Cette étude exploratoire a pour but d’identifier les facteurs pouvant influencer l’implantation des adjoints au médecin dans le système de santé québécois, selon les perceptions de médecins omnipraticiens et de médecins spécialistes. La collecte de données pour cette étude qualitative s’est effectuée à l’aide d’entrevues semi-structurées effectuées auprès de 13 omnipraticiens et spécialistes provenant d’hôpitaux de Montréal et de la clinique médicale des Forces canadiennes de St-Jean (Québec). L’étude a démontré que des obstacles perçus, tels que le corporatisme et le manque d’information sur la profession, pourraient interférer avec l’intégration des adjoints au médecin au Québec. Cependant, les participants s’entendent pour dire que ces obstacles ne seraient pas insurmontables et ont, par la même occasion, identifié de nombreux éléments pouvant faciliter cette intégration. Les adjoints au médecin ont des compétences uniques et travaillent déjà dans d’autres provinces canadiennes qui ont un réseau de santé similaire au nôtre. Cette étude permet donc d’approfondir les connaissances à l’égard de cette profession, en plein essor au pays, dans l’éventualité d’une intégration de ce groupe professionnel au Québec. / The intent of this exploratory study is to identify the factors that could influence the implementation of Physician Assistants in Quebec's medical system based on the opinions of General Physicians and Medical Specialists. The data were collected from semi-structured interviews conducted with 13 generalists and specialists from hospitals in the Montréal area and from the military clinic located in the Canadian Forces Base St-Jean (Québec). The study showed that perceived obstacles such as corporatism and the lack of information about the profession could interfere with physician assistant integration. However, participants agreed that these obstacles could be overcome. They have, at the same time, identified many very positive factors in favour of this integration. Physician assistants have unique skills and are already fully integrated in provincial health care systems in other Canadian provinces. This study offers a better understanding and greater awareness of this expanding profession in our country and the progress towards its possible integration in Québec.

Face aux forts besoins de réduction de la consommation énergétique et de l’impact environnemental,le bâtiment d’aujourd’hui vise la performance en s’appuyant sur des sourcesd’énergie de plus en plus diversifiées (énergies renouvelables), une enveloppe mieux conçue(isolation) et des systèmes de gestion plus avancés. Plus la conception vise la basse consommation,plus les interactions entre ses composants sont complexes et peu intuitives. Seule unerégulation plus intégrée permettrait de prendre en compte cette complexité et d’optimiser lefonctionnement pour atteindre la basse consommation sans sacrifier le confort.Les techniques de commande prédictive, fondées sur l’utilisation de modèles dynamiqueset de techniques d’optimisation, promettent une réduction des consommations et de l’inconfort.Elles permettent en effet d’anticiper l’évolution des sources et des besoins intermittentstout en tirant parti de l’inertie thermique du bâtiment, de ses systèmes et autres élémentsde stockage. Cependant, dans le cas du bâtiment, l’obtention d’un modèle dynamique suffisammentprécis présente des difficultés du fait d’incertitudes importantes sur les paramètresdu modèle et les sollicitations du système. Les avancées récentes dans le domaine de l’instrumentationdomotique constituent une opportunité prometteuse pour la réduction de cesincertitudes, mais la conception d’un tel système pour une telle application n’est pas triviale.De fait, il devient nécessaire de pouvoir considérer les problématiques de monitoring énergétique,d’instrumentation, de commande prédictive et de modélisation de façon conjointe.Cette thèse vise à identifier les liens entre commande prédictive et instrumentation dansle bâtiment, en proposant puis exploitant une méthode générique de modélisation du bâtiment,de simulation thermique et de résolution de problèmes d’optimisation. Cette méthodologiemet en oeuvre une modélisation thermique multizone du bâtiment, et des algorithmesd’optimisation reposant sur un modèle adjoint et les outils du contrôle optimal. Elle a étéconcrétisée dans un outil de calcul permettant de mettre en place une stratégie de commandeprédictive comportant des phases de commande optimale, d’estimation d’état et decalibration.En premier lieu, nous étudions la formulation et la résolution d’un problème de commandeoptimale. Nous abordons les différences entre un tel contrôle et une stratégie de régulationclassique, entre autres sur la prise en compte d’indices de performance et de contraintes. Nousprésentons ensuite une méthode d’estimation d’état basée sur l’identification de gains thermiquesinternes inconnus. Cette méthode d’estimation est couplée au calcul de commandeoptimale pour former une stratégie de commande prédictive.Les valeurs des paramètres d’un modèle de bâtiment sont souvent très incertaines. Lacalibration paramétrique du modèle est incontournable pour réduire les erreurs de prédictionet garantir la performance d’une commande optimale. Nous appliquons alors notreméthodologie à une technique de calibration basée sur des mesures de températures in situ.Nous ouvrons ensuite sur des méthodes permettant d’orienter le choix des capteurs à utiliser(nombre, positionnement) et des paramètres à calibrer en exploitant les gradients calculéspar la méthode adjointe.La stratégie de commande prédictive a été mise en oeuvre sur un bâtiment expérimentalprès de Chambéry. Dans le cadre de cette étude, l’intégralité du bâtiment a été modélisé,et les différentes étapes de notre commande prédictive ont été ensuite déployées de mainière séquentielle. Cette mise en oeuvre permet d’étudier les enjeux et les difficultés liées àl’implémentation d’une commande prédictive sur un bâtiment réel.Cette thèse est issue d’une collaboration entre le CEA Leti, l’IFSTTAR de Nantes et leG2ELab, et s’inscrit dans le cadre du projet ANR PRECCISION. / More efficient energy management of buildings through the use of Model Predictive Control(MPC) techniques is a key issue to reduce the environmental impact of buildings. Buildingenergy performance is currently improved by using renewable energy sources, a betterdesign of the building envelope (insulation) and the use of advanced management systems.The more the design aims for high performance, the more interactions and coupling effectsbetween the building, its environment and the conditions of use are important and unintuitive.Only a more integrated regulation would take in account this complexity, and couldhelp to optimize the consumption without compromising the comfort.Model Predictive Control techniques, based on the use of dynamic models and optimizationmethods, promise a reduction of consumption and discomfort. They can generate energysavings by anticipating the evolution of renewable sources and intermittent needs, while takingadvantage of the building thermal inertia and other storage items. However, in the caseof buildings, obtaining a good dynamic model is tough, due to important uncertainties onmodel parameters and system solicitations.Recent advances in the field of wireless sensor networks are fostering the deployment ofsensors in buildings, and offer a promising opportunity to reduce these errors. Nevertheless,designing a sensor network dedicated to MPC is not obvious, and energy monitoring,instrumentation, modeling and predictive control matters must be considered jointly.This thesis aims at establishing the links between MPC and instrumentation needs inbuildings. We propose a generic method for building modeling, thermal simulation andoptimization. This methodology involves a multi-zone thermal model of the building, andefficient optimization algorithms using an adjoint model and tools from the optimal controltheory. It was implemented in a specific toolbox to develop a predictive control strategywith optimal control phases, state estimation phases and model calibration.At first, we study the formulation and resolution of an optimal control problem. We discussthe differences between such a control and a conventional regulation strategy, throughperformance indicators. Then, we present a state estimation method based on the identificationof unknown internal gains. This estimation method is subsequently coupled with theoptimal control method to form a predictive control strategy.As the parameters values of a building model are often very uncertain, parametric modelcalibration is essential to reduce prediction errors and to ensure the MPC performance. Consequently,we apply our methodology to a calibration technique based on in situ temperaturemeasurements. We also discuss how our approach can lead to selection techniques in orderto choose calibrated parameters and sensors for MPC purposes.Eventually, the predictive control strategy was implemented on an experimental building,at CEA INES, near Chambéry. The entire building was modeled, and the different steps ofthe control strategy were applied sequentially through an online supervisor. This experimentgave us a useful feedback on our methodology on a real case.This thesis is the result of a collaboration between CEA Leti, IFSTTAR Nantes andG2ELab, and is part of the ANR PRECCISION project.

Submitted by Aelson Maciera (aelsoncm@terra.com.br) on 2017-06-27T20:42:50Z No. of bitstreams: 1 DissLFL.pdf: 818058 bytes, checksum: a45e8f4dbdc692c6f31fde1d45f6574d (MD5) / Approved for entry into archive by Ronildo Prado (ronisp@ufscar.br) on 2017-07-03T17:56:46Z (GMT) No. of bitstreams: 1 DissLFL.pdf: 818058 bytes, checksum: a45e8f4dbdc692c6f31fde1d45f6574d (MD5) / Approved for entry into archive by Ronildo Prado (ronisp@ufscar.br) on 2017-07-03T17:56:52Z (GMT) No. of bitstreams: 1 DissLFL.pdf: 818058 bytes, checksum: a45e8f4dbdc692c6f31fde1d45f6574d (MD5) / Made available in DSpace on 2017-07-03T18:01:59Z (GMT). No. of bitstreams: 1 DissLFL.pdf: 818058 bytes, checksum: a45e8f4dbdc692c6f31fde1d45f6574d (MD5) Previous issue date: 2016-12-20 / Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) / In this thesis, we present a priori estimates for solutions of a mean-field game (MFG) defined over a bounded domain Ω ⊂ ℝd. We propose an application of these results to a model of capital and wealth accumulation. In Chapter 1, an introduction to mean-field games is presented. We also put forward some of the motivation from Economics and discuss previous developments in the theory of differential games. These comments aim at indicating the connection between mean-field games theory, its applications and the realm of Mathematical Analysis. In Chapter 2, we present an optimal control problem. Here, the agents are supposed to be undistinguishable, rational and intelligent. Undistinguishable means that every agent is governed by the same stochastic differential equation. Rational means that all efforts of the agent is to maximize a payoff functional. Intelligent means that they are able to solve an optimal control problem. Once we describe this (stochastic) optimal control problem, we produce a heuristic derivation of the mean-field games system, which is summarized in a Verification Theorem; this gives rise to the Hamilton-Jacobi equation (HJ). After that, we obtain the Fokker-Plank equation (FP). Finally, we present a representation formula for the solutions to the (HJ) equation, together with some regularity results. In Chapter 3, a specific optimal control problem is described and the associated MFG is presented. This MFG is prescribed in a bounded domain Ω ⊂ ℝd, which introduces substantialadditional challenges from the mathematical view point. This is due to estimates for the solutionsat the boundary in Lp. The rest of the chapter puts forward two well known tips of estimates: theso-called Hopf-Lax formula and the First Order Estimate. In Chapter 4, the wealth and capital accumulation mean-field game model is presented. The relevance of studying MFG in a bounded domain then becomes clear. In light of the results obtained in Chapter 3, we close Chapter 4 with the Hopf-Lax formula, and the First Order estimates. Three appendices close this thesis. They gather elementary material on Stochastic Calculus and Functional Analysis. / Nesta dissertação são apresentadas algumas estimativas a priori para soluções de sistemas mean-field games (MFG), definidos em domínios limitados Ω ⊂ ℝd. Tais estimativas são aplicadas em um modelo mean-field específico, que descreve o acúmulo de riqueza e capital. No Capítulo 1, é apresentada uma breve introdução histórica sobre os mean-field games. Nesta introdução, exploramos sua relação com a teoria dos jogos, cujos alicerces foram construídos por economistas e matemáticos ao longo do século XX. O objetivo do capítulo é transmitir. No Capítulo 2, apresentamos um problema de controle ótimo em que cada agente é suposto ser indistinguível, racional e inteligente. Indistinguível no sentido de que cada um é governado pela mesma equação diferencial estocástica. Racional no sentido de que todos os esforços do agente são no sentido de maximizar um funcional de recompensa e, inteligente no sentido de que são capazes de resolver um problema de controle ótimo. Descreve-se este problema de controle ótimo, e apresenta-se a derivação heurística dos mean-field games; obtém-se através de um Teorema de Verificação, a equação de Hamilton-Jacobi (HJ) associada, e em seguida, obtémse a equação de Fokker-Planck. De posse destas equações, apresentamos alguns resultados preliminares, como uma fórmula de representação para soluções da equação de HJ e alguns resultados de regularidade. No Capítulo 3, descreve-se um problema específico de controle ótimo e apresenta-se a respectiva derivação heurística culminando na descrição de um MFG com condições não periódicas na fronteira; esta abordagem é original na literatura de MFG. O restante do capítulo é dedicado à exposição de dois tipos bem conhecidos de estimativas: a fórmula de Hopf-Lax e estimativa de Primeira Ordem. Uma observação relevante, é a de que o trabalho em obter-se estimativas a priori é aumentado substancialmente neste caso, devido ao fato de lidarmos com estimativas para os termos de fronteira com normas em Lp. ao leitor, as origens da Teoria Econômica contemporânea, que surgem à partir da utilização da Matemática na formulação e resolução de problemas econômicos. Tal abordagem é motivada principalmente pelo rigor e clareza da Matemática em tais circunstâncias. No Capítulo 4, apresenta-se o modelo de jogo do tipo mean-field de acúmulo de capital e riqueza, o que deixa claro a relevância do estudo dos MFG em um domínio limitado. À luz dos resultados obtidos no Capítulo 3, encerramos o Capítulo 4 com as estimativas do tipo Hopf-Lax e de Primeira Ordem. Três apêndices encerram o texto desta dissertação de mestrado; estes reúnem material elementar sobre Cálculo Estocástico e Análise Funcional.

Depuis quelques années un métier émergent de l'éducatif se développe dans l'Enseignement Catholique , celui d'A.P.S. ou Adjoint en Pastorale Scolaire. A la croisée entre deux réalités institutionnelles, celle de l'éducation et celle de l'église catholique, ce métier se structure et se développe en tension entre une sur prescription à caractère discrétionnaire et le réel de l'activité des acteurs éducatifs dans leur contexte de travail. En nous appuyant sur la théorie de l'analyse de l'activité, notre intervention-recherche en collaboration avec cinq personnes en formation continue, a permis de mettre en évidence l'existence de plusieurs genres professionnels au sein d'un même métier qui peut être rattaché aux métiers "avec et pour autrui". Le dispositif expérimental a favorisé la constitution d'un collectif de travail, l'émergence de règles de métier, de compétences particulières , ainsi que le développement de l'activité des professionnels. Notre étude a joué le rôle d'analyseur des transformations des métiers qui consistent à agir "avec et pour autrui". Dans une perspectives d'évolution du système scolaire français, la réforme des collèges et la prise en compte des enjeux de société concernant la mixité sociale et culturelle, l'analyse de l'activité de ce métier émergent ouvre une réflexion sur les métiers de l'éducatif" avec et pour autrui" et la formation. / In recent years , a job as emerged from the education grows in Catholic teaching, the A.P.S. or school pastoral assistants. At the jonction between two institutional realities, that of Education and that of the Church, the profession has structured and developped stretching between a discretionary over prescription and the real activity of educational actors in heir work context. Based on the theory of activity analysis, our intervention research which has been conducted in collaboration with five people in in-house training as working, has highlighted the existence of several professionnal within' the same job which can be attached to trades that consist of acting with and for others. The experimental device has encouraged the formation of a collective work, the emergence of business rules, news skills, and the" development of the activity of professionals. Our study has also payed a role of the analyzer of business transformations "are to act for and whith another one".In an evlutionary perpective with the french school system reform in colleges an the inclusion of social issues concerning the social and cultural diversity, the analysis of the activity of emerging job has opened reflections on the job of education " for and with another one".

Originally developed in the 1970s for the optimal control of systems governed by partial differential equations, the adjoint method has found several successful applications, e.g. in meteorology with large-scale 3D or 4D atmospheric data assimilation schemes, for carbon cycle data assimilation in biogeochemistry and climate research, or in oceanographic modelling with efficient adjoint codes of ocean general circulation models.<p><p>Despite the variety of applications in these research fields, adjoint methods have only very recently drawn attention from the ocean acoustics community. In ocean acoustic tomography and geoacoustic inversion, where the inverse problem is to recover unknown acoustic properties of the water column and the seabed from acoustic transmission data, the solution approaches are typically based on travel time inversion or standard matched-field processing in combination with metaheuristics for global optimization. <p><p>In order to complement the adjoint schemes already in use in meteorology and oceanography with an ocean acoustic component, this thesis is concerned with the development of the adjoint of a full-field acoustic propagation model for shallow water environments. <p><p>In view of the increasing importance of global ocean observing systems such as the European Seas Observatory Network, the Arctic Ocean Observing System and Maritime Rapid Environmental Assessment (MREA) systems for defence and security applications, the adjoint of an ocean acoustic propagation model can become an integral part of a coupled oceanographic and acoustic data assimilation scheme in the future. <p><p>Given the acoustic pressure field measured on a vertical hydrophone array and a modelled replica field that is calculated for a specific parametrization of the environment, the developed adjoint model backpropagates the mismatch (residual) between the measured and predicted field from the receiver array towards the source.<p><p>The backpropagated error field is then converted into an estimate of the exact gradient of the objective function with respect to any of the relevant physical parameters of the environment including the sound speed structure in the water column and densities, compressional/shear sound speeds, and attenuations of the sediment layers and the sub-bottom halfspace. The resulting environmental gradients can be used in combination with gradient descent methods such as conjugate gradient, or Newton-type optimization methods tolocate the error surface minimum via a series of iterations. This is particularly attractive for monitoring slowly varying environments, where the gradient information can be used to track the environmental parameters continuously over time and space.<p><p>In shallow water environments, where an accurate treatment of the acoustic interaction with the bottom is of outmost importance for a correct prediction of the sound field, and field data are often recorded on non-fully populated arrays, there is an inherent need for observation over a broad range of frequencies. For this purpose, the adjoint-based approach is generalized for a joint optimization across multiple frequencies and special attention is devoted to regularization methods that incorporate additional information about the desired solution in order to stabilize the optimization process.<p><p>Starting with an analytical formulation of the multiple-frequency adjoint approach for parabolic-type approximations, the adjoint method is progressively tailored in the course of the thesis towards a realistic wide-angle parabolic equation propagation model and the treatment of fully nonlocal impedance boundary conditions. A semi-automatic adjoint generation via modular graph approach enables the direct inversion of both the geoacoustic parameters embedded in the discrete nonlocal boundary condition and the acoustic properties of the water column. Several case studies based on environmental data obtained in Mediterranean shallow waters are used in the thesis to assess the capabilities of adjoint-based acoustic inversion for different experimental configurations, particularly taking into account sparse array geometries and partial depth coverage of the water column. The numerical implementation of the approach is found to be robust, provided that the initial guesses are not too far from the desired solution, and accurate, and converges in a small number of iterations. During the multi-frequency optimization process, the evolution of the control parameters displays a parameter hierarchy which clearly relates to the relative sensitivity of the acoustic pressure field to the physical parameters. <p><p>The actual validation of the adjoint-generated environmental gradients for acoustic monitoring of a shallow water environment is based on acoustic and oceanographic data from the Yellow Shark '94 and the MREA '07 sea trials, conducted in the Tyrrhenian Sea, south of the island of Elba.<p> <p>Starting from an initial guess of the environmental control parameters, either obtained through acoustic inversion with global search or supported by archival in-situ data, the adjoint method provides an efficient means to adjust local changes with a couple of iterations and monitor the environmental properties over a series of inversions. <p><p>In this thesis the adjoint-based approach is used, e.g. to fine-tune up to eight bottom geoacoustic parameters of a shallow-water environment and to track the time-varying sound speed profile in the water column. <p><p>In the same way the approach can be extended to track the spatial water column and bottom structure using a mobile network of sparse arrays.<p><p>Work is currently being focused on the inclusion of the adjoint approach into hybrid optimization schemes or ensemble predictions, as an essential building block in a combined ocean acoustic data assimilation framework and the subsequent validation of the acoustic monitoring capabilities with long-term experimental data in shallow water environments. / Doctorat en Sciences de l'ingénieur / info:eu-repo/semantics/nonPublished

L’adaptation de maillage est un outil puissant pour l’obtention de simulations aérodynamiques précises à coût limité. Dans le cas particulier des simulations visant au calcul de fonctions aérodynamiques (efforts, moments, rendements...), plusieurs méthodes dites de raffinement ciblé (ou, en anglais, « goal-oriented ») basées sur le vecteur adjoint de la fonction d’intérêt ont été proposées. L’objectif de la thèse est l’extension d’une méthode de ce type basée sur la dérivée totale dJ/dX de la grandeur aérodynamique d’intérêt, J, par rapport aux coordonnées du maillage volumique, X. Les trois méthodes usuelles de calcul de gradient discret – la méthode de différentiation directe, la méthode adjointe-"paramètres" et la méthode adjointe-"maillage" évaluant dJ/dX – ont tout d’abord été étudiées et codées dans le logiciel elsA de l’ONERA pour des maillages non-structurés, pour des écoulements compressibles de fluide parfait et des écoulements laminaires. La seconde étape du travail a consisté à créer un senseur local θ basé sur dJ/dX qui identifie les zones du maillage volumique où la position des nœuds a une forte incidence sur l’évaluation de la fonction J. Ce senseur sert d’indicateur pour l’adaptation de différents maillages, pour différents régimes d’écoulement (subsonique, transsonique, supersonique), pour des configurations d’aérodynamique interne (aube et tuyère) et externe (profil d’aile). La méthode proposée est comparée à une méthode de raffinement ciblée très populaire (Venditti et Darmofal, 2001) et à une méthode de raffinement basée sur les caractéristiques de l’écoulement (ou, en anglais, « feature-based ») ; elle conduit à des résultats très satisfaisants. / Mesh adaptation is a powerful tool to obtain accurate aerodynamic simulations with limited cost. In the specific case of computation of aerodynamic functions (forces, moments, efficiency ...), goal-oriented methods based on the adjoint vector have been proposed. The aim of the thesis is the extension of a method of this type based on the total derivative dJ/dX of the aerodynamic output of interest, J, with respect to the volume mesh coordinates, X. The three common methods for calculating discrete gradient – the direct differentiation method, the parameter-adjoint method and mesh-adjoint method evaluating dJ/dX – have been studied first and coded in the elsA ONERA software for unstructured grids, for compressible inviscid and laminar flows. The second part of this work was has been to define a local sensor θ based on dJ/dX in order to identify zones where the volume mesh nodes position has a strong impact on the evaluation of the function J. This sensor is the selected indicator for different mesh adaptations for different flow regimes (subsonic, transonic, supersonic) for internal (blade and nozzle) and external (wing profile) aerodynamic configurations. The proposed method is compared to a well-known goal-oriented method (Darmofal and Venditti, 2001) and to a feature-based method ; it leads to very consistent results. very consistent results.

Vacuum handling is a widespread technology in automated production systems for gripping of workpieces. Unfortunately, this solution accounts for a considerable share of industrial energy consumption. This consumption is mainly due to the losses in the involved steps like air compression, distribution, vacuum generation and gripper suction. However, the energy efficiency of vacuum handling systems is still relatively unexplored. The consortium partners are working together in the BiVaS project, which is funded by the Federal Ministry for Economic Affairs and Energy (BMWi). In this project, the drawback of high energy consumption is approached by the development of an energyefficient ejector, a biomimetic suction gripper as well as on system integration and the development of energy-efficient operating strategies and their energetic balancing in order to reduce the consumption of compressed air by 20 %. This reduction will be quantified experimentally in a pilot plant benchmark, where a realistic handling process is developed and examined. This paper shows the state of the art focusing on energy consumption of vacuum handling technology and the behaviour of the involved components during different handling operations. Furthermore, first biomimetic concepts and an estimation of the increase of energy efficiency will be presented for the running project.

<p>The thesis intends to explain in full detail the developed techniques and approaches for the modeling, design, and sensitivity analysis of nano-plasmoic structures. However, some examples are included for audiences of general microwave background. Although the thesis is mainly focused on simulation-based techniques, analytical and convex optimization approaches are also demonstrated. The thesis is organized into two parts. Part 1 includes Chapters 2-4, which cover the simulation-based modeling and sensitivity analysis approaches and their applications. Part 2 includes Chapters 5 and 6, which cover the analytical optimization approaches.</p> / <p>We propose novel techniques for modeling, adjoint sensitivity analysis, and optimization of photonic and nano-plasmonic devices. The scope of our work is generalized to cover microwave, terahertz and optical regimes. It contains original approaches developed for different categories of materials including dispersive and plasmonic materials. Artificial materials (metamaterials) are also investigated and modeled. The modeling technique exploits the time-domain transmission line modeling (TD-TLM) technique. Generalized adjoint variable method (AVM) techniques are developed for sensitivity analysis of the modeled devices. Although TLM-based, they can be generalized to other time-domain modeling techniques like finite difference time-domain method (FDTD) and time-domain finite element method (FEM).</p> <p>We propose to extend the application of TLM-based AVM to photonic devices. We develop memory efficient approaches that overcome the limitation of excessive memory requirement in TLM-based AVM. A memory reduction of 90% can be achieved without loss of accuracy and at a more efficient calculation procedure. The developed technique is applied to slot waveguide Bragg gratings and a challenging dielectric resonator antenna problem.</p> <p>We also introduce a novel sensitivity analysis approach for materials with dispersive constitutive parameters. To our knowledge, this is the first wide-band AVM approach that takes into consideration the dependence of material properties on the frequency. The approach can be utilized for design optimization of innovative nano-plasmonic structures. The design of engineered metamaterial is systematic and efficient. Beside working with engineered new designs, dispersive AVM can be utilized in bio-imaging applications. The sensitivity of the objective function with respect to dispersive material properties enables the exploitation of parameter and gradient based optimization for imaging in the terahertz and optical regimes. Material resonance interaction can be easily investigated by the provided sensitivity information.</p> <p>In addition to the developed techniques for simulation-based optimization, several analytical optimization algorithms are proposed to foster the parameter extraction and design optimization in terahertz and optical regimes. In terahertz time-domain spectroscopy, we have developed an efficient parameter based approach that utilizes the pre-known information about the material. The algorithm allows for the estimation of the optical properties of sample materials of unknown thicknesses. The approach has been developed based on physical analytical dispersive models. It has been applied with the Debye, Lorentz, Cole-Cole, and Drude model.</p> <p>Furthermore, we propose various algorithms for design optimization of coupled resonators. The proposed algorithms are utilized to transform a highly non-linear optimization problem into a linear one. They exploit an approximate transfer function of the coupled resonators that avoids negligible multiple reflections among them. The algorithms are successful for the optimization of very large-scale coupled microcavities (150 coupled ring resonators).</p> / Doctor of Philosophy (PhD)

Nous analysons dans cette thèse différents aspects associés à la modélisation des écoulements à surface libre en eaux peu profondes (Shallow Water). Nous étudions tout d’abord le système d’équations de Saint-Venant à deux dimensions et leur résolution par la méthode numérique des volumes finis, en portant une attention particulière sur les aspects hyperboliques et conservatifs. Ces schémas permettent de traiter les équilibres stationnaires, les interfaces sec/mouillé et aussi de modéliser des écoulements subcritique, transcritique et supercritique. Nous présentons ensuite la théorie de la méthode d’assimilation variationnelle de données adaptée à ce type d’écoulement. Son application au travers des études de sensibilité est longuement discutée dans le cadre de l'hydraulique à surface libre. Après cette partie à caractère théorique, la partie tests commence par une qualification de l’ensemble des méthodes numériques qui sont implémentées dans le code DassFlow, développé à l’Université de Toulouse, principalement à l’IMT mais aussi à l’IMFT. Ce code résout les équations Shallow Water par une méthode de volumes finis et est validé par comparaison avec les solutions analytiques pour des cas tests classiques. Ces mêmes résultats sont comparés avec un autre code d’hydraulique à surface libre aux éléments finis en deux dimensions, Telemac 2D. Une particularité notable du code DassFlow est de permettre l’assimilation variationnelle de données grâce au code adjoint permettant le calcul du gradient de la fonction coût. Ce code adjoint a été obtenu en utilisant l'outil de différentiation automatique Tapenade (Inria). Nous testons ensuite sur un cas réel, hydrauliquement complexe, différentes qualités de Modèles Numériques de Terrain (MNT) et de bathymétrie du lit d’une rivière. Ces informations proviennent soit d’une base de données classique type IGN, soit d’informations LIDAR à très haute résolution. La comparaison des influences respectives de la bathymétrie, du maillage et du type de code utilisé, sur la dynamique d’inondation est menée très finement. Enfin nous réalisons des études cartographiques de sensibilité aux paramètres du modèle sur DassFlow. Ces cartes montrent l’influence respective des différents paramètres ou de la localisation des points de mesure virtuels. Cette localisation optimale de ces points est nécessaire pour une future assimilation de données efficiente. / We analyze in this thesis various aspects associated with the modeling of free surface flows in shallow water approximation. We first study the system of Saint-Venant equations in two dimensions and its resolution with the numerical finite volumes method, focusing in particular on aspects hyperbolic and conservative. These schemes can process stationary equilibria, wetdry interfaces and model subcritical, transcritical and supercritical flows. After, we present the variational data assimilation method theory fitted to this kind of flow. Its application through sensitivity studies is fully discussed in the context of free surface water. After this theoretical part, we test the qualification of numerical methods implemented in the code Dassflow, developed at the University of Toulouse, mainly at l'IMT, but also at IMFT. This code solves the Shallow Water equations by finite volume method and is validated by comparison with analytical solutions for standard test cases. These results are compared with another hydraulic free surface flow code using finite elements in two dimensions: Telemac2D. A significant feature of the Dassflow code is to allow variational data assimilation using the adjoint method for calculating the cost function gradient. The adjoint code was obtained using the automatic differentiation tool Tapenade (INRIA). Then, the test is carried on a real hydraulically complex case using different qualities of Digital Elevation Models (DEM) and bathymetry of the river bed. This information are provided by either a conventional database types IGN or a very high resolution LIDAR information. The comparison of the respective influences of bathymetry, mesh size, kind of code used on the dynamics of flooding is very finely explored. Finally we perform sensitivity mapping studies on parameters of the Dassflow model. These maps show the respective influence of different parameters and of the location of virtual measurement points. This optimal location of these points is necessary for an efficient data assimilation in the future.

此篇論文，主要是討論遞迴式神經網路。在文中，我們將架構一個單層的神經網路結構。並利用三種不同的學習法則來套用此架構。我們也做了圓軌跡和圖形８的模擬，以及討論了此架構的收斂性。 / Our task in this paper is to discuss the Recurrent Neural Network. We construct a singal layer neural network and apply three different learning rules to simulate circular trajectory and figure eight. Also, we present the proof of convergence.