Orientador: Silvio Antonio Sachetto Vitiello / Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Física Gleb Wataghin / Made available in DSpace on 2018-08-24T22:26:40Z (GMT). No. of bitstreams: 1 Pedroso_ViniciusZampronio_M.pdf: 1006322 bytes, checksum: a4558189ae209bae9a6212be5b0bd54f (MD5) Previous issue date: 2014 / Resumo: We present a trial wave function to describe systems formed from 4He atoms. This function is able to describe the system in the liquid and solid phases, it is translationally invariant and satisfies the Bose-Einstein statistics. This wave function depends on a set of auxiliary variables integrated all over the space and favors the exchange between particles, mainly at higher densities where it is more important. We also calculated the equation of state and the pair distribution function of the liquid and solid phases of the system and results are in good agreement with the experimental data / Abstract: Apresentamos uma função de onda tentativa para descrever sistemas de átomos de 4He. Esta função é capaz de descrever o sistema nas fases sólida e liquida, é translacionalmente invariante e satisfaz a estatística de Bose-Einstein. A função proposta depende de um conjunto de variáveis auxiliares integradas em todo o espaço e favorece os processos de troca de partículas, principalmente para densidades mais altas onde ela é importante. Calculamos a equação de estado e a função de distribuição radial das fases liquida e sólida do sistema e obtivemos um ótimo acordo com dados experimentais / Mestrado / Física / Mestre em Física

Orientador: Marcelo da Silva Montenegro / Tese (doutorado) - Universidade Estadual de Campinas, Instituto de Matemática Estatística e Computação Científica / Made available in DSpace on 2018-08-23T11:08:01Z (GMT). No. of bitstreams: 1 Abreu_RafaeldosReis_D.pdf: 1052192 bytes, checksum: 6fe840dfe507b4cc6c5ad9f908da486a (MD5) Previous issue date: 2013 / Resumo: Neste trabalho, tratamos de resultados de existência de soluções ground state para algumas classes de problemas elípticos sobre espaços euclidianos ou sobre domínios exteriores. Nos casos em que consideramos um domínio exterior, consideramos a condição de fronteira de Dirichlet ou de Neumann. O fato de se considerar domínios não limitados naturalmente implica em algumas dificuldades como, por exemplo, a falta de compacidade. Quando isso ocorre, em geral, a condição Palais-Smale não é válida. Para contornar esta e outras dificuldades, usamos o Teorema do Passo da Montanha sem condição Palais-Smale, Lema de Lions e Teorema de Vitali. Em nosso estudo, utilizamos métodos variacionais explorando diversas técnicas para a obtenção de pontos críticos de funcionais associados a cada problema. Pontos críticos não nulos de cada funcional são soluções de seu respectivo problema / Abstract: In this work, we deal with existence of ground state solutions for some classes of elliptic problems on Euclidean spaces or on exterior domains. In cases where we consider an exterior domain, we consider the Dirichlet boundary condition or the Neumann boundary condition. Elliptic problems involving unbounded domains naturally have some difficulties, por example, the lack of compactness. When it occurs, in general, the Palais-Smale condition is not valid. To overcome this difficulty and others, we use the Mountain Pass Theorem without Palais-Smale condition, results due to Lions and the Vitali's Theorem. In our study, we use variational methods exploring techniques to obtain critical points of functionals related to each problem. Nonzero critical points of each functional are solutions of its respective problem / Doutorado / Matematica / Doutor em Matemática

The cross-situational word learning paradigm argues that word meanings can be approximated by word-object associations, computed from co-occurrence statistics between words and entities in the world. Lexicon acquisition involves simultaneously guessing (1) which objects are being talked about (the ”meaning”) and (2) which words relate to those objects. However, most modeling work focuses on acquiring meanings for isolated words, largely neglecting relationships between words or physical entities, which can play an important role in learning. Semantic parsing, on the other hand, aims to learn a mapping between entire utterances and compositional meaning representations where such relations are central. The focus is the mapping between meaning and words, while utterance meanings are treated as observed quantities. Here, we extend the joint inference problem of word learning to account for compositional meanings by incorporating a semantic parsing model for relating utterances to non-linguistic context. Integrating semantic parsing and word learning permits us to explore the impact of word-word and concept-concept relations. The result is a joint-inference problem inherited from the word learning setting where we must simultaneously learn utterance-level and individual word meanings, only now we also contend with the many possible relationships between concepts in the meaning and words in the sentence. To simplify design, we factorize the model into separate modules, one for each of the world, the meaning, and the words, and merge them into a single synchronous grammar for joint inference. There are three main contributions. First, we introduce a novel word learning model and accompanying semantic parser. Second, we produce a corpus which allows us to demonstrate the importance of structure in word learning. Finally, we also present a number of technical innovations required for implementing such a model.

L'objectif de cette thèse est de proposer une démarche de modélisation à partir de niveau microstructure pour obtenir un modèle d'endommagement à gradient. Elle se fonde, d'une part, sur la méthode d'homogénéisation pour construire un milieu effectif à partir de la microstructure donnée, et d'autre part, sur la formulation variationnelle, à l'échelle macroscopique, d'une loi d'évolution d'endommagement à partir du milieu homogénéisé. Nous construisons dans un premier temps une approche basée sur le développement asymptotique et de la méthode variationnelle pour homogénéiser un milieu élastique périodique. Afin de modéliser le phénomène de localisation d'endommagement , cette approche a été étendue à un milieu hétérogène quasi périodique. Par un exemple du milieu micro fissuré quasi périodique, nous obtenons une énergie élastique qui dépend non seulement du gradient de l'endommagement mais aussi du gradient de déformation. Dans la deuxième partie, nous construisons un modèle d'endommagement à gradient à partir de l'énergie élastique homogénéisé en se basant sur la principe de minimisation d'énergie. En ajoutant des hypothèse pour simplifier le modèle, nous pouvons construit explicitement des états localisés d'endommagement et de déformation. Enfin, un schéma de résolution numérique, basé sur un algorithme de minimisation alternée, a été proposé pour le cas d'un barre en traction. A partir des résultats numériques, les avantages et les inconvénients du modèle sont discutés. / The aim of this work is to propose a general framework to obtain a gradient damage model from the micro-structural level. It is based, firstly, on the homogenization method to derive an effective medium from the microstructure, and secondly, on the variational formulation of a damage evolution law from the homogenized medium. We propose, as a first step, an approach based on asymptotic expansion and the variational method for homogenizing a periodic elastic medium. To model the localization of damage, this approach has been extended to a quasi-periodic heterogeneous medium. From an example of quasi periodically micro-cracked solid, we obtain an elastic energy that not only depends on the gradient of the damage but also the strain gradients. Based on the principle of energy minimization, we propose the construction of a gradient damage model from the elastic energy homogenized in the second part. By adding some hypothesis to simplify the model, we can construct localized damage and strain solutions in closed form. Finally, a numerical resolution scheme, which is based on an alternate minimization algorithm, is proposed for the one-dimensional traction bar test. From the numerical results, the advantages and disadvantages of the model are discussed.

A novel idea for perception of object surfaces is presented by so called "shape descriptors". Such idea is as an abstract level to represent the object surface by some real numbers. It has the similar idea like as the Fourier coefficients of mapping a function f(x) to frequency domain by Fourier transform. The main goal of this thesis is to define some of the key issues in understanding of an object shape and also to find a modeling methodology to create the "shape descriptors". The modeling methodology is designed based on a variational interpolation technique. Such technique is used to generate a group of variational implicit functions with help of radial basis functions. In our modeling methodology, we randomly choose some reference points on a set of related concentric spheres around a 3D point cloud data as known values in variational implicit functions. The "shape descriptors" are found from these implicit functions implementing LU decomposition. We show that the "shape descriptors" are invariant to size and positioning (rotation and translation) changes of a shape and they are also effective tools for matching of two similar objects surfaces.

Nous traitons le sujet de l’estimation conjointe des signaux aléatoires dépendant de paramètres déterministes et inconnus. Premièrement, on aborde le sujet du côté applicatif en proposant deux algorithmes de séparation de la parole voisée mono-capteur. Dans le premier, nous utilisons le modèle autorégressif de la parole qui décrit les corrélations court et long termes (quasi-périodique) pour formuler un modèle d’état dépendant de paramètres inconnus. EM-Kalman est ainsi utilisé pour estimer conjointement les sources et les paramètres. Dans le deuxième, nous proposons une méthode fréquentielle pour le même modèle de la parole où les sources et les paramètres sont estimés séparément. Les observations sont découpées à l’aide d’un fenêtrage bien conçu pour assurer une reconstruction parfaite des sources après. Les paramètres (de l’enveloppe spectrale) sont estimés en maximisant le critère du GML exprimé avec la matrice de covariance paramétrée que nous modélisons plus correctement en tenant compte de l’effet du fenêtrage. Le filtre de Wiener est utilisé pour estimer les sources. Deuxièmement, on aborde l’estimation conjointe d’un point de vue plus théorique en s'interrogeant sur les performances relatives de l’estimation conjointe par rapport à l’estimation séparée d’une manière générale. Nous considérons le cas conjointement Gaussien (observations et variables cachées) et trois méthodes itératives d'estimation conjointe: MAP en alternance avec ML, biaisé même asymptotiquement pour les paramètres, EM qui converge asymptotiquement vers ML et VB que nous prouvons converger asymptotiquement vers la solution ML pour les paramètres déterministes. / The thesis is composed of two parts. In the first part, we deal with the monaural speech separation problem. We propose two algorithms. In the first algorithm, we exploit the joint autoregressive model that models short and long (periodic) correlations of Gaussian speech signals to formulate a state space model with unknown parameters. The EM-Kalman algorithm is then used to estimate jointly the sources (involved in the state vector) and the parameters of the model. In the second algorithm, we use the same speech model but this time in the frequency domain (quasi-periodic Gaussian sources with AR spectral envelope). Observation data is sliced using a well-designed window. Parameters are estimated separately from the sources by optimizing the Gaussian ML criterion expressed using the sample and parameterized covariance matrices. Classical frequency domain asymptotic methods replace linear convolution by circulant convolution leading to approximation errors. We show how the introduction of windows can lead to slightly more complex frequency domain techniques, replacing diagonal covariance matrices by banded covariance matrices, but with controlled approximation error. The sources are then estimated using the Wiener filtering. The second part is about the relative performance of joint vs. marginalized parameter estimation. We consider jointly Gaussian latent data and observations. We provide contributions to Cramer-Rao bounds, then, we investigate three iterative joint estimation approaches: Alternating MAP/ML which suffers from inconsistent parameter bias, EM which converges to ML and VB that we prove converges asymptotically to the ML solution for parameter estimation.

D'une part, les équations de Navier-Stokes permettent de décrire les écoulements fluides, la littérature est riche de méthodes numériques permettant la résolution de celle-ci. D'autre part, nous sommes capables de mesurer de manière non-intrusive différentes caractéristique d'un écoulement (champ de vitesse et pression, etc.). Dans le cadre de cette thèse, nous nous intéressons aux techniques d'assimilation de données qui combinent les modèles numériques avec les observations afin de déterminer une meilleure approximation du système. Cette thèse s'articule autour de l'assimilation de donnée variationnelle (4DVar) qui est plus précise par construction. Nous avons mené une première application sur la reconstruction de la hauteur et vitesse de la surface libre d'un fluide contenu dans un récipient rectangulaire à fond plat. L'écoulement est modélisé par les équations de shallow water et résolues numériquement. Les observations de l'évolution de la hauteur de la surface libre ont été prélevées par un capteur de profondeur (Kinect). Nous avons comparé les résultats de la reconstruction par 4DVar avec plusieurs version de la méthode d'assimilation hybride 4DEnVar. Enfin, nous avons appliqué la technique 4DVar à la reconstruction volumique de l'aval d'un sillage de cylindre à Reynolds 300. L'écoulement turbulent a été simulé par un code DNS parallèle Incompact3D. La reconstruction a été effectué en combinant tout d'abord des observations synthétiques en trois dimension, puis en combinant des observations de plans orthogonales en stéréo PIV. / In the one hand, flow dynamics are usually described by the NavierStokes equations and the literature provides a wide range of techniques to solve such equations. On the other hand, we can nowadays measure different characteristics of a flow (velocity, pressure, temperature etc...) with non-intrusive Particle Image Velocimetry techniques. Within this thesis, we take interest in the data assimilation techniques, that combine a dynamics model with measurements to determine a better approximation of the system. This thesis focus on the classic variational assimilation technique (4DVar) which ensures a high accuracy of the solution by construction. We carry out a first application of the 4DVar technique to reconstruct the characteristics (height and velocity field) of a uni directional wave at its free surface. The fluid evolution is simulated by the shallow water equations and solved numerically. We use a simple experimental setup envolving a depth sensor (Kinect sensor) to extract the free surface height. We compared the results of the 4DVar reconstruction with different versions of the hybrid data assimilation technique 4DEnVar. Finally, we apply the 4DVar technique to reconstruct the downstream of a three dimensional cylinder wake at Reynolds 300. The turbulent flow is simulated by the high-performance multi-threading DNS code Incompact3d. This dynamics model is first combined with synthetic three dimensional observations, then with real orthogonal-plane stereo PIV observations to reconstruct the full three dimensional flow.

Le méthane est un puissant gaz à effet de serre direct et indirect, dont l’évolution récente reste mal comprise et débattue. L’objectif de ma thèse est l’étude de la capacité des nouvelles mesures du méthane par satellite à quantifier les flux annuels de méthane et leurs variations interannuelles. J’assimile les observations de trois systèmes d’observations satellitaire et du réseau de mesures traditionnelles de surface dans un système d’inversion bayésien variationnel sur des fenêtres temporelles longues cohérentes avec la durée de vie du méthane. Dans un premier temps, je montre que le réglage objectif des statistiques d’erreur de chaque système permet un bon accord entre les estimations des bilans annuels régionaux de méthane obtenus avec TANSO-FTS, IASI et le réseau de surface.Ce résultat permet d’envisager la combinaison de ces mesures pour mieux contraindre les estimations des émissions de méthane. En revanche, les résultats obtenus avec les mesures de SCIAMACHY alors en fin de vie, restent incohérents, probablement à cause d’une structure d’erreur difficile à modéliser. Dans un deuxième temps, je montre que le réseau de surface et IASI détectent la plupart des anomalies de flux de méthane en Afrique du Sud et en Asie de l’Est alors que TANSO-FTS détecte la quasi-totalité desanomalies en Afrique du Nord. De plus, négativement corrélées à l’humidité du sol, les anomalies récentes en Afrique du Nord et en Asie de l’Est suggèrent, respectivement, une augmentation des émissions des feux et un changement de pratique dans la riziculture.Je mets aussi en évidence une contribution majoritaire des terres des basses latitudes del’hémisphère nord à la tendance récente du méthane. / Methane is a powerful greenhouse gas with direct and indirect effect on global warming but its recent trend is misunderstood and still debated. My PhD aims at evaluating the ability of new satellite methane measurements to quantify the methane annual fluxes and their interannual variability. I assimilate the measurements of three satellite observing systems and the traditional observing surface network in a bayesian variational inversion system over long temporal windows consistent with the methane lifetime.First, I show that the tuning of input error statistics of each observing system allows a good agreement between the annual regional methane budgets inferred from TANSO-FTS, IASI and the surface network. This result opens the possibility to combine these measurements to better constrain the methane emission estimates. However, the results inferred from SCIAMACHY measurements acquired at the end of the life of the instrument, remain inconsistent, probably because of an error structure that is difficult to model.Secondly, I show that the surface network and IASI detect the main methane flux anomalies in South Africa and in East Asia whereas TANSO-FTS detects almost all the anomalies in North Africa. Moreover, negatively correlated with soil moisture, the recent anomalies observed in North Africa and in East Asia suggest, respectively, an increase of fire emissions and a change in rice culture practices. I also show that the lands over the Northern Hemisphere low latitudes have a major contribution to the recent methane trend.

[pt] Mais de três décadas atrás, Przemieniecki introduziu uma formulação para análise de elementos de barra e treliça baseada em uma expansão em série de freqüências. Recentemente esta formulação foi generalizada para análise de sistemas elásticos submetidos a carregamento qualquer e deslocamentos iniciais. Baseado no método da superposição modal, um sistema acoplado, com equações diferenciais de movimento de alta ordem, é transformado em um sistema desacoplado com equações diferenciais de segunda ordem, que pode ser resolvido por qualquer método conhecido na literatura. A motivação para este desenvolvimento é o Método Híbrido dos Elementos de Contorno, que tem sido desenvolvido para problemas dependentes do tempo e problemas dependentes da freqüência. Esta formulação, assim como a introduzida por Pian para o Método dos Elementos Finitos, obtém uma matriz de rigidez utilizando apenas integrais de contorno, para um domínio de forma qualquer contendo vários graus de liberdade. O uso de termos com freqüências de alta ordem melhora muito a precisão numérica. A análise modal de um problema dinâmico, conforme se apresenta, é aplicável a qualquer formulação de elementos finitos, em geral, desde que a matriz de rigidez generalizada possa ser obtida. Este trabalho é uma tentativa de consolidação da formulação teórica proposta, em que se faz uso de integrais exclusivamente no contorno, com a discussão de diversos casos particulares e a conseqüente avaliação numérica: estruturas restringidas ou não; consideração de deslocamentos e velocidades iniciais, tanto em termos de valores nodais quanto de campos prescritos no domínio (incluindo deslocamentos de corpo rígido); deslocamentos forçados dependentes do tempo; forças de massa dependentes do tempo; cálculo de resultados em pontos internos. Vários exemplos acadêmicos para problemas de potencial bidimensionais ilustram este trabalho. / [en] More than three decades ago, Przemieniecki introduced a formulation for the free vibration analysis of bar and beam elements based on a power series of frequencies. Recently, this formulation was generalized for the analysis of the dynamic response of elastic systems submitted to arbitrary nodal loads as well as initial displacements. Based on the mode-superposition method, a set of coupled, higher-order differential equations of motion is transformed into a set of uncoupled second order differential equations, which may be integrated by means of standard procedures. Motivation for this theoretical achievement is the hybrid boundary element method, which has been developed for time-dependent as well as frequency-dependent problems. This formulation, as a generalization of Pian`s previous achievements for finite elements, yields a stiffness matrix for which only boundary integrals are required, for arbitrary domain shapes and any number of degrees of freedom. The use of higher-order frequency terms drastically improves numerical accuracy. The introduced modal assessment of the dynamic problem is applicable to any kind of finite element for which a generalized stiffness matrix is available. The present work is an attempt of consolidating this boundary- only theoretical formulation, in which a series of particular cases are conceptually outlined and numerically assessed: Constrained and unconstrained structures; initial displacements and velocities as nodal values as well as prescribed domain fields (including rigid body movement); forced time-dependent displacements; time-dependent body forces; evaluation of results at internal points. Several academic examples for 2D problems of potential illustrate the formulation.

[pt] Este trabalho trata o Método Híbrido dos Elementos de Contorno com vista à análise de problemas que envolvam simetria ou antissimetria. Nestes casos, apenas uma parte da estrutura, que pode ser a metade, um quarto ou um oitavo, deve ser discretizada e capaz de representar o todo. Os métodos de contorno apresentam a vantagem, quando comparados com os de domínio, de não ser necessário nenhum tipo de discretização ao longo dos eixos ou planos de simetria, sem a introdução de mais aproximações, visto que apenas o contorno é discretizado. Embora estas simplificações venham a restringir alguns deslocamentos de corpo rígido (para problemas de elasticidade), no Método dos Elementos de Contorno convencional (colocação ou Galerkin) a ausência de tais deslocamentos não acarreta alterações na sistemática do método. Nos Métodos Híbridos de Elementos de Contorno, por outro lado, os deslocamentos de corpo rígido são necessários direta ou indiretamente para a aplicação de condições de ortogonalidade e avaliação das propriedades espectrais que são essenciais na obtenção da diagonal principal de certas matrizes inerentes ao método, tais como de flexibilidade, de deslocamentos e de tensões. Esta necessidade de avaliação é uma característica de suma importância do método e, quando não houver possibilidade de fazê-la, deve-se procurar uma forma substituta conceitualmente equivalente. Verifica-se que, apesar de este método ser baseado em funções singulares de Green, é capaz de representar estados simples de tensões, tanto por trabalhos virtuais quanto por interpolações no domínio. Como objetivo principal deste trabalho, será demonstrado que para cada deslocamento de corpo rígido perdido, devido às restrições impostas pela simetria ou antissimetria, poderá ser utilizado um estado simples de tensão (constantes na maioria dos casos), que permitirá o estabelecimento de propriedades espectrais apropriadas. De forma a se garantir uma sistemática estruturada para o trabalho, faz-se uma abordagem de conceitos fundamentais aplicados a problemas da elastostática e potencial estacionário, na formulação variacional do Método Híbrido dos Elementos de Contorno com posteriores considerações especiais de estados simples de tensão (representados polinomialmente), para elasticidade tridimensional em geral, visto que para problemas bidimensionais o caso se torna uma particularização. Todas as combinações de simetria e antissimetria são avaliadas com a implementação numérica. Diversos exemplos de problemas bidimensionais ilustram a formulação teórica. / [en] The boundary element methods are suited for the analysis of symmetric and antisymmetric problems - in which only a part (half, quadrant or octant) of the structure needs to be explicitly considered - since, as an additional advantage when compared with a domain discretization method, no interpolation is required along the symmetry axes (for 2D problems) or planes (for 3D problems) and, consequently, no approximations are introduced thereon. Although such computational simplification may prevent some of the structures allowable rigid body movements (elasticity problems considered), this fact may be completely ignored as concerning the implementation of the traditional (collocation or Galerkin) boundary element methods. In the hybrid boundary element methods, on the other hand, special orthogonality conditions, directly or indirectly related to rigid body displacements, are required for the evaluation of elements about the main diagonal of some matrices (flexibility, displacement and stress matrices). Then, a central issue in such methods is the assessment of these matrices spectral properties for any combination of symmetry and antisymmetry and, most important, the investigation of conceptually equivalent, substitutive properties. As presented in this work, the hybrid boundary element methods, although based on singular Green s functions, are able to simulate, in terms of both virtual work and field interpolation, the simplest stress states. Then, one demonstrates that for every missing rigid body displacement - brought about by some symmetry or antisymmetry consideration - one may lay hold of a simple (in most cases constant) stress state, which enables establishing appropriate spectral properties. This work introduces the underlying variational concepts of the hybrid boundary element method and outlines the special consideration of simple (polynomial) stress states, as generally formulated for 3D elasticity, since 2D elasticity and problems of potential may be dealt with as particular cases. All combinations of symmetry and antisymmetry are outlined with the aim of numerical implementation. A series of 2D examples for problems of potential illustrate the theoretical