Este trabalho descreve a implementação de algoritmos de inferência para modelos de primeira ordem. Três algoritmos foram implementados: ve, c-fove e ac-fove. Este último e o estado da arte no calculo de probabilidades em Redes Bayesianas Relacionais e não possua nenhuma implementação disponível. O desenvolvimento foi feito segundo uma metodologia ágil que resultou em um pacote de software que pode ser utilizado em outras implementações. Mostra-se que o software criado possui o desempenho esperado em teoria, embora apresente algumas limitações. Esta dissertação contribui também com novos tópicos teóricos que complementam o algoritmo. / In this work, we describe the implementation of inference algorithms for first order models. Three algorithms were implemented: ve, c-fove and ac-fove. The latter is the state of the art in probability calculations for Relational Bayesian Networks and had no implementation available. The development was done according to an agile methodology, which resulted in a software that can be used in other packages. We show that the resulting software has the expected performance from the theory, although with some limitations. This work also contributes with new theoretical topics that complement the algorithm.

Orientador: Maurice de Koning / Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Fisica Gleb Wataghin / Made available in DSpace on 2018-08-13T12:16:33Z (GMT). No. of bitstreams: 1 CarvajalJara_DiegoAlejandro_M.pdf: 4649880 bytes, checksum: d014cd5c835938e4643e478fa70353be (MD5) Previous issue date: 2009 / Resumo: Recentemente tem sido proposta a existência de uma transição líquido-líquido em substâncias puras com o propósito de explicar alguns comportamentos anômalos como os incrementos de funções resposta (compressibilidade isotérmica, coeficiente de expansão térmica, calor específico) ao diminuir a temperatura de um líquido. A existência deste tipo de transição foi demostrada experimentalmente para o fósforo por dispersão de raios X, e através de simulações atomísticas ou de primeiros princípios para a água, o silício, o carbono, etc. A compreensão detalhada deste tipo de transição está intimamente relacionada à questão fundamental de quais fatores físicos controlam as propriedades de um líquido, e portanto o estudo desta é de grande importância para o desenvolvimento de novas tecnologias, a síntese de novos materiais e o controle de suas propriedades. Neste trabalho realizamos simulações atomísticas sobre um sistema de 1152 partículas de Gálio submetidas a um potencial semi-empírico MEAM, com condições periódicas de contorno. Com estas simulações procuramos uma transição líquido-líquido no Gálio através de um processo de eliminação de três teorias. Inicialmente mostramos que o limite de metaestabilidade do líquido super-aquecido tem um comportamento monótono decrescente no plano de fase P-T. Posteriormente nosso sistema apresenta histerese, uma descontinuidade no volume, estruturas locais diferentes, duas fases que fluem, e um calor latente característico de uma transição de fase de primeira ordem. Por todas estas razões concluímos que o sistema simulado apresenta um transição líquido-líquido de primeira ordem. Adicionalmente foram realizadas compressões e expansões isotérmicas para temperaturas diferentes, observando que estes processos também apresentam histerese e que ela diminui com o aumento da temperatura, indicando assim a possível existência de um segundo ponto crítico e a finalização da transição líquido-líquido. Finalizamos o trabalho com a obtenção e a caracterização de uma possível nova fase cristalina do Gálio cuja estrutura ainda não tem sido obtida experimentalmente. Esta fase foi obtida por casualidade durante o estudo da existência de uma transição líquido-líquido no Gálio ao tencionar o Gálio a -1.6GPa. Sua estrutura é do tipo ortorrômbica com uma simetria Cmcm (grupo espacial 63) e sua principal diferença do Gálio-I é a orientação dos dímeros de Gálio, que nesta nova fase estão dispostos paralelamente. Simulações por DFT mostraram que esta nova fase é metaestável a pressão nula e chega a ser estável a pressões negativas abaixo de ~ 1.5GPa. / Abstract: Recently, the existence of a liquid-liquid transition in pure substances has been proposed as an explanation of anomalous behaviors such as the increase of response functions (isothermal compressibility, coefficient of thermal expansion, specific heat) with decreasing temperature displayed by some liquids. The existence of this type of transition has been demonstrated experimentally for phosphorous by X-ray diÿraction, and through atomistic simulations for water, silicon and carbon. The detailed understanding of this type of transition is closely related to the fundamental question of which physical factors control the properties of a liquid. Therefore, the study of this phenomenon is of great importance for the development of new technologies, the synthesis of new materials and the control of their properties. In this work, we carry out a series of atomistic simulations of a system containing 1152 Gallium atoms described by a semi-empirical Modified Embedded-Atom Model subject to periodic boundary conditions. By means of these simulations we search for a liquid-liquid transition in Gallium by means of a process of elimination of three theories. Initially we show that the limit of metastability of the superheated liquid has a decreasing monotonous behavior in the pressure-temperature phase diagram. Subsequently, our system presents hysteresis, a discontinuity in volume, two phases that have different local atomic structures and display diffusion, and a latent heat, all characteristic of a first-order phase transition. For all these reasons we conclude that the simulated system presents a liquid-liquid phase transition of first-order in the supercooled regime. In addition, we carried out several simulations of isothermal compressions and expansions for different temperatures. These results also show hysteresis although it is found to decrease with increasing temperature, thus indicating the possible existence of a second critical point at which the liquid-liquid transition ends. We finish our studies with the discovery and characterization of a crystalline phase of Gallium whose structure has not been observed experimentally. This phase was obtained by chance during the study of the existence of a liquid-liquid transition in Gallium under tension of -1.6GPa. Its structure is of the orthorhombic type with Cmcm symmetry (space group 63). Its main difference with respect to the Gallium-I phase is that in the new phase the Gallium dimers are disposed in a parallel fashion. Subsequent DFT simulations show that this new phase is metastable at zero pressure and predict it to become stable with respect to Gallium-I arrives at negative pressures below ~1.5GPa. / Mestrado / Física da Matéria Condensada / Mestre em Física

Neste trabalho, estudamos a classificação de singularidades de curvas espaciais simples que não são intersecções completas. O Teorema de Hilbert-Burch nos permite usar a matriz de representação para estudar a variedade definida pelo ideal gerado por seus menores maximais. Da mesma forma, as deformações da variedade determinantal podem ser representadas por perturbações da matriz e qualquer perturbação da matriz fornece uma deformação da variedade. Assim, o estudo das singularidades de curvas determinantais pode ser formulado em termos da matriz de representação da curva / In this work, we study the classification of simple space curve singularities which are not complete intersections. The Theorem of Hilbert-Burch enables us to deal with the presentation matrices instead of the ideals defined by their maximal minors. In the same way, deformations of the determinantal variety can be represented by perturbations of the matrix and any perturbation of the matrix gives rise to a deformation of the variety. Therefore, the study of determinantal curves can be formulated in terms of the presentation matrices

Este trabalho descreve o desenvolvimento de um sistema de controle com compensação de tempo morto aplicado a um ventilador para simular os efeitos do vento em modelos de embarcações. O sistema é instalado no tanque de provas do Departamento de Engenharia Naval da Escola Politécnica da USP. Um sistema de controle baseado em computador é elaborado para a realização dos testes e validação, procedendo-se inicialmente a identificação do sistema a ser controlado. Na malha de controle, o sinal de referência é a velocidade do vento desejada em uma distância pré-definida do ventilador e pode ser tanto um valor constante (representando um vento constante) ou um valor variante no tempo (representando uma rajada de vento). O sistema atuador consiste de um inversor de frequência que aciona o ventilador e a velocidade do vento é medida por um anemômetro, cujo sinal é usado como realimentação para o controlador. A sintonia da malha é realizada pelo método do modelo interno (IMC) e o efeito do tempo morto é observado nos testes, aplicando-se o compensador baseado no Preditor de Smith para minimizá-lo. Pelos resultados dos ensaios, conclui-se que o Preditor de Smith melhora o desempenho do sistema de geração de vento. / This paper describes the development of a control system with dead time compensation applied to a fan to simulate wind effects in models of vessels. The system is installed on the academic towing tank of the Naval Architecture and Oceanic Engineering Department of the University of São Paulo. A control system based on computer is designed for testing and validation, proceeding initially to identify the system to be controlled. In the control loop, the set-point is the desired speed of the wind on a pre-defined distance from the fan and it can be either a constant value (representing a constant wind) or a time-varying value (representing a wind gust). The actuator system consists of a frequency-inverter that drives the fan and the wind speed is measured by an anemometer, whose signal is used as feedback to the controller. The tuning of the controller is made by the internal model control (IMC) and the effect of dead time is observed in the tests, applying the compensator based on Smith Predictor to minimize it. Through the results from the tests, it was concluded that the Smith Predictor improves the performance of the wind generation system.

Mechanisation of Mathematics refers to use of computers to generate or check proofs in Mathematics. It involves translation of relevant mathematical theories from one system of logic to another, to render these theories implementable in a computer. This process is termed formalisation of mathematics. Two among the many ways of mechanising are: 1 Generating results using automated theorem provers. 2 Interactive theorem proving in a proof assistant which involves a combination of user intervention and automation. In the first part of this thesis, we reformulate the question of equivalence of two Links in first order logic using braid groups. This is achieved by developing a set of axioms whose canonical model is the braid group on infinite strands B∞. This renders the problem of distinguishing knots and links, amenable to implementation in first order logic based automated theorem provers. We further state and prove results pertaining to models of braid axioms. The second part of the thesis deals with formalising knot Theory in Higher Order Logic using the interactive proof assistant -Isabelle. We formulate equivalence of links in higher order logic. We obtain a construction of Kauffman bracket in the interactive proof assistant called Isabelle proof assistant. We further obtain a machine checked proof of invariance of Kauffman bracket.

Les systèmes hyperboliques interviennent dans de nombreuses branches des sciences : théorie cinétique, mécanique des fluides non visqueux, magnéto hydrodynamique, dynamique des gaz non visqueux, trafic routier, flux d’une rivière ou d’un glacier, processus de sédimentation, processus d’échanges chimiques, etc. Et souvent, les systèmes qui régissent ces évènements font intervenir des petits paramètres, dont l’étude asymptotique permet d’envisager des simplifications mathématiques et/ou informatiques notoires. L’existence locale et l’existence globale de solutions, uniformément par rapport à ces paramètres, sont des questions fondamentales. Cette thèse regroupe à la fois des résultats généraux sur l’existence locale uniforme de solutions pour des systèmes hyperboliques quasi-linéaires du premier ordre ; et sur l’existence globale uniforme de solutions autour d’un équilibre constant pour ces mêmes systèmes. Le cas du système d’Euler-Maxwell ne satisfaisant pas les conditions requises pour l’existence uniforme globale, nous le traitons à part. / Hyperbolic systems arise in a large field of sciences : kinetic theory, inviscid reactive flow, magnetohydrodynamics, inviscid gas dynamics, traffic flow, river or glacier flow, sedimentation processes, chemical exchange processes, etc. In these kind of systems, small paramaters often appear, and an asymptotic study may lead to mathematical or computational simplifications. One fundamental problem that we may work on is local and global existence of solutions for these systems, uniformly with respect to these parameters. This Ph.D. thesis includes, on one hand, general results on uniform local existence of solutions for first order quasi-linear hyperbolic systems ; and on the other hand, results on uniform global existence of solutions near constant equilibriums for these same systems. In the case of Euler-Maxwell systems, required conditions are not fulfilled for uniform global existence, then we treat it separately.

La phase d’apprentissage des méthodes d’apprentissage statistique automatique correspondent à la résolution d’un problème d’optimisation mathématique dont la fonction objectif se décompose en deux parties: a) le risque empirique, construit à partir d’une fonction de perte, dont la forme est déterminée par la métrique de performance et les hypothèses sur le bruit; b) la pénalité de régularisation, construite a partir d’une norme ou fonction jauge, dont la structure est déterminée par l’information à priori disponible sur le problème a résoudre.Les fonctions de perte usuelles, comme la fonction de perte charnière pour la classification supervisée binaire, ainsi que les fonctions de perte plus avancées comme celle pour la classification supervisée avec possibilité d’abstention, sont non-différentiables. Les pénalités de régularisation comme la norme l1 (vectorielle), ainsi que la norme nucléaire (matricielle), sont également non- différentiables. Cependant, les algorithmes d’optimisation numériques les plus simples, comme l’algorithme de sous-gradient ou les méthodes de faisceaux, ne tirent pas profit de la structure composite de l’objectif. Le but de cette thèse est d’étudier les problèmes d’apprentissage doublement non-différentiables (perte non- différentiable et régularisation non-différentiable), ainsi que les algorithmes d’optimisation numérique qui sont en mesure de bénéficier de cette structure composite.Dans le premier chapitre, nous présentons une nouvelle famille de pénalité de régularisation, les normes de Schatten par blocs, qui généralisent les normes de Schatten classiques. Nous démontrons les principales propriétés des normes de Schatten par blocs en faisant appel à des outils d’analyse convexe et d’algèbre linéaire; nous retrouvons en particulier des propriétés caractérisant les normes proposées en termes d’enveloppe convexes. Nous discutons plusieurs applications potentielles de la norme nucléaire par blocs, pour le filtrage collaboratif, la compression de bases de données, et l’annotation multi-étiquettes d’images.Dans le deuxième chapitre, nous présentons une synthèse de différentes tech- niques de lissage qui permettent d’utiliser des algorithmes de premier ordre adaptes aux objectifs composites qui de décomposent en un terme différentiable et un terme non-différentiable. Nous montrons comment le lissage peut être utilisé pour lisser la fonction de perte correspondant à la précision au rang k, populaire pour le classement et la classification supervises d’images. Nous décrivons dans les grandes lignes plusieurs familles d’algorithmes de premier ordre qui peuvent bénéficier du lissage: i) les algorithmes de gradient conditionnel; ii) les algorithmes de gradient proximal; iii) les algorithmes de gradient incrémental.Dans le troisième chapitre, nous étudions en profondeur les algorithmes de gradient conditionnel pour les problèmes d’optimisation non-différentiables d’apprentissage statistique automatique. Nous montrons qu’une stratégie de lis- sage adaptative associée à un algorithme de gradient conditionnel donne lieu à de nouveaux algorithmes de gradient conditionnel qui satisfont des garanties de convergence théoriques. Nous présentons des résultats expérimentaux prometteurs des problèmes de filtrage collaboratif pour la recommandation de films et de catégorisation d’images. / Training machine learning methods boils down to solving optimization problems whose objective functions often decomposes into two parts: a) the empirical risk, built upon the loss function, whose shape is determined by the performance metric and the noise assumptions; b) the regularization penalty, built upon a norm, or a gauge function, whose structure is determined by the prior information available for the problem at hand.Common loss functions, such as the hinge loss for binary classification, or more advanced loss functions, such as the one arising in classification with reject option, are non-smooth. Sparse regularization penalties such as the (vector) l1- penalty, or the (matrix) nuclear-norm penalty, are also non-smooth. However, basic non-smooth optimization algorithms, such as subgradient optimization or bundle-type methods, do not leverage the composite structure of the objective. The goal of this thesis is to study doubly non-smooth learning problems (with non-smooth loss functions and non-smooth regularization penalties) and first- order optimization algorithms that leverage composite structure of non-smooth objectives.In the first chapter, we introduce new regularization penalties, called the group Schatten norms, to generalize the standard Schatten norms to block- structured matrices. We establish the main properties of the group Schatten norms using tools from convex analysis and linear algebra; we retrieve in particular some convex envelope properties. We discuss several potential applications of the group nuclear-norm, in collaborative filtering, database compression, multi-label image tagging.In the second chapter, we present a survey of smoothing techniques that allow us to use first-order optimization algorithms designed for composite objectives decomposing into a smooth part and a non-smooth part. We also show how smoothing can be used on the loss function corresponding to the top-k accuracy, used for ranking and multi-class classification problems. We outline some first-order algorithms that can be used in combination with the smoothing technique: i) conditional gradient algorithms; ii) proximal gradient algorithms; iii) incremental gradient algorithms.In the third chapter, we study further conditional gradient algorithms for solving doubly non-smooth optimization problems. We show that an adaptive smoothing combined with the standard conditional gradient algorithm gives birth to new conditional gradient algorithms having the expected theoretical convergence guarantees. We present promising experimental results in collaborative filtering for movie recommendation and image categorization.

Many agricultural fields that have received long-term applications of P often contain levels of P exceeding those required for optimal crop production. Knowledge of the effect of the P remaining in the soil (residual effect) is of great importance for fertilization management. In order to characterize P forms in soils, a wide variety of methods have been proposed. The use of dialysis membrane tubes filled with hydrous ferric oxide (DMT-HFO) has recently been reported as an effective way to characterize P desorption over a long-term in laboratoty studies. However, there is little information on the relationship between kinetics of P release using this new method and plant P uptake. This method consist of a procedure of shaking a sample for a long period of time there by exploiting the whole volume of the soil which is in contrast to the actual plant mode of uptake. This method has also practical limitations in employing it for a routine soil analysis, as it is very expensive and time consuming. The objectives of this study were (i) to study the changes in labile, non-labile and residual P using successive P desorption by DMT-HFO followed by a subsequent fractionation method (combined method) (ii) to assess how the information gained from P desorption kinetic data relates to plant growth at green house and field trials (iii) to investigate the effect of varying shaking time on DMT-HFO extractable P and (iv) to propose a short cut approach to the combined method. The release kinetics of the plots from long term fertilizer trials at the University of Pretoria and Ermelo were studied. P desorption kinetics were described relatively well by a two-component first-order model (R2 = 0.947, 0.918,&0.993 for NPK, MNK,&MNPK treatments respectively). The relative contributions of both the labile pool (SPA) and the less labile pool (SPB) to the total P extracted increased with increased P supply levels. Significant correlations were observed between the rate coefficients and maize grain yield for both soil types. The correlation between the cumulative P extracted and maize yield (r = 0.997**) however was highly significant for Ermelo soils. This method was also used to determine the changes in the different P pools and to relate these P fractions with maize yield. Highly significant correlations were observed between maize grain yield and the different P fractions including total P. In both soil types the contribution of both the labile and non-labile inorganic P fractions in replenishing the solution Pi was significant where as the contributions from the organic fractions were limited. The C/HCl-Pi is the fraction that decreased most in both cases as well. Investigation was carried out to evaluate the effect of varying shaking periods on the extractable DMT-HFO-Pi for UP soils of varying P levels. Four shaking options were applied. Significant difference was observed for the treatment of high P application. Shaking option 2 seemed relatively better than the others since it showed the strongest correlation. Thus for soils with high releasing kinetics and high total P content, provided that the P release from the soil is a rate limiting step, reducing the length of shaking time could shorten the duration one needs to complete the experiment with out influencing the predicting capacity of the methodology. The other objective of this thesis was also to present a short cut method alternative to the combined fractionation method. Comparison of the sum of DMT-HFO-Pi, NaHCO3-Pi, NaOH-Pi, D/HCl-Pi and C/HCl-Pi extracted by a conventional step-by-step method with the sum of DMT-HFO-Pi and a single C/HCl-Pi extraction as a short cut approach for all extraction periods resulted in strong and significant correlations. The C/HCl-Pi fraction extracted by both methods was correlated with maize grain yield and it was found to be highly significant. This study revealed that this short cut approach could be a simplified and economically viable option to study the P dynamics of soils especially for soils where the P pool acting as a source in replenishing the labile portion of P is already identified. The method employed here therefore could act as an analytical tool to approximate successive cropping experiments carried out under green house or field condition. However, data from a wider range of soils is needed to evaluate the universality of this method. More work is also required in relating desorption indices of this method with yield parameters especially at field level. / Thesis (PhD)--University of Pretoria, 2009. / Plant Production and Soil Science / PhD / unrestricted

Automated deduction procedures for modal logics, and related decidable fragments of first-order logic, are used in many real-world applications. A popular way of obtaining decision procedures for these logics is to base them on semantic tableau calculi. We focus on calculi that use unification, instead of the more widely employed approach of generating ground instantiations over the course of a derivation. The most common type of tableaux with unification are so-called free-variable tableaux, where variables are treated as global to the entire tableau. A long-standing open problem for procedures based on free-variable tableaux is how to ensure fairness, in the sense that "equivalent" applications of the closure rule are prevented from being done over and over again. Some solutions such as using depth-first iterative deepening are known, but those are unnecessary in theory, and not very efficient in practice. This is a main reason why there are hardly any decision procedures for modal logics based on free-variable tableaux. In this thesis, we review existing work on incorporating unification into first-order and modal tableau procedures, show how the closure fairness problem arises, and discuss existing solutions to it. For the first-order case, we outline a calculus which addresses the closure fairness problem. As opposed to free-variable tableaux, closure fairness is much easier to achieve in disconnection tableaux and similar clausal calculi. We therefore focus on using clausal first-order tableau calculi for decidable classes, in particular the two-variable fragment. Using the so-called unrestricted blocking mechanism for enforcing termination, we present the first ground tableau decision procedure for this fragment. Even for such a ground calculus, guaranteeing that depth-first terminations terminate is highly non-trivial. We parametrise our procedure by a so-called lookahead amount, and prove that this parameter is crucial for determining whether depth-first derivations terminate or not. Extending these ideas to tableaux with unification, we specify a preliminary disconnection tableau procedure which uses a non-grounding version of the unrestricted blocking rule.

Tato diplomová práce prezentuje problematiku dopravního toku a jeho modelování. Zabývá se především několika LWR modely, které následně rozebírá a hledá řešení pro počáteční úlohy. Ukazuje se, že ne pro všechny počáteční úlohy lze řešení definovat na celém prostoru, ale jen v určitém okolí počáteční křivky. Proto je dále odvozena metoda výpočtu velikosti tohoto okolí a to nejen zcela obecně, ale i pro dané modely. Teoretický rozbor LWR modelů a řešení počátečních úloh jsou demonstrovány několika příklady, které zřetelně ukazují, jak se dopravní tok simulovaný danými modely chová.