<div>Enhancements of gas turbine engines efficiency are critical for the development of the next generation of clean and efficient aircraft. With the increase in combustion temperatures, cooling of the turbine blades poses one of the most important thermal management issues. The current and most adopted solution is to flow cooling air bled from the compressor through channels inside turbine blades. Fuel preheating, meant to increase combustion efficiency, could be used to cool such air flow in fuel-air heat exchangers. However, when fuel thermodynamic states approach supercritical pressures and temperatures, large amplitude oscillations have been known to occur with catastrophic hardware failures. For this reason, the use of supercritical fuels in fuel-air heat exchangers has been avoided, thereby reducing the fuel's cooling potential and the overall efficiency of the aircraft. Engine manufacturers desire a model capable of predicting the onset of such disruptive thermoacoustic oscillations. To this goal, we study theoretically and numerically transcritical thermoacoustic oscillations, i.e., thermoacoustic instabilities manifesting themselves when a fluid is heated close to its critical point, where abrupt changes of thermodynamic properties appear. Details of this work will be on the development of a transcritical thermoacoustic theory and on numerical results from linear stability analysis and high-fidelity Navier-Stokes simulations. Meeting the needs of industry and with the intent of pushing technological and scientific barriers, we propose to exploit such powerful oscillations for energy conversion through the use of the first-ever-built transcritical thermoacoustic engine.</div>

This work considers the synthesis of brass instrument sounds using time-domain numerical methods. The operation of such a brass instrument is as follows. The player's lips are set into motion by forcing air through them, which in turn creates a pressure disturbance in the instrument mouthpiece. These disturbances produce waves that propagate along the air column, here described using one spatial dimension, to set up a series of resonances that interact with the vibrating lips of the player. Accurate description of these resonances requires the inclusion of attenuation of the wave during propagation, due to the boundary layer effects in the tube, along with how sound radiates from the instrument. A musically interesting instrument must also be flexible in the control of the available resonances, achieved, for example, by the manipulation of valves in trumpet-like instruments. These features are incorporated into a synthesis framework that allows the user to design and play a virtual instrument. This is all achieved using the finite-difference time-domain method. Robustness of simulations is vital, so a global energy measure is employed, where possible, to ensure numerical stability of the algorithms. A new passive model of viscothermal losses is proposed using tools from electrical network theory. An embedded system is also presented that couples a one-dimensional tube to the three-dimensional wave equation to model sound radiation. Additional control of the instrument using a simple lip model as well a time varying valve model to modify the instrument resonances is presented and the range of the virtual instrument is explored. Looking towards extensions of this tool, three nonlinear propagation models are compared, and differences related to distortion and response to changing bore profiles are highlighted. A preliminary experimental investigation into the effects of partially open valve configurations is also performed.

abstract: Coordination and control of Intelligent Agents as a team is considered in this thesis. Intelligent agents learn from experiences, and in times of uncertainty use the knowl- edge acquired to make decisions and accomplish their individual or team objectives. Agent objectives are defined using cost functions designed uniquely for the collective task being performed. Individual agent costs are coupled in such a way that group ob- jective is attained while minimizing individual costs. Information Asymmetry refers to situations where interacting agents have no knowledge or partial knowledge of cost functions of other agents. By virtue of their intelligence, i.e., by learning from past experiences agents learn cost functions of other agents, predict their responses and act adaptively to accomplish the team’s goal. Algorithms that agents use for learning others’ cost functions are called Learn- ing Algorithms, and algorithms agents use for computing actuation (control) which drives them towards their goal and minimize their cost functions are called Control Algorithms. Typically knowledge acquired using learning algorithms is used in con- trol algorithms for computing control signals. Learning and control algorithms are designed in such a way that the multi-agent system as a whole remains stable during learning and later at an equilibrium. An equilibrium is defined as the event/point where cost functions of all agents are optimized simultaneously. Cost functions are designed so that the equilibrium coincides with the goal state multi-agent system as a whole is trying to reach. In collective load transport, two or more agents (robots) carry a load from point A to point B in space. Robots could have different control preferences, for example, different actuation abilities, however, are still required to coordinate and perform load transport. Control preferences for each robot are characterized using a scalar parameter θ i unique to the robot being considered and unknown to other robots. With the aid of state and control input observations, agents learn control preferences of other agents, optimize individual costs and drive the multi-agent system to a goal state. Two learning and Control algorithms are presented. In the first algorithm(LCA- 1), an existing work, each agent optimizes a cost function similar to 1-step receding horizon optimal control problem for control. LCA-1 uses recursive least squares as the learning algorithm and guarantees complete learning in two time steps. LCA-1 is experimentally verified as part of this thesis. A novel learning and control algorithm (LCA-2) is proposed and verified in sim- ulations and on hardware. In LCA-2, each agent solves an infinite horizon linear quadratic regulator (LQR) problem for computing control. LCA-2 uses a learning al- gorithm similar to line search methods, and guarantees learning convergence to true values asymptotically. Simulations and hardware implementation show that the LCA-2 is stable for a variety of systems. Load transport is demonstrated using both the algorithms. Ex- periments running algorithm LCA-2 are able to resist disturbances and balance the assumed load better compared to LCA-1. / Dissertation/Thesis / Masters Thesis Electrical Engineering 2018

O principal objetivo deste trabalho é apresentar técnicas de solução para equações não lineares. Especificamente, consideramos equações compostas por funções elementares, dentre elas polinomiais, racionais, trigonométricas, exponenciais e logarítmicas, e por operações algébricas de soma, subtração, multiplicação, divisão, potência e raiz. Exploramos técnicas de resolução analítica e numérica. Como não existem fórmulas resolventes de extensão geral, a técnica analítica consiste em aplicar operações elementares que nos levam a equações equivalentes (que têm a mesma solução) até que se consiga uma equação simples, de fácil resolução. Os métodos numéricos abrangem um conjunto maior de equações e obtêm uma aproximação para a solução por meio de um processo que gera uma sequência de aproximações. Entre os métodos numéricos estudados estão Bissecção, de Newton, das Secantes e do Ponto Fixo (ou Iteração Linear). Recursos Computacionais como calculadora, planilha eletrônica e o software Maxima foram utilizados com objetivo de automatizar os cálculos, tornando essa tarefa mais rápida, e também buscando extrair informações adicionais do processo de resolução como criar tabelas e traçar gráficos. Realizamos testes numéricos com equações de diversos graus de dificuldade. Observamos as vantagens, as desvantagens e as limitações de cada método e de cada recurso. / The goal of this work is to present solution techniques for nonlinear equations. Specifically, we consider equations compounded of elementary functions, among them polynomials, rational, trigonometric, exponential and logarithmic, and of algebraic operations of addition, subtraction, multiplication, division, power and root. We explore analytical and numerical resolution techniques. Since there are no general resolvent formulas, the analytic technique consists of applying elementary operations that lead to equivalent equations (which have the same solution) until a simple and easily to solve equation is obtained. Numerical methods cover a larger set of equations and obtain an approximation to the solution by a process which generates a sequence of approximations. Among the numerical methods we studied Bisection, Newton, Secant and Fixed Point (or Linear Iteration) methods. Computational resources such as calculator, spreadsheet and the software Maxima were used in order to automate calculations, making this task faster, as well as seeking for additional information from the resolution process, such as creating tables and graphics. We perform numerical tests, with equations of varying degrees of difficulty. We note the advantages, disadvantages and limitations of each method and resource.

Les relevés à venir comme Euclid, LSST et WFIRST vont nous ouvrir la perspective d’étudier l’univers profond. Pour ces grands relevés, l’astigmatisme cosmique correspond à une sonde indispensable pour étudier la nature de l’énergie noire et la matière noire. Compte tenu de la précision attendue par ces observations, nous devons faire des prédictions basées sur des simulations correspondant à l’état de l’art afin de quantifier avec précision la variance, les biais et les dégénérescences potentielles liés aux baryons. Dans ce contexte, ma thèse se focalise sur la construction d’estimateurs précis basés sur les observables de lentillage. La première partie de ma thèse consiste à caractériser la géométrie des grandes structures par astigmatisme cosmique (Gouin et al. 2017). Une décomposition multipolaire du signal est appliquée afin de quantifier la distribution azimutale de la matière noire, centrée sur les amas. Les propriétés statistiques de ces moments sont estimées à partir d’une simulation cosmologique. Les distorsions harmoniques calculées dans le voisinage des amas tracent la structure filamentaire. Un plus grand nombre de filaments semblent connectés aux amas de forte masse. Dans la dernière partie de ma thèse, je synthétise le signal d’astigmatisme cosmique dans le cône de lumière de la simulation Horizon AGN. Pour ce faire, je propage les rayons de lumière le long du cône dans l’approximation des plans de lentillage multiples. L’effet des baryons est significatif dans la statistique du cisaillement aux échelles angulaires inférieures à l’arc-minute. Le signal de cisaillement galaxie-galaxie est comparée aux observations récentes, et semble être en bon accord. / Upcoming weak lensing surveys such as Euclid, LSST and WFIRST will provide an unprecedented opportunity to investigate the dark Universe. Through these large scale surveys, gravitational lensing is an indispensable cosmological probe to investigate the dark energy and the dark matter. Due to the new level of accuracy in observations, we must perform cosmological predictions in state-of-art simulations, to precisely quantify its variances, biases and potential degeneracies coming from baryonic physics. In this context, my thesis focuses on the construction of accurate weak lensing observables. The first part of my PhD work characterises the geometry of large-scale structure through weak lensing (Gouin et al. 2017). I relied on multipolar decomposition of weak lensing signal to quantify the azimuthal distribution of dark matter centred on galaxy clusters. The statistical properties of these moments are estimated from a large N-body simulation. The harmonic distortions computed in the vicinity of clusters appear to trace the filamentary structure. Larger number of filaments seem to be connected to high-mass clusters.The detection level of this statistical estimator is estimated. In the last part of my thesis, I mock the weak gravitational lensing signal in the light-cone of the Horizon-AGN simulation (Gouin et al. 2019). To do so, I propagate light-rays along the light-cone in the multiple-lens-plane approximation. The impact of baryons is significant in cosmic shear statistics for angular scales below a few arcmins. In addition, the galaxy-galaxy lensing signal is compared to current observational measurements (Leauthaud et al. 2017), and seems in good agreement.

Les propriétés optiques des nanomatériaux, tels que les cristaux photoniques ou les métamatériaux, ont reçu beaucoup d’attention dans les dernières années [1–9]. La dérivation numérique de ces propriétés se révèle pourtant très compliquée, en particulier dans le cas des structures métallo-diélectriques, qui comportent des résonances plasmoniques. C’est pourquoi des méthodes numériques avancées et des modèles semi-analytiques sont nécessaires. Dans cette thèse, nous montrerons que le formalisme de la matrice de diffraction peut satisfaire ces deux aspects. La méthode de la matrice de diffraction est un concept très général en physique. Dans le cas des structures périodiques, on peut dériver la matrice de diffraction à l’aide de la méthode modale de Fourier [10]. Pour la description exacte des géométries planes, nous avons développé la méthode des coordonnées adaptées [11], qui nous donne un nouveau système de coordonnées, dans lequel les interfaces des matériaux sont des surfaces de coordonnées constantes. En combinaison avec la méthode de la résolution spatiale adaptative, la méthode des coordonnées adaptées permet d’améliorer considérablement la convergence de la méthode modale de Fourier, de telle sorte qu’on peut calculer des structures métalliques compliquées très efficacement. Si on utilise la matrice de diffraction, il est non seulement possible de dériver les propriétés optiques en illumination de champ lointain, comme la transmission, la réflexion, l’absorption, et le champ proche, mais aussi de décrire l’émission d’un objet à l’intérieur d’une structure et d’obtenir les résonances optiques d’un sytème. Dans cette thèse, nous présenterons une méthode efficace pour la dérivation des résonances optiques tridimensionnelles, utilisant directement la matrice de diffraction [14]. Si on connaît les résonances d’un système isolé, il est aussi possible d’obtenir une approximation des résonances dans le cas d’un système combiné à l’aide de notre méthode du couplage des résonances [15, 16]. Cette méthode permet de décrire le régime de couplage des champs lointain et proche, y compris le couplage fort avec les résonances Fabry-Perot, pour des systèmes qui se composent d’un empilement de deux structures planes et périodiques. Pour cette raison, on peut étudier efficacement le couplage de ces systèmes. Cette thèse est écrite de manière à donner une idée d’ensemble du formalisme de la matrice de diffraction et de la méthode modale de Fourier. En outre, nous décrivons notre généralisation de ces méthodes et nous montrons la validité de nos approches pour différents exemples. / The optical properties of nanostructures such as photonic crystals and metamaterials have drawn a lot of attention in recent years [1–9]. The numerical derivation of these properties, however, turned out to be quite complicated, especially in the case of metallo-dielectric structures with plasmonic resonances. Hence, advanced numerical methods as well as semi-analytical models are required. In this work, we will show that the scattering matrix formalism can provide both. The scattering matrix approach is a very general concept in physics. In the case of periodic grating structures, the scattering matrix can be derived by the Fourier modal method [10]. For an accurate description of non-trivial planar geometries, we have extended the Fourier modal method by the concept of matched coordinates [11], in which we introduce a new coordinate system that contains the material interfaces as surfaces of constant coordinates. In combination with adaptive spatial resolution [12,13], we can achieve a tremendously improved convergence behavior which allows us to calculate complex metallic shapes efficiently. Using the scattering matrix, it is not only possible to obtain the optical properties for far field incidence, such as transmission, reflection, absorption, and near field distributions, but also to solve the emission from objects inside a structure and to calculate the optical resonances of a system. In this work, we provide an efficient method for the ab initio derivation of three-dimensional optical resonances from the scattering matrix [14]. Knowing the resonances in a single system, it is in addition possible to obtain approximated resonance positions for stacked systems using our method of the resonant mode coupling [15, 16]. The method allows describing both near field and far field regime for stacked two-layer systems, including the strong coupling to Fabry-Perot resonances. Thus, we can study the mutual coupling in such systems efficiently. The work will provide the reader with a basic understanding of the scattering matrix formalism and the Fourier modal method. Furthermore, we will describe in detail our extensions to these methods and show their validity for several examples.

La modélisation structurale consiste à approximer les structures géologiques du sous-sol en un modèle numérique afin d'en visualiser la géométrie et d'y effectuer des calculs d'estimation et de prédiction. L'approche implicite de la modélisation structurale utilise des données de terrain interprétées pour construire une fonction volumétrique sur le domaine d'étude qui représente la géologie. Cette fonction doit honorer les observations, interpoler entre ces dernières, et extrapoler dans les zones sous-échantillonnées tout en respectant les concepts géologiques. Les méthodes actuelles portent cette interpolation soit sur les données, soit sur un maillage. Ensuite, le problème de modélisation est posé selon la discrétisation choisie : par krigeage dual sur les points de donnée ou en définissant un critère de rugosité sur les éléments du maillage. Dans cette thèse, nous proposons une formulation continue de la modélisation structurale par méthodes implicites. Cette dernière consiste à minimiser une somme de fonctionnelles arbitraires. Les contraintes de donnée sont imposées avec des fonctionnelles discrètes, et l'interpolation est contrôlée par des fonctionnelles continues. Cette approche permet de (i) développer des liens entre les méthodes existantes, (ii) suggérer de nouvelles discrétisations d'un même problème de modélisation, et (iii) modifier le problème de modélisation pour mieux honorer certains cas géologiques sans dépendre de la discrétisation. Nous portons également une attention particulière à la gestion des discontinuités telles que les failles et les discordances. Les méthodes existantes nécessitent soit la création de zones volumétriques avec des géométries complexes, soit la génération d'un maillage volumétrique dont les éléments sont conformes aux surfaces de discontinuité. Nous montrons, en explorant des méthodes sans maillage locales et des concepts de réduction de maillage, qu'il est possible d'assurer l'interpolation des structures tout en réduisant les contraintes liées à la gestion des discontinuités. Deux discrétisations de notre problème de minimisation sont suggérées : l'une utilise les moindres carrés glissants avec des critères optiques pour la gestion des discontinuités, et l'autre utilise des fonctions issues de la méthode des éléments finis avec le concept de nœuds fantômes pour les discontinuités. Une étude de sensibilité et une comparaison des deux méthodes sont proposées en 2D, ainsi que quelques exemples en 3D. Les méthodes développées dans cette thèse ont un grand impact en termes d'efficacité numérique et de gestion de cas géologiques complexes. Par exemple, il est montré que notre problème de minimisation au sens large apporte plusieurs solutions pour la gestion de cas de plis sous-échantillonnés et de variations d'épaisseur dans les couches stratigraphiques. D'autres applications sont également présentées tels que la modélisation d'enveloppe de sel et la restauration mécanique. / Implicit structural modeling consists in approximating geological structures into a numerical model for visualization, estimations, and predictions. It uses numerical data interpreted from the field to construct a volumetric function on the domain of study that represents the geology. The function must fit the observations, interpolate in between, and extrapolate where data are missing while honoring the geological concepts. Current methods support this interpolation either with the data themselves or using a mesh. Then, the modeling problem is posed depending on these discretizations: performing a dual kriging between data points or defining a roughness criterion on the mesh elements. In this thesis, we propose a continuous formulation of implicit structural modeling as a minimization of a sum of generic functionals. The data constraints are enforced by discrete functionals, and the interpolation is controlled by continuous functionals. This approach enables to (i) develop links between the existing methods, (ii) suggest new discretizations of the same modeling problem, and (iii) modify the minimization problem to fit specific geological issues without any dependency on the discretization. Another focus of this thesis is the efficient handling of discontinuities, such as faults and unconformities. Existing methods require either to define volumetric zones with complex geometries, or to mesh volumes with conformal elements to the discontinuity surfaces. We show, by investigating local meshless functions and mesh reduction concepts, that it is possible to reduce the constraints related to the discontinuities while performing the interpolation. Two discretizations of the minimization problem are then suggested: one using the moving least squares functions with optic criteria to handle discontinuities, and the other using the finite element method functions with the concept of ghost nodes for the discontinuities. A sensitivity analysis and a comparison study of both methods are performed in 2D, with some examples in 3D. The developed methods in this thesis prove to have a great impact on computational efficiency and on handling complex geological settings. For instance, it is shown that the minimization problem provides the means to manage under-sampled fold structures and thickness variations in the layers. Other applications are also presented such as salt envelope surface modeling and mechanical restoration.

The main objective of this dissertation is to present an adaptation of some finite volume methods used in the resolution of problems arising in sedimentation processes of flocculated suspensions (or sedimentation with compression).<br />This adaptation is based on the utilization of multiresolution techniques, originally designed to reduce the computational cost incurred in solving using high resolution schemes in the numerical solution of hyperbolic systems of conservation laws.

This thesis studies the behavior of the Eulerian scheme, with "Wave Confinement" (WC), when propagating periodic waves. WC is a recently developed method that was derived from the scheme "vorticity confinement" used in fluid mechanics, and it efficiently solves the linear wave equation. This new method is applicable for numerous simulations such as radio wave propagation, target detection, cell phone and satellite communications. The WC scheme adds a nonlinear term to the discrete wave equation that adds stability with negative and positive diffusion, conserves integral quantities such as total amplitude and wave speed, and it allows wave propagation over long distances with minimal numerical diffusion, which contrasts to other numerical methods where wave propagation is affected by numerical dissipation. Previous studies have shown that WC propagates short pulses/surfaces as thin nonlinear solitary waves. In this thesis, a one-dimensional (1D) periodic wave is propagated by WC using the advection and wave equations. For the advection equation, the parameters and the initial condition (IC) used in WC are analyzed to establish for which conditions the method can be implemented. When the IC is a positive periodic wave, the converged solution consists of a series of hyperbolic secants where the number of cycles of the IC represents the number of hyperbolic secants. Waves with varying signs are analyzed by changing the wave confinement term. For this case, the converged solution is a series of positive and negative hyperbolic secants where each hyperbolic secant is represented by half cycle of the IC. For the wave equation, parameters and different IC's are studied to determine when WC is feasible. For positive periodic waves, the converged solution retains its sinusoidal shape and does not converge to a series of hyperbolic secants. The waves with varying signs, however, converge to a series of hyperbolic secants as seen for the advection equation. WC is stable for various periodic waves for both advection and wave equations, which shows WC is useful for numerically propagating periodic waveforms. Convergence depends on the wave number of the IC and on the parameters (convection speed, positive diffusion, negative diffusion) used in WC.

Since 1980, many convertible bonds have been issued by Swedish companies. Most of these issues have been aimed at the employees. The great number of these employee issues gave rise to a new tax law. This tax law made it necessary to obtain a value on a convertible bond certificate at issue. In the first part of the dissertation, the institutional setting for the issuing of convertible bonds in Sweden is discussed. The relevant tax laws and recommendations given by different organizations are described. Also other features related to the issues are described. Furthermore, an empirical study of convertible bonds issues to emplyees in listed companies is carried out. The main purpose of the study is to quantify the volume of convertible bond issues to employees which have defaulted. Issues with a nominal value of around 500 million Swedish Crowns have been involved in some form of default. In this study, several models are compared to investigate whether the choice of model for valuing convertible bonds is important. These models all fall within the framework of Contingent Claims Analysis. Contingent Claims Analysis is an option based technique for determining the value of a claim whose payoffs depend upon the development of one or several underlying variables. In the study, it is shown in great detail how to set up and use those models. It is shown that the choice of model is important for the value of a convertible bond in certain situations. Those situations are identified by an empirical study of Swedish convertible bonds and through sensitivity analysis. / <p>Diss. Stockholm : Handelshögskolan, 1993</p>