Avanços recentes nos conhecimentos sobre a infecção viral e AIDS tem levado pacientes soropositivos a uma melhor qualidade de vida. A determinação de quais populações celulares ou qual mecanismo imunológico seja mais relevante para instalação da epidemia conduz a novos patamares de possibilidades de novas drogas antiretrovirais e tratamento mais eficientes. O uso de modelagem matemática, para a epidemiologia, correlaciona indivíduos (neste caso células) e doença (o vírus) através de equações diferenciais, onde se quer observar as condições necessárias para a instalação ou não da doença. Neste trabalho, observou-se através das simulações, que o componente mais importante, depois do linfócito TCD4+, é a célula macrófago (por ser um reservatório de proliferação viral), que a infecção ocorre várias vezes ao longo do tempo (devido o processo de apresentação de antígenos) e que os linfócitos CTL são ineficientes em erradicar a infecção pelo vírus HIV-1, que pode ser um simples fenômeno de co-adaptação. / Recent advances in knowledge about the viral infection and AIDS seropositive patients has led to a better life quality. The determination of what people or cellular immune mechanism which is more relevant for the epidemic installation leads to new levels of possibilities to new antiretroviral drugs discovers and more efficient treatment. Mathematical modeling use on epidemiology, correlates individuals (this case cells) and illness (the virus) through differential equations, where want to observe the conditions necessary to the installation or not the disease. In this study, it was observed through simulations, that the most important component, after lymphocyte CD4 T cells, macrophages is the cell (as a reservoir of viral proliferation) that the infection occurs repeatedly over time (because of the antigen presenting process) and CTL lymphocytes are inefficient in eradicating the infection by HIV-1, which may be a simple phenomenon of co-adaptation.

Les neurosciences computationnelles consistent en l’étude du système nerveux par la modélisation et la simulation. Plus le modèle sera proche de la réalité et plus les ressources calculatoires exigées seront importantes. La question de la complexité et de la précision est un problème bien connu dans la simulation. Les travaux de recherche menés dans le cadre de cette thèse visent à améliorer la simulation de modèles mathématiques représentant le comportement physique et chimique de récepteurs synaptiques. Les modèles sont décrits par des équations différentielles ordinaires (EDO), et leur résolution passe par des méthodes numériques. Dans le but d’optimiser la simulation, j’ai implémenté différentes méthodes de résolution numérique des EDO. Afin de faciliter la sélection du meilleur algorithme de résolution numérique, une méthode nécessitant un minimum d’information a été proposée. Cette méthode permet de choisir l’algorithme qui optimise la simulation. La méthode a permis de démontrer que la dynamique d’un modèle de récepteur synaptique influence plus les performances des algorithmes de résolution que la structure cinétique du modèle lui-même. De plus, afin de caractériser des comportements pathogènes, une phase d’optimisation est réalisée. Cependant, certaines valeurs de paramètres rendent le modèle instable. Une étude de stabilité a permis de déterminer la stabilité du modèle pour des paramètres fournis par la littérature, mais également de remonter à des contraintes de stabilité sur les paramètres. Le respect de ces contraintes permet de garantir la stabilité des modèles étudiés, et donc de garantir le succès de la procédure permettant de rendre un modèle pathogène. / Computational Neuroscience consists in studying the nervous system through modeling and simulation. It is to characterize the laws of biology by using mathematical models integrating all known experimental data. From a practical point of view, the more realistic the model, the largest the required computational resources. The issue of complexity and accuracy is a well known problem in the modeling and identification of models. The research conducted in this thesis aims at improving the simulation of mathematical models representing the physical and chemical behavior of synaptic receptors. Models of synaptic receptors are described by ordinary differential equations (ODE), and are resolved with numerical procedures. In order to optimize the performance of the simulations, I have implemented various ODE numerical resolution methods. To facilitate the selection of the best solver, a method, requiring a minimum amount of information, has been proposed. This method allows choosing the best solver in order to optimize the simulation. The method demonstrates that the dynamic of a model has greater influence on the solver performances than the kinetic scheme of the model. In addition, to characterize pathogenic behavior, a parameter optimization is performed. However, some parameter values lead to unstable models. A stability study allowed for determining the stability of the models with parameters provided by the literature, but also to trace the stability constraints depending to these parameters. Compliance with these constraints ensures the stability of the models studied during the optimization phase, and therefore the success of the procedure to study pathogen models.

Several existing pricing models of financial derivatives as well as the effects of volatility risk are analyzed. A new option pricing model is proposed which assumes that stock price follows a diffusion process with square-root stochastic volatility. The volatility itself is mean-reverting and driven by both diffusion and compound Poisson process. These assumptions better reflect the randomness and the jumps that are readily apparent when the historical volatility data of any risky asset is graphed. The European option price is modeled by a homogeneous linear second-order partial differential equation with variable coefficients. The case of underlying assets that pay continuous dividends is considered and implemented in the model, which gives the capability of extending the results to American options. An American option price model is derived and given by a non-homogeneous linear second order partial integro-differential equation. Using Fourier and Laplace transforms an exact closed-form solution for the price formula for European call/put options is obtained.

本研究中討論了非線性微分方程式之解的正則性。在這之中發現了一些有趣的現象，得到了方程式解可以做任意次的微分，並且得到對該解任意次微分後其值趨近到無限大時之爆破速率、爆破常數及當其值遞減至零時的爆破速率、爆破常數。 / In this paper we work with the regularity of solutions for the non-linear ordinary differential equation u''-u^p=0 for some well-defined functions u^p. We have found some interesting phenomena, u belongs to C^q for any q in positive integer, blow-up constant, blow-up rate, null point and decay rate of u^(n) are obtained in this work, through that we get the characterization for these equations in this case.

Modern control theory is today an interdisciplinary area of research. Just as much as this can be problematic, it also provides a rich research environment where practice and theory meet. This Thesis is conducted in the borderline between computing science (numerical analysis) and applied control theory. The design and analysis of a modern control system is a complex problem that requires high qualitative software to accomplish. Ideally, such software should be based on robust methods and numerical stable algorithms that provide quantitative as well as qualitative information. The introduction of the Thesis is dedicated to the underlying control theory and to introduce the reader to the main subjects. Throughout the Thesis, the theory is illustrated with several examples, and similarities and differences between the terminology from mathematics, systems and control theory, and numerical linear algebra are highlighted. The main contributions of the Thesis are structured in two parts, dealing with two mainly unrelated subjects. Part I is devoted to the qualitative information which is provided by the stratification of orbits and bundles of matrices, matrix pencils and system pencils. Before the theory of stratification is established the reader is introduced to different canonical forms which reveal the system characteristics of the model under investigation. A stratification reveals which canonical structures of matrix (system) pencils are near each other in the sense of small perturbations of the data. Fundamental concepts in systems and control, like controllability and observability of linear continuous-time systems, are considered and it is shown how these system characteristics can be investigated using the stratification theory. New results are presented in the form of the cover relations (nearest neighbours) for controllability and observability pairs. Moreover, the permutation matrices which take a matrix pencil in the Kronecker canonical form to the corresponding system pencil in (generalized) Brunovsky canonical form are derived. Two novel algorithms for determining the permutation matrices are provided. Part II deals with numerical methods for solving periodic Riccati differential equations (PRDE:s). The PRDE:s under investigation arise when solving the linear quadratic regulator (LQR) problem for periodic linear time-varying (LTV) systems. These types of (periodic) LQR problems turn up for example in motion planning of underactuated mechanical systems, like a humanoid robot, the Furuta pendulum, and pendulums on carts. The constructions of the nonlinear controllers are based on linear versions found by stabilizing transverse dynamics of the systems along cycles. Three different methods explicitly designed for solving the PRDE are evaluated on both artificial systems and stabilizing problems originating from experimental control systems. The methods are the one-shot generator method and two recently proposed methods: the multi-shot method (two variants) and the SDP method. As these methods use different approaches to solve the PRDE, their numerical behavior and performance are dependent on the nature of the underlying control problem. Such method characteristics are investigated and summarized with respect to different user requirements (the need for accuracy and possible restrictions on the solution time). / Modern reglerteknik är idag i högsta grad ett interdisciplinärt forskningsområde. Lika mycket som detta kan vara problematiskt, resulterar det i en stimulerande forskningsmiljö där både praktik och teori knyts samman. Denna avhandling är utförd i gränsområdet mellan datavetenskap (numerisk analys) och tillämpad reglerteknik. Att designa och analysera ett modernt styrsystem är ett komplext problem som erfordrar högkvalitativ mjukvara. Det ideala är att mjukvaran består av robusta metoder och numeriskt stabila algoritmer som kan leverera både kvantitativ och kvalitativ information.Introduktionen till avhandlingen beskriver grundläggande styr- och reglerteori samt ger en introduktion till de huvudsakliga problemställningarna. Genom hela avhandlingen illustreras teori med exempel. Vidare belyses likheter och skillnader i terminologin som används inom matematik, styr- och reglerteori samt numerisk linjär algebra. Avhandlingen är uppdelade i två delar som behandlar två i huvudsak orelaterade problemklasser. Del I ägnas åt den kvalitativa informationen som ges av stratifiering av mångfalder (orbits och bundles) av matriser, matrisknippen och systemknippen. Innan teorin för stratifiering introduceras beskrivs olika kanoniska former, vilka var och en avslöjar olika systemegenskaper hos den undersökta modellen. En stratifiering ger information om bl.a. vilka kanoniska strukturer av matrisknippen (systemknippen) som är nära varandra med avseende på små störningar i datat. Fundamentala koncept i styr- och reglerteori behandlas, så som styrbarhet och observerbarhet av linjära tidskontinuerliga system, och hur dessa systemegenskaper kan undersökas med hjälp av stratifiering. Nya resultat presenteras i form av relationerna för täckande (närmsta grannar) styrbarhets- och observerbarhets-par. Dessutom härleds permutationsmatriserna som tar ett matrisknippe i Kroneckers kanoniska form till motsvarande systemknippe i (generaliserade) Brunovskys kanoniska form. Två algoritmer för att bestämma dessa permutationsmatriser presenteras. Del II avhandlar numeriska metoder för att lösa periodiska Riccati differentialekvationer (PRDE:er). De undersökta PRDE:erna uppkommer när ett linjärt kvadratiskt regulatorproblem för periodiska linjära tidsvariabla (LTV) system löses. Dessa typer av (periodiska) regulatorproblem dyker upp till exempel när man planerar rörelser för understyrda (underactuated) mekaniska system, så som en humanoid (mänsklig) robot, Furuta-pendeln och en vagn med en inverterad (stående) pendel. Konstruktionen av det icke-linjära styrsystemet är baserat på en linjär variant som bestäms via stabilisering av systemets transversella dynamik längs med cirkulära banor. Tre metoder explicit konstruerade för att lösa PRDE:er evalueras på både artificiella system och stabiliseringsproblem av experimentella styrsystem. Metoderna är sk. en- och flerskotts metoder (one-shot, multi-shot) och SDP-metoden. Då dessa metoder använder olika tillvägagångssätt för att lösa en PRDE, beror dess numeriska egenskaper och effektivitet på det underliggande styrproblemet. Sådana metodegenskaper undersöks och sammanfattas med avseende på olika användares behov, t.ex. önskad noggrannhet och tänkbar begränsning i hur lång tid det får ta att hitta en lösning.

Residual-distribution methods offer several potential benefits over classical methods, such as a means of applying upwinding in a multi-dimensional manner and a multi-dimensional positivity property. While it is apparent that residual-distribution methods also offer higher accuracy than finite-volume methods on similar meshes, few studies have directly compared the performance of the two approaches in a systematic and quantitative manner. In this study, comparisons between residual distribution and finite volume are made for steady-state smooth and discontinuous flows of gas dynamics, governed by hyperbolic conservation laws, to illustrate the strengths and deficiencies of the residual-distribution method. Deficiencies which reduce the accuracy are analyzed and a new nonlinear scheme is proposed that closely reproduces or surpasses the accuracy of the best linear residual-distribution scheme. The accuracy is further improved by extending the scheme to fourth order using established finite-element techniques. Finally, the compact stencil, arithmetic workload, and data parallelism of the fourth-order residual-distribution scheme are exploited to accelerate parallel computations on an architecture consisting of both CPU cores and a graphics processing unit. Numerical experiments are used to assess the gains to efficiency and possible monetary savings that may be provided by accelerated architectures.

Gas sensors are today used in many different application areas, and one growing future market is battery operated sensors. As many gas sensor components are heated, one major limit of the operation time is caused by the power dissipated as heat. AppliedSensor is a company that develops and produces gas sensor components, modules and solutions, among which battery operated gas sensors are one targeted market. The aim of the diploma work has been to simulate the heat transfer on a hydrogen gas sensor component and its closest surroundings consisting of a carrier mounted on a printed circuit board. The component is heated in order to improve the performance of the gas sensing element. Power dissipation occurs by all three modes of heat transfer; conduction from the component through bond wires and carrier to the printed circuit board as well as convection and radiation from all the surfaces. It is of interest to AppliedSensor to understand which factors influence the heat transfer. This knowledge will be used to improve different aspects of the gas sensor, such as the power consumption. Modeling and simulation have been performed in FEMLAB, a tool for solving partial differential equations by the finite element method. The sensor system has been defined by the geometry and the material properties of the objects. The system of partial differential equations, consisting of the heat equation describing conduction and boundary conditions specifying convection and radiation, was solved and the solution was validated against experimental data. The convection increases with the increase of hydrogen concentration. A great effort was made to finding a model for the convection. Two different approaches were taken, the first based on known theory from the area and the second on experimental data. When the first method was compared to experiments, it turned out that the theory was insufficient to describe this small system involving hydrogen, which was an unexpected but interesting result. The second method matched the experiments well. For the continuation of the project at the company, a better model of the convection would be a great improvement.

Residual-distribution methods offer several potential benefits over classical methods, such as a means of applying upwinding in a multi-dimensional manner and a multi-dimensional positivity property. While it is apparent that residual-distribution methods also offer higher accuracy than finite-volume methods on similar meshes, few studies have directly compared the performance of the two approaches in a systematic and quantitative manner. In this study, comparisons between residual distribution and finite volume are made for steady-state smooth and discontinuous flows of gas dynamics, governed by hyperbolic conservation laws, to illustrate the strengths and deficiencies of the residual-distribution method. Deficiencies which reduce the accuracy are analyzed and a new nonlinear scheme is proposed that closely reproduces or surpasses the accuracy of the best linear residual-distribution scheme. The accuracy is further improved by extending the scheme to fourth order using established finite-element techniques. Finally, the compact stencil, arithmetic workload, and data parallelism of the fourth-order residual-distribution scheme are exploited to accelerate parallel computations on an architecture consisting of both CPU cores and a graphics processing unit. Numerical experiments are used to assess the gains to efficiency and possible monetary savings that may be provided by accelerated architectures.

We apply some recent numerical solutions to diffusion bridges written in Iacus (2008). One is an approximate scheme from Bladt and S{\o}rensen (2007), another one, from  Beskos et al (2006), is an algorithm which is exact: no numerical error at given grid points!

In this paper we investigate (d + 1)-point D-optimal designs for d-th degree polynomial regression with weight function w(x) > 0 on the interval [a, b]. Suppose that w'(x)/w(x) is a rational function and the information of whether the optimal support contains the boundary points a and b is available. Then the problem of constructing (d + 1)-point D-optimal designs can be transformed into a differential equation problem leading us to a certain matrix with k auxiliary unknown constants. We characterize the weight functions corresponding to the cases when k= 0 and k= 1. Then, we can solve (d + 1)-point D-optimal designs directly from differential equation (k = 0) or via eigenvalue problems (k = 1). The numerical results show us an interesting relationship between optimal designs and ordered eigenvalues.