No presente trabalho estudam-se diferentes técnicas explícitas, em diferenças finitas, no emprego de algoritmos do tipo velocidade-pressão para simulação de escoamentos incompressíveis. O método de resolução da pressão de maneira direta e explícita, introduzido por Bravo e Claeyssen [BRA 97a], é analisado. Faz-se uma aproximação para o erro causado por esta técnica, e verifica-se como isto afeta a equação da continuidade. As simulações são realizadas na cavidade quadrada, comparando-se os diferentes métodos e validando as aproximações realizadas no estudo do método de resolução da pressão. Além disso, simula-se o escoamento em cavidades profundas e rasas, observandose a formação de vórtices e distribuição de energia cinética. Simulações do escoamento na cavidade cúbica também são apresentadas. / In this work different explicit technics in finite defferences in the application of velocity-pressure algorithm to simulate incompressible flows have been studied. The direct and explicit method of pressure resolution, introduced by Bravo anel Claeyssen [BRA 97a] is analyzed. An approximation to the error caused by this method is made, anel how this affects the continuity equation is verified. The simulations are maele in a square cavity, comparing the differents methoels anel valielating the approximations maele in the study of the pressure resolution method. Besieles this, flow in eleep and shallow cavities is simulateel, observing the formation of vortices and kinetic energy distribution. Simulations of the flow in the cubical cavity are also considered.

Les bassins de retenue des eaux pluviales sont utilisés pour préserver la qualité des eaux réceptrices par sédimentation pendant le temps de pluie. Cependant, les efficacités du bassins n'étaient pas satisfaisants en raison de la mal compréhension du processus de sédimentation. Afin de mieux comprendre ces processus dans des ouvrages in situ, cette thèse porte à la fois sur des expérimentations in situ et sur les modélisations de l'hydrodynamique et du transport particulaire dans les bassins de retenue pilotes et in situ. Cette recherche s’est appuyée en grande partie sur le bassin Django Reinhardt (BDR) à Chassieu (volume: 32000 m3, surface: 11000 m2) dans le cadre de l’OTHU et sur les données expérimentales obtenues par Dufresne (2008) et Vosswinkel et al. (2012). Les échantillons de sédiments ont été prélevés et leurs caractéristiques physiques ont été analysées en laboratoire dans le but de cerner leur distribution spatiale. Concernant la modélisation numérique, les simulations de l’hydrodynamique en régime permanent ont été réalisées à l'aide du logiciel CFD Fluent et ont été évaluées à partir de l’analyse de corrélation entre le comportement hydrodynamique du bassin et la distribution spatiale des caractéristiques physiques des sédiments. Les conditions limites sur le fond couramment utilisées et largement décrites dans la littérature ont été testées dans le but de représenter la distribution spatiale des sédiments et l’efficacité de décantation du BDR. Les conditions testées sont : i) contrainte de cisaillement critique ou bed shear stress – BSS et ii) énergie cinétique turbulente critique ou bed turbulent kinetic energy - BTKE. L’approche Euler-Lagrange dite « particle tracking » a été mise en œuvre. En raison de l'échec de prédiction des zones de dépôt à l’aide des conditions limites disponibles (BSS et BTKE), une nouvelle relation a été proposée pour estimer le seuil BTKE. La condition à la limite obtenue en utilisant cette nouvelle relation a été testée sur un bassin pilote (Dufresne, 2008) et sur le BDR en régime permanent. Les résultats obtenus n’étaient pas très satisfaisants concernant la prédiction des zones de dépôt et l’efficacité de décantation dans le bassin BDR, même en considérant une distribution granulométrique non uniforme. Afin de mieux prédire les zones de dépôt dans le BDR, une nouvelle méthode a été proposée en considérant le transport des particules, leur décantation et leur érosion en régime transitoire. Sur la base de la méthode proposée pour le transport des particules, la décantation et l'érosion en régime transitoire, plusieurs modélisations avec différentes conditions limites ont été réalisées dans un bassin de retenue pilote rectangulaire (Vosswinkel et al., 2012). Les prédictions des efficacités et des zones de dépôt en régime transitoire avec la méthode proposée sont satisfaisantes / Stormwater detention basins are used to preserve the quality of receiving waters by sedimentation during the wet weather. However, the removal efficiencies of basin were not satisfactory due to the not well understanding of the sedimentation processes. In order to further understand these processes in the real facilities, this thesis therefore focuses both on in situ experiments and modeling of the hydrodynamic and sediment transport in field detention basin and in small scale basin in laboratory. This research was supported by large part on the Django Reinhardt basin (DRB) in Chassieu within the OTHU program and the experimental data deriving from Dufresne (2008) and Vosswinkel et al. (2012). Samples of sediments accumulated in the basin were collected and their physical characteristics were analyzed in order to determine their spatial distribution. Concerning numerical modeling, the hydrodynamic simulations in steady state were performed using CFD software Fluent and were evaluated by the correlation analysis between the hydrodynamic behavior of DRB and the spatial distribution of the physical characteristics of sediments. The bed boundary conditions used in the literatures were tested in order to represent the spatial distribution of sediments and removal efficiency of DRB. The conditions tested were: i) critical bed shear stress - BSS and ii) critical bed turbulent kinetic energy - BTKE. Because of the failure prediction of DRB deposit zones with usual bed boundary conditions, a new relationship based on particle settling velocities has been proposed to estimate the BTKE threshold for the bed boundary condition. The proposed boundary condition was tested in a pilot basin (Dufresne, 2008) and the DRB using the Euler-Lagrange approach under steady flow conditions. The results were not very satisfactory regarding the DRB deposit zones, even considering non-uniform grain size. In order to better predict the deposit zones and settling efficiency in field detention basins, a new method has been proposed accounting for the sediment transport, settling and erosion under unsteady conditions. Based on this proposed method for representing the particle transport, settling and erosion processes under unsteady conditions, various simulations with different bed boundary conditions were carried out in a pilot rectangular basin (Vosswinkel et al., 2012). The predictions of removal efficiencies and deposition zones are satisfactory. Hence, taking into account transient effects on both hydrodynamics and sediment transport leads to drastically improve the spatial and temporal distributions of sediments in settling detention basins.

The structure of exotic nuclei is one of the main interests in current nuclear physics. Exotic nuclei present unusual properties, such as a low breakup energy, a short lifetime and/or a halo structure. Because of their short lifetimes, they can not be studied by usual spectroscopic techniques. Indeed, targets of such nuclei are impossible to build. But since the availability of radioactive beams, nuclear reactions have provided possibilities of exploring nuclei far from stability.<p><p>The investigation of exotic nuclei has been recently reactivated by the development of intense radioactive nuclear beams. As firstly observed for the deuteron, and then for other exotic projectiles such as $^6$He and $^{11}$Be, the internal structures of the interacting nuclei can have a significant effect on the elastic cross sections. Due to their low binding energy, the projectile dissociation process, leaving the target in its ground state, highly affects elastic cross sections but also other measurements such as transfer and fusion reactions. Accurate reaction theories are therefore needed. The coupled discretized-continuum channel (CDCC) method is one of those theories and assumes a projectile made of N clusters (usually N=2 or 3) impinging on a target which is structureless. The N+1-body Schrödinger equation is approximately solved by expanding the total wave function over the bound and continuum states of the projectile. These latter take into account the dissociation events and are approximately described by a truncated set of square-integrable wave functions. There are two available methods for discretizing the continuum, the pseudostate method where the projectile Hamiltonian is diagonalized within a finite basis of square-integrable functions, or the bin method where exact scattering wave functions of the projectile are averaged over bins in a finite region of space. In both cases, the N+1-body Schrödinger equation is replaced by a set of coupled-channel differential equations, which provides the physical quantities such as the collision matrix. In principle, the CDCC method can be very close to the exact N+1-body wave function and is adapted to low as well as to high energy reactions. However, its main interest consists in the low-energy domain.<p><p>In the present work, we propose a new approach to solve the CDCC equations. This method is based on the R-matrix theory associated with a Lagrange mesh basis. We will show that the combination of both approaches provides a fast and accurate technique to solve the CDCC equations, even for large systems, where traditional methods meet convergence problems. Before investigating collisions with exotic projectiles, we restrict ourselves to the simplest nucleus, the deuteron. Then we make a step towards a more complicated system, the $^6$Li which is a well known stable nucleus. We apply the CDCC method to the d + $^{58}$Ni and $^6$Li + $^{40}$Ca elastic scattering and breakup. These systems are considered in the literature as test cases. They have been investigated by several authors who showed the importance of the breakup channels in the elastic cross sections.<p><p>After having validated the present version of the CDCC method, we focus on $^{11}$Be, a typical example of a halo nucleus, with low binding energy and large quadrupole moment. Elastic, inelastic and breakup cross sections are computed in the CDCC formalism, at energies near the Coulomb barrier, where continuum effects in the scattering of exotic nuclei, and more specifically on the $^{11}$Be + $^{64}$Zn scattering, are observed. We show that converged cross sections need high angular momenta as well as large excitation energies in the wave functions of the projectile.<p><p>A Borromean nucleus is made of three constituents which are weakly linked together, but where each pair of those three constituents does not form a bound system. The name "Borromean" comes from the Borromean rings where, if any one of three rings is removed, the remaining two become unbound. Collisions with $^6$He and $^9$Be Borromean projectiles are studied in the present work. Again we compare our method with the $^6$He + $^{208}$Pb and $^6$He + $^{12}$C benchmark calculations. Afterwards, the convergence against the parameters of the description of the $^9$Be projectile is tested for the elastic cross section. The sensitivity to the technique employed to remove the forbidden states and also the sensitivity to the collision energy are investigated. / Doctorat en Sciences de l'ingénieur / info:eu-repo/semantics/nonPublished

<p align="justify">Les fonctions de Lagrange sont des fonctions indéfiniment dérivables qui s'annulent en tous les points d'un réseau sauf un. Ces fonctions sont utilisées comme fonctions de base d'un calcul variationnel. Les éléments de matrice de ce calcul sont évalués à l'aide de la règle de quadrature de Gauss définie par le réseau de points. Les équations à résoudre prennent ainsi la forme d'équations sur réseau.</p><p><p align="justify">La méthode des réseaux de Lagrange allie simplicité et précision. La matrice représentant le potentiel est diagonale et ne dépend que des valeurs prises par le potentiel aux points du réseau. Contrairement à la méthode des différences finies, une expression analytique est obtenue pour la solution. Nous cherchons clans cette thèse à cerner les avantages et inconvénients de la méthode des réseaux de Lagrange, ainsi qu'à étendre son champ d'application en mécanique quantique. Nous montrons notamment que cette méthode peut être reliée à d'autres méthodes sur réseau, telles que les méthodes de la variable discrétisée (DVR) ou du réseau de Fourier, qui sont fort utilisées en physique atomique et moléculaire.</p><p><p align="justify">Dans les problèmes à deux corps, nous appliquons la méthode à l'étude des états liés et nous l'étendons au cas des collisions, c'est-à-dire aux états libres. Une nouvelle technique de calcul de la longueur de diffusion et de la portée effective est également considérée. Dans certains cas, la solution exacte du problème à deux corps existe sous forme analytique, ce qui permet une étude de la précision de la méthode en ce qui concerne les valeurs propres et les vecteurs propres de la matrice hamiltonienne. L'extension de la méthode aux problèmes à deux corps régis par une dynamique semi-relativiste est également examinée.</p><p><p align="justify">Dans le cas des problèmes à trois corps, nous effectuons une comparaison entre plusieurs systèmes de coordonnées auxquels sont couplés différents réseaux de Lagrange. Les résultats de cette comparaison dépendent de la présence de singularités dans les potentiels, celles-ci pouvant limiter fortement la précision de la méthode.</p><p><p align="justify">En physique nucléaire, nous comparons deux approches sur réseaux de Lagrange lors de l'étude de l'état fondamental du noyau 6He. Il s'agit d'un noyau à halo de neutrons, pour lequel il existe une forte probabilité de trouver deux des neutrons loin des autres nucléons. Le noyau 6He peut ainsi être traité comme un système à trois corps, constitué d'une particule alpha et de deux neutrons. Nous étendons également le modèle à trois corps pour ce noyau au cas d'interactions à deux corps plus générales, c'est-à-dire contenant différents opérateurs agissant sur les spins des nucléons.</p><p><p align="justify">En physique atomique et moléculaire, où les interactions sont, en première approximation, purement coulombiennes, nous nous sommes intéressé aux états S et P des principaux systèmes à trois corps que sont l'atome d'hélium He, les ions hydrogène H-et positronium Ps-, l'ion moléculaire d'hydrogène HZ et la molécule muonique dt"mu". Les fonctions d'onde approchées obtenues lors de la détermination des états liés sont utilisées pour évaluer des rayons quadratiques moyens et les rayons de masse de ces systèmes.</p><p> / Doctorat en sciences appliquées / info:eu-repo/semantics/nonPublished

Notre étude porte sur des matériaux à effets relativistes importants. L'hamiltonien semi-relativiste, couplé aux équations de Maxwell (EM), permet de déduire les sources de courant et de densité, incluant des termes de second ordre (polarisations de spin et de Darwin). Différents modèles sont développés par expansion des EM. L'étude ab initio montre que (1) le désordre atomique peut produire le ferrimagnétisme (FM) dans GaFeO3 (GFO) multiferroïque, (2) les états 3d Fe des octaèdres déformés ont une levée de dégénérescence tétraédrique (théorie du champ cristallin), (3) la polarisation électrique concorde avec l'expérience, (4) le mécanisme magnétoélectrique (ME) direct est insuffisant pour expliquer le ME observé. Pour Cr2O3, le calcul de l'état massif sous contraintes biaxiales n'explique pas son FM, la taille de l'échantillon ou l'excès d'oxygène pourrait de fait être important. Enfin, nous avons développé le XAS et le XMCD dans le code VASP et calculé ces spectres pour GFO. / We studied the physics of materials where relativistic effects are important. We first coupled the semi-relativistic Hamiltonian with the Maxwell's equations, and derived the current and density sources, which included second-order terms like the spin and Darwin polarizations. Different models were developed, by expanding the Maxwell's equations. We then performed ab initio studies to explain (1) site disorders as the origin of ferrimagnetism in multiferroic GaFeO3 (GFO), (2) crystal-field theory where the Fe 3d states at the deformed octahedra have tetrahedral splittings, (3) the electric polarization as a function of temperature, and (4) the insufficiency of the direct magnetoelectric (ME) mechanism to explain observed ME behavior. For Cr2O3, bulk calculations for different biaxial strains failed to explain its ferromagnetism, indicating that size or excess-O effects might be important. Finally, we implemented XAS and XMCD in VASP and computed these spectra for GFO.

This diploma thesis deals with separation of backgroud and moving objects in video. Video can be represented as series of frames and each frame represented as low - rank structure - matrix. This thesis describe sparse representation of signals and robust principal component analysis. It also presents and implements algorithms - models for reconstruction of real video.

This diploma thesis deals with separation of backgroud and moving objects in video. Video can be represented as series of frames and each frame represented as low - rank structure - matrix. This thesis describe sparse representation of signals and robust principal component analysis. It also presents and implements algorithms - models for reconstruction of real video.

This diploma thesis deals with separation of backgroud and moving objects in video. Video can be represented as series of frames and each frame represented as low - rank structure - matrix. This thesis describe sparse representation of signals and robust principal component analysis. It also presents and implements algorithms - models for reconstruction of real video.

This thesis is split up into two parts each revolving around Floer homology and quantum cohomology of closed monotone symplectic manifolds. In the first part we consider symplectic manifolds obtained by symplectic reduction. Our main result is that a quantum version of an abelianization formula of Martin holds, which relates the quantum cohomologies of symplectic quotients by a group and by its maximal torus. Also we show a quantum version of the Leray-Hirsch theorem for Floer homology of Lagrangian intersections in the quotient. The second part is devoted to Floer homology of a pair of monotone Lagrangian submanifolds in clean intersection. Under these assumptions the symplectic action functional is degenerated. Nevertheless Frauenfelder defines a version of Floer homology, which is in a certain sense an infinite dimensional analogon of Morse-Bott homology. Via natural filtrations on the chain level we were able to define two spectral sequences which serve as a tool to compute Floer homology. We show how these are used to obtain new intersection results for simply connected Lagrangians in the product of two complex projective spaces. The link between both parts is that in the background the same technical methods are applied; namely the theory of holomorphic strips with boundary on Lagrangians in clean intersection. Since all our constructions rely heavily on these methods we also give a detailed account of this theory although in principle many results are not new or require only straight forward generalizations.:1. Introduction 2. Overview of the main results 2.1. Abelianization . 2.2. Quantum Leray-Hirsch theorem 2.3. Floer homology of Lagrangians in clean intersection 3. Background 3.1. Symplectic geometry . 3.2. Hamiltonian action functional 3.3. Morse homology . 3.4. Floer homology 4. Asymptotic analysis 4.1. Main statement . 4.2. Mean-value inequality . 4.3. Isoperimetric inequality 4.4. Linear theory 4.5. Proofs 5. Compactness 5.1. Cauchy-Riemann-Floer equation . 5.2. Local convergence . 5.3. Convergence on the ends 5.4. Minimal energy . 5.5. Action, energy and index estimates 6. Fredholm Theory 6.1. Banach manifold . 6.2. Linear theory 7. Transversality 7.1. Setup 7.2. R-dependent structures 7.3. R-invariant structures . 7.4. Regular points . 7.5. Floer’s ε-norm . 8. Gluing 8.1. Setup and main statement 8.2. Pregluing . 8.3. A uniform bounded right inverse 8.4. Quadratic estimate 8.5. Continuity of the gluing map 8.6. Surjectivity of the gluing map 8.7. Degree of the gluing map 8.8. Morse gluing . 9. Orientations 9.1. Preliminaries and notation 9.2. Spin structures and relative spin structures 9.3. Orientation of caps 9.4. Linear theory . 10.Pearl homology 10.1. Overview . 10.2. Pearl trajectories . 10.3. Invariance . 10.4. Spectral sequences 11.Proofs of the main results 11.1. Abelianization Theorem 11.2. Quantum Leray-Hirsch Theorem . 12.Applications 12.1. Quantum cohomology of the complex Grassmannian 12.2. Lagrangian spheres in symplectic quotients A. Estimates A.1. Derivative of the exponential map A.2. Parallel Transport A.3. Estimates for strips B. Operators on Hilbert spaces B.1. Spectral gap B.2. Flow operator C. Viterbo index D. Quotients of principal bundles by maximal tori D.1. Compact Lie groups D.2. The cohomology of the quotient of principle bundles by maximal tori

Die Arbeit behandelt Methoden zur Analyse bzw. Synthese elektromechanischer Systeme mit endlichem Freiheitsgrad (EMS). Dabei wird von einer einheitlichen mathematischen Modellierung solcher Systeme basierend auf dem Prinzip der virtuellen Arbeit in Lagrange'scher Fassung ausgegangen. Als Analysemethoden für strukturfeste EMS werden neben der numerischen Integration die Bestimmung von Gleichgewichtszuständen und die Herleitung der linearisierten Gleichungen zur Schwingungsanalyse dargelegt. Auf die Analyse von strukturvariablen EMS wird ausführlich eingegangen. Dazu werden Phänomene der Strukturvariabilität domänenunabhängig als unilaterale Bindungen aufgefasst und mittels komplementärer Variablen beschrieben. Die kombinatorische Aufgabe der Strukturfindung wird mittels eines linearen Komplementaritätsproblems gelöst. Die Synthese eines EMS wird als inverses Problem der Dynamik aufgefasst. Bei fester Gesamtkonfiguration führt das auf die nichtlineare dynamische Steuerung solcher Systeme. Dazu wird ein so genannter erweiterter PD-Regler - bestehend aus einer nichtlinearen Vorsteuerung auf Basis der inversen Dynamik des EMS und einer linearen Rückführung des Lage- und Geschwindigkeitsfehlers - entworfen. Die globale asymptotische Stabilität dieses Regelgesetzes wird durch explizite Konstruktion einer Lyapunov-Funktion nachgewiesen. Einige Beispiele zur Anwendung der aufgeführten Analyse- und Synthesemethoden runden die Arbeit ab. The thesis deals with methods for analysis and synthesis of electromechanical systems with finite degrees of freedom (EMS). Starting point is a unified mathematical approach to modelling such systems based on the principle of virtual work in Lagrange's formulation. Numerical integration, determination of equilibrium states and derivation of linearized equations are used as analytical methods for EMS with fixed structure. Electromechanical systems with variable structure are regarded explicitly. Phenomena of structural variability are comprehended as unilateral constraints and described using complementary variables. The combinatorial task of finding a valid structure is solved using a linear complementarity problem. The synthesis of EMS is understood as an inverse task of dynamics. Using a fixed configuration, this approach leads to non-linear dynamic control of such systems. A so-called augmented PD-controller - consisting, on the one hand, of a non-linear feedforward based on inverse dynamics of the EMS and, on the other hand, of a linear feedback using position and velocity errors - is designed. Global asymptotic stability is proven by explicit construction of a Lyapunov-function. Some examples showing the usage of the corresponding analytical and synthetic methods are given.