithmétiques des jeunes enfants sont basées sur le comptage ou d’autres stratégies procédurales qui sont peu à peu remplacées par la récupération directe en mémoire (Barrouillet & Fayol, 1998). Ces premiers résultats ont été rapportés par Groen et Parkman (1972) en étudiant les temps de résolution. Mais moyenner des temps de latence d’essais impliquant différentes procédures peut conduire à des conclusions erronées sur la façon dont les problèmes sont résolus. D’autres auteurs ont préféré la méthode des protocoles verbaux. Cependant Kirk et Ashcraft (2001) remettent en question ce protocole. Nous proposons un nouveau paradigme pour faire la lumière sur la façon dont des problèmes additifs, soustractifs et multiplicatifs sont résolus par les adultes et les enfants. Ce paradigme tire avantage du fait que les procédures de calcul dégradent les traces en mémoire des opérandes impliquées dans les calculs (Thevenot, Barrouillet & Fayol, 2001). Le temps nécessaire à l’algorithme pour parvenir à la réponse et son coût cognitif entraînent une réduction du niveau d’activation des opérandes. Cette baisse d’activation résulte d’un phénomène de déclin mémoriel qui entraîne une détérioration des traces en mémoire (Towse & Hitch, 1995 ; Towse, Hitch & Hutton, 1998) et de l’activation concurrente de résultats transitoires provoquant un partage attentionnel entre les opérandes, leurs composantes et les résultats intermédiaires nécessaires pour arriver à la solution (Anderson, 1993). Par conséquent, lorsque l’algorithme aboutit à la réponse les traces des opérandes sont dégradées et la récupération en mémoire est difficile. Ce phénomène devrait être plus prononcé pour les grands nombres car parvenir au résultat nécessite plus d’étapes et plus de temps. En contrastant la difficulté rencontrée par des adultes pour reconnaître des opérandes après leur implication soit dans une opération arithmétique (addition, soustraction ou multiplication) soit dans une comparaison (qui ne nécessite aucune décomposition) avec un troisième nombre, nous pourrons déterminer si l’opération a été résolue par procédure algorithmique. Si la difficulté est la même dans les deux conditions, l’opération aura été résolue par récupération, une telle activité ne nécessite pas la décomposition des opérandes / Numerous studies show that adults solve simple arithmetic problems more or less exclusively by retrieval the response in a network of associations stored in long-term memory (Ashcraft, 1992, 1995 and Campbell, 1995). It is recognized that the arithmetic performance of young children are based on counting or other procedural strategies that are gradually replaced by direct memory retrieval (Barrouillet & Fayol, 1998). These first results were reported by Groen and Parkman (1972) by studying the resolution time. But average latency of trials involving different procedures can lead to erroneous conclusions about how problems are solved. Other authors have preferred the method of verbal reports. But Kirk and Ashcraft (2001) question this paradigm. We propose a new paradigm to shed light on how the addition problems, subtraction and multiplication are resolved by adults and children. This paradigm takes advantage of the fact that algorithmic computation degrades the memory traces of the operands involved in the calculation (Thevenot, Barrouillet, & Fayol, 2001). The time needed by the algorithm to reach the answer and its cognitive cost lead to a reduction in the level of activation of the operands. This decrease in activation would result both from a memory decay phenomenon, which damages memory traces (Towse & Hitch, 1995; Towse, Hitch & Hutton, 1998) and from the necessary concurrent activation of transitory results, which induces a sharing of the attentional resources between the operands, their components and the intermediate results necessary to be reached to solve the problem (Anderson, 1993). Therefore, when the algorithm leads to the response, traces of the operands are degraded and retrieval the operand in memory is more difficult. This phenomenon should be more pronounced for large numbers since arriving at the result requires more steps and more time. Thus, contrasting the relative difficulty that adults or children encounter in recognizing operands after either their involvement in an arithmetic problem or their simple comparison with a third number can allow us to determine if the arithmetic problem has been solved by an algorithmic procedure or by retrieval of the result from memory: If operands are more difficult to recognize after the operation than after their comparison, we can assume that an algorithmic procedure has been used. On the contrary, if the difficulty is the same in both conditions, then the operation has most probably been solved by retrieval, a fast activity that does not imply the decomposition of the operands

O processo de projeto de arquitetura, engenharia e construção (AEC) tem evoluído consideravelmente com a finalidade de responder à complexidade dos sistemas interconectados de um edifício (TOULOUPAKI e THEODOSIOU, 2017). Vêm sendo solicitados à arquitetura contemporânea, processos de projeto capazes de quantificar e qualificar o maior número de variáveis presentes nos sistemas de um edifício, com a intenção de balancear suas dinâmicas e inter-relações (SHI e YANG 2013). A presente pesquisa utiliza o processo de projeto orientado ao desempenho como condutor de sua investigação. Os projetos e pesquisas que exploram o desempenho através da otimização de projetos em AEC são predominantemente internacionais. Além desse aspecto, grande parte dos processos e seus conjuntos de procedimentos não são publicados, especialmente os métodos para gerar os algoritmos utilizados. Muitas destas não detalham, tampouco disponibilizam suas implementações. Tal condição impossibilita conclusões seguras sobre as características das otimizações produzidas, dificulta a reprodução dos métodos e desampara discussões sobre soluções alternativas. Desta maneira, torna-se reduzida a reflexão na arquitetura sobre quais métodos são mais apropriados para a otimização de determinados problemas do projeto arquitetônico (WORTMANN et al., 2015). Neste contexto, a presente pesquisa estrutura-se na metodologia Design Science (SIMON, 1996), dada sua natureza de produção de conhecimento através da abordagem pragmática e interdisciplinar, envolvendo conteúdo da arquitetura e da ciência da computação. Pode-se por fim afirmar que com o apoio da metodologia empregada na presente pesquisa, desde a revisão bibliográfica ao plug-in ArchOptimum desenvolvido nos experimentos realizados, obteve-se resultados que colaboram com a compreensão e abertura dos potenciais da associação entre a arquitetura e a ciência da computação através do design computacional. / The architecture, engineering, and construction (AEC) design process has evolved considerably in order to respond to the complexity of the interconnected building systems (TOULOUPAKI and THEODOSIOU, 2017). In contemporary architecture, computational design processes are capable of quantifying and qualifying better variables present in the building, balancing their dynamics and interrelationships (SHI and YANG 2013). This research uses the process of Performance-oriented Architectural Design as the driver of its investigation. Research and projects in AEC that explore performance through optimization are predominantly international. In addition, most of the processes and their sets of procedures are not published, especially the methods to generate the algorithms used. Many of these do not detail and make their implementations available. This makes it impossible to secure conclusions about the characteristics of the optimizations produced, makes it difficult to reproduce the methods, and disregards discussions about alternative solutions. These facts reduce the reflection on which methods are more appropriate for the optimization of specific problems of the architectural design (WORTMANN et al., 2015). In this context, the present research is structured in the Design Science methodology (SIMON, 1996), given its nature of knowledge production through the pragmatic and interdisciplinary approach, involving the content of architecture and computer science. Finally, it is possible to state that with the support of the methodology used in the present research, from the bibliographic review to the ArchOptimum plug-in developed in the experiments, we obtained results that collaborate with the understanding and opening the potencies of the association between architecture and computer science through Computational Design.

<p>In order that epistemic formulas might be coupled with approximate databases, it is necessary to have a well-defined semantics for the knowledge operator and a method of reducing epistemic formulas to approximate formulas. In this thesis, two possible definitions of a semantics for the knowledge operator are proposed for use together with an approximate relational database: </p><p>* One based upon logical entailment (being the dominating notion of knowledge in literature); sound and complete rules for reduction to approximate formulas are explored and found not to be applicable to all formulas. </p><p>* One based upon algorithmic computability (in order to be practically feasible); the correspondence to the above operator on the one hand, and to the deductive capability of the agent on the other hand, is explored.</p><p>Also, an inductively defined semantics for a"know whether"-operator, is proposed and tested. Finally, an algorithm implementing the above is proposed, carried out using Java, and tested.</p>

In this master's thesis, hidden Markov models (HMM) are evaluated as a tool for forecasting movements in a currency cross. With an ever increasing electronic market, making way for more automated trading, or so called algorithmic trading, there is constantly a need for new trading strategies trying to find alpha, the excess return, in the market. HMMs are based on the well-known theories of Markov chains, but where the states are assumed hidden, governing some observable output. HMMs have mainly been used for speech recognition and communication systems, but have lately also been utilized on financial time series with encouraging results. Both discrete and continuous versions of the model will be tested, as well as single- and multivariate input data. In addition to the basic framework, two extensions are implemented in the belief that they will further improve the prediction capabilities of the HMM. The first is a Gaussian mixture model (GMM), where one for each state assign a set of single Gaussians that are weighted together to replicate the density function of the stochastic process. This opens up for modeling non-normal distributions, which is often assumed for foreign exchange data. The second is an exponentially weighted expectation maximization (EWEM) algorithm, which takes time attenuation in consideration when re-estimating the parameters of the model. This allows for keeping old trends in mind while more recent patterns at the same time are given more attention. Empirical results shows that the HMM using continuous emission probabilities can, for some model settings, generate acceptable returns with Sharpe ratios well over one, whilst the discrete in general performs poorly. The GMM therefore seems to be an highly needed complement to the HMM for functionality. The EWEM however does not improve results as one might have expected. Our general impression is that the predictor using HMMs that we have developed and tested is too unstable to be taken in as a trading tool on foreign exchange data, with too many factors influencing the results. More research and development is called for.