No presente trabalho estudamos duas variações do problema de empacotamento de itens retangulares idênticos, permitindo rotações de 90 graus, em um poliedro. Uma variação consiste em encontrar a maior quantidade de itens retangulares idênticos que podem ser empacotados em um poliedro. A outra consiste em encontrar o poliedro de um determinado tipo com menor área para empacotar uma quantidade fixa de itens retangulares idênticos. Desenvolvemos restrições de eliminação de simetrias para estes problemas, o que tornou a resolução dos mesmos mais eficiente, por métodos do tipo branch-&-bound. Estudamos também o problema de corte no qual há uma determinada demanda (de itens) a ser cortada e um conjunto de objetos disponíveis. Desejamos satisfazer a demanda minimizando o custo dos objetos utilizados e, dentre as diferentes possibilidades de se fazer isso, desejamos aquela que maximize as sobras aproveitáveis. De forma geral, sobras aproveitáveis podem ser entendidas como regiões retangulares de um objeto que possuem altura e largura iguais ou superiores a de um item de referência e representam sobras do processo de corte que podem se tornar objetos e serem reaproveitadas em um novo procedimento de corte. Apresentamos modelos de otimização em dois níveis para duas variações do problema de corte com sobras aproveitáveis a saber: o problema de corte de itens retangulares em dois estágios e o problema de corte de itens retangulares não guilhotinado. Como formas de resolver os modelos propostos, apresentamos reformulações destes modelos de programação em dois níveis em modelos de programação inteira mista. Lidamos também com uma variação do problema de corte com sobras aproveitáveis considerando a minimização da quantidade de sobras. Aplicamos restrições de eliminação de simetrias aos modelos desenvolvidos para o problema de corte de itens retangulares com sobras aproveitáveis, a fim de resolver instâncias maiores, e desenvolvemos uma estratégia de solução alternativa para os modelos. Os modelos desenvolvidos foram implementados computacionalmente e fomos capazes de resolver instâncias pequenas dos problemas em questão. / In this work we study two variations of the packing problem where identical rectangular items must be packed into a polyhedron. One of the variations consists in finding the largest amount of rectangular items that can fit in a polyhedron. The other one consists in finding a minimal area polyhedron of a certain type that packs a set of rectangular identical items. We present some symmetry-breaking constraints that reduce the computational effort in solving those problems through a branch-&-bound method. We also studied the cutting stock problem where there are some items to be cut from a set of rectangular objects and we need to satisfy the demand of items to be cut minimizing the cost of the used objects and, among the different ways of doing this, we want that which maximize the usable leftovers. Loosely speaking,usable leftovers can be understood as rectangular regions in an object that has the width and the height greater than or equal to the ones of a reference item. These leftovers can be seen as leftovers from a cutting process that will become items in a new cutting process. We present bilevel programming models to two variations of this problem with usable leftovers: the two-stage cutting stock problem of rectangular items and the non-guillotine cutting stock problem of rectangular items. In order to solve the proposed models we present also MIP reformulations of these bilevel programming problem models. We also developed some symmetry breaking constraints in order to accelerate the solving process of those models. The developed models were computationally programmed and we were able to solve small instances of the proposed problems

The Enigma code was one of the most dangerous and effective weapons the Germans wielded at the outbreak of the Second World War. The Enigma machine was capable of encrypting radio messages that seemed virtually unbreakable. In fact, there were 158,900, 000,000,000 possible combinations in any given message transmitted. On the eve of the war’s outbreak, the British had recently learned that the Poles had made significant progress against this intimidating cipher in the early 1930s. Incensed and with little help, the British Government Code & Cipher School began the war searching for a solution. Drawing from their experiences from the First World War, and under the visionary guidance of Alan Turing, Gordon Welchman, and countless others, the British created a new, mechanical approach to breaking the seemingly impossible German code. By breaking the code, they could very well save Britain.

The SUPERTANK dataset is analyzed to examine the upper limit of beach change in response to elevated water level induced by wave runup. Thirty SUPERTANK runs are investigated, including both erosional and accretionary wave conditions under random and monochromatic waves. Two experiments, one under a spilling and one under a plunging breaker-type, from the Large-Scale Sediment Transport Facility (LSTF) are also analyzed. The upper limit of beach change approximately equals the maximum vertical excursion of swash runup. Exceptions to this direct relationship are those with beach or dune scarps when gravity-driven changes, i.e., avalanching, become significant. The vertical extent of wave runup, Rmax, above mean water level on a beach without a scarp is found to approximately equal the significant breaking wave height, Hbs. Therefore, a simple formula Rmax = Hbs is proposed. The linear relationship between maximum runup and breaking wave height is supported by a conceptual derivation. This predictive formula reproduced the measured runup from a large-scale 3-dimensional movable bed physical model. Beach and dune scarps substantially limit the uprush of swash motion, resulting in a much reduced maximum runup. Predictions of wave runup are not improved by including a slope-dependent surf-similarity parameter. The limit of wave runup is substantially less for monochromatic waves than for random waves, attributed to absence of low-frequency motion for monochromatic waves.

This research project addresses the substance abuse cycle intervention and prevention strategies used amongst individuals who are a part of generational substance abuse. Participants were recruited from an inpatient drug rehabilitation center in Southern California. Qualitative interviews were conducted in order to gain additional knowledge on this topic. The interviews were audio recorded, transcribed, and then analyzed by the researcher. Participants’ first hand experiences contributed to a better understanding of effective ways to address this cycle. The results consisted of the following themes: childhood substance abuse, environmental influence, Child Protective Services involvement, problem recognition, structure, counseling, relationships, higher power, positive attitude and hope for the future. Contributions to social work practice are discussed along with recommendations for future research. Findings were given to California State University, San Bernardino and were provided to the drug rehabilitation center used in this study.

Le projet ANR ACORFORMed, dans lequel s’inscrit cette thèse, a pour objectif la création (par des informaticiens) d’un agent conversationnel animé « patient » comme outil de formation à l’annonce, par la simulation et à l’aide d’un environnement virtuel. A l’aide de la méthodologie issue des études de la gestuelle et des apports de la littérature sur l’organisation des interactions, nous tentons de répondre à la question suivante : quelle est l’organisation structurelle globale de l’interaction patient-médecin, lorsque ce dernier se forme à l’annonce d’un dommage associé aux soins ? Les analyses menées dans cette thèse nous permettent de décrire l’interaction patient/médecin lors de formations à l’annonce en proposant différentes phases composants l’interaction, ainsi que des précisions quant à leur découpage et leurs articulations. Elles constituent une base indispensable et utilisable par les informaticiens pour concevoir et implémenter un agent conversationnel « patient » crédible qui pourra être utilisé dans la formation des médecins. Située au coeur d’un projet interdisciplinaire, cette thèse en linguistique permet donc de transposer les pratiques interactionnelles des médecins en vue de l’implémentation d’un agent virtuel par des informaticiens. / The ACORFORMed ANR project, in which this PhD project is integrated, aims at the creation (by computer scientists) of a "patient" animated conversational agent as a training tool for announcing, simulating and using a virtual environment. Using the methodology resulting from gestures studies and contributions of the literature on the organisation of interactions, we try to answer the following question: what is the overall structural organisation of the doctor/patient interaction, when this last is training to break the news of a damage associated with care? The analyses carried out in this thesis allow us to describe the doctor/patient interaction during training to break the news by proposing different phases that make up the interaction, as well as details about their division and their articulations. They are an indispensable and usable base for computer scientists to design and implement a credible "patient" conversational agent that can be used in physician training. Located at the heart of an interdisciplinary project, this thesis in linguistics makes it possible to transpose the interactional practices of physicians with a view to the implementation of a virtual agent by computer scientists.

This study reports on a new formulation of the spectral dissipation source term Sds for wind-wave modelling applications. This new form of Sds features a nonlinear dependence on the local wave spectrum, expressed in terms of the azimuthally integrated saturation parameter B(k)=k^4 F(k). The basic form of this saturation-dependent Sds is based on a new framework for the onset of deep-water wave breaking due to the nonlinear modulation of wave groups. The new form of Sds is succesfully validated through numerical experiments that include exact nonlinear computations of fetch-limited wind-wave evolution and hindcasts of two-dimensional wave fields made with an operational wind-wave model. The newly-proposed form of Sds generates integral spectral parameters that agree more closely with observations when compared to other dissipation source terms used in state-of-the-art wind-wave models. It also provides more flexibility in controlling properties of the wave spectrum within the high wavenumber range. Tests using a variety of wind speeds, three commonly-used wind input source functions and two alternative full-development evolution limits further demonstrate the robustness and flexibility of the new saturation-dependent dissipation source term. Finally, improved wave hindcasts obtained with an implementation of the new form of Sds in a version of the WAM model demonstrate its potential usefulness in operational wind-wave forecasting applications.

Le Modèle Standard de la physique des particules a été remarquablement testé auprès des grands accélérateurs jusqu'à une énergie de l'ordre du TeV. Malgré ses succès, un grand nombre de problèmes restent sans solution. Par exemple, le Modèle Standard ne permet pas d'expliquer l'origine de la prépondérance de la matière sur l'antimatière, ni l'origine du rapport 1016 entre l'échelle de brisure électrofaible et l'échelle de Planck. Les interactions électrofaibles et fortes ne sont pas unifiées et de plus, la gravité est absente. Cette thèse s'interesse à différents aspects phénoménologiques des théories d'extension du Modèle Standard. La première partie porte sur la baryogénèse électro- faible dans le Modèle Standard Supersymétrique Minimal et plus spécifiquement sur la dynamique de la transition de phase électrofaible. La deuxième partie est consacrée au problème de la brisure spontanée de la supergravité dans les théories de cordes effectives, les seules théories offrant l'espoir d'une description unifiée des interactions gravitationnelles avec les autres forces fondamentales. Enfin, la dernière partie discute un certain nombre de modèles phénoménologiques utilisant les propriétés des branes et la présence de dimensions supplémentaires afin de reformuler le problème de hiérarchie de jauge en termes géométriques.

The thesis presents the results of our research on symmetry for some semilinear elliptic problems and on existence of solution for quasilinear problems involving singularities. The text is composed of two parts, each of which begins with a specific introduction. The first part is devoted to symmetry and symmetry-breaking results. We study a class of partial differential equations involving radial weights on balls, annuli or $R^N$ --where these weights are unbounded--. We show in particular that on unbounded domains, focusing on symmetric functions permits to recover compactness, which implies existence of solutions. Then, we stress the fact that symmetry-breaking occurs on bounded domains, depending both on the weights and on the nonlinearity of the equation. We also show that for the considered class of problems, the multibumps-solution phenomenon appears on the annulus as well as on the ball. The second part of the thesis is devoted to partial and ordinary differential equations with singularities. Using concentration-compactness tools, we show that a rather large class of functionals is lower semi-continuous, leading to the existence of a ground state solution. We also focus on the unicity of solutions for such a class of problems.

Monoatomic chain formation for Ag, Au, Pd and Pt has been investigated using a model for the tip structure. First-principles calculations, mostly spin polarized, were performed within the framework of the Density Functional theory. Results are presented and discussed on the basis of the electronic structure. Tendencies for chain formation were noted for Ag, Au and Pt.

The work presented in this thesis is dedicated to developing inexpensive quantum-chemical models that are able to produce smooth and physically correct potential energy curves for the dissociation of single covalent bonds. It is well known that the energies produced by many ab initio theories scaling as the fifth order with the system size (for instance, second-order Moller-Plesset (MP2) and Epstein-Nesbet perturbation theories) diverge at large interatomic separations. We show that the divergent behavior of such perturbation schemes is due to a small number of terms in the energy expressions. Then, we demonstrate that the self-consistent replacement of these terms by their analogs from the coupled cluster theory (such as CCSD) allows one to redress the erroneous behavior of the perturbation theories without the damage to the overall scaling. We also investigate the accuracy of these hybrid perturbation theory-coupled cluster theories near equilibrium geometry. Judging from the computed spectroscopic constants and shapes of the potential energy curves, one such model, denoted MP2-CCSD(II) in this work, performs consistently better than the MP2 theory at essentially the same computational cost.