Philosophiae Doctor - PhD / The global impact of the converging dual epidemics of tuberculosis (TB) and human immunodeficiency virus (HIV) is one of the major public health challenges of our time, because in many countries, human immunodeficiency virus (HIV) and mycobacterium tuberculosis (TB) are among the leading causes of morbidity and mortality. It is found that infection with HIV increases the risk of reactivating latent TB infection, and HIV-infected individuals who acquire new TB infections have high rates of disease progression. Research has shown that these two diseases are enormous public health burden, and unfortunately, not much has been done in terms of modeling the dynamics of HIV-TB co-infection at a population level. In this thesis, we study these models and design and analyze robust numerical methods to solve them. To proceed in this direction, first we study the sub-models and then the full model. The first sub-model describes the transmission dynamics of HIV that accounts for behavior change. The impact of HIV educational campaigns is also studied. Further, we explore the effects of behavior change and different responses of individuals to educational campaigns in a situation where individuals may not react immediately to these campaigns. This is done by considering a distributed time delay in the HIV sub-model. This leads to Hopf bifurcations around the endemic equilibria of the model. These bifurcations correspond to the existence of periodic solutions that oscillate around the equilibria at given thresholds. Further, we show how the delay can result in more HIV infections causing more increase in the HIV prevalence. Part of this study is then extended to study a co-infection model of HIV-TB. A thorough bifurcation analysis is carried out for this model. Robust numerical methods are then designed and analyzed for these models. Comparative numerical results are also provided for each model. / South Africa

Dans cette thèse, nous étudions plusieurs problèmes de mathématiques financières liés à la valorisation des produits dérivés. Par différentes approches asymptotiques, nous développons des méthodes pour calculer des approximations précises du prix de certains types d’options dans des cas où il n’existe pas de formule explicite.Dans le premier chapitre, nous nous intéressons à la valorisation des options dont le payoff dépend de la trajectoire du sous-jacent par méthodes de Monte-Carlo, lorsque le sous-jacent est modélisé par un processus affine à volatilité stochastique. Nous prouvons un principe de grandes déviations trajectoriel en temps long, que nous utilisons pour calculer, en utilisant le lemme de Varadhan, un changement de mesure asymptotiquement optimal, permettant de réduire significativement la variance de l’estimateur de Monte-Carlo des prix d’options.Le second chapitre considère la valorisation par méthodes de Monte-Carlo des options dépendant de plusieurs sous-jacents, telles que les options sur panier, dans le modèle à volatilité stochastique de Wishart, qui généralise le modèle Heston. En suivant la même approche que dans le précédent chapitre, nous prouvons que le processus vérifie un principe de grandes déviations en temps long, que nous utilisons pour réduire significativement la variance de l’estimateur de Monte-Carlo des prix d’options, à travers un changement de mesure asymptotiquement optimal. En parallèle, nous utilisons le principe de grandes déviations pour caractériser le comportement en temps long de la volatilité implicite Black-Scholes des options sur panier.Dans le troisième chapitre, nous étudions la valorisation des options sur variance réalisée, lorsque la volatilité spot est modélisée par un processus de diffusion à volatilité constante. Nous utilisons de récents résultats asymptotiques sur les densités des diffusions hypo-elliptiques pour calculer une expansion de la densité de la variance réalisée, que nous intégrons pour obtenir l’expansion du prix des options, puis de leur volatilité implicite Black-Scholes.Le dernier chapitre est consacré à la valorisation des dérivés de taux d’intérêt dans le modèle Lévy de marché Libor qui généralise le modèle de marché Libor classique (log-normal) par l’ajout de sauts. En écrivant le premier comme une perturbation du second et en utilisant la représentation de Feynman-Kac, nous calculons explicitement l’expansion asymptotique du prix des dérivés de taux, en particulier, des caplets et des swaptions. / In this thesis, we study several mathematical finance problems, related to the pricing of derivatives. Using different asymptotic approaches, we develop methods to calculate accurate approximations of the prices of certain types of options in cases where no explicit formulas are available.In the first chapter, we are interested in the pricing of path-dependent options, with Monte-Carlo methods, when the underlying is modelled as an affine stochastic volatility model. We prove a long-time trajectorial large deviations principle. We then combine it with Varadhan’s Lemma to calculate an asymptotically optimal measure change, that allows to reduce significantly the variance of the Monte-Carlo estimator of option prices.The second chapter considers the pricing with Monte-Carlo methods of options that depend on several underlying assets, such as basket options, in the Wishart stochastic volatility model, that generalizes the Heston model. Following the approach of the first chapter, we prove that the process verifies a long-time large deviations principle, that we use to reduce significantly the variance of the Monte-Carlo estimator of option prices, through an asymptotically optimal measure change. In parallel, we use the large deviations property to characterize the long-time behaviour of the Black-Scholes implied volatility of basket options.In the third chapter, we study the pricing of options on realized variance, when the spot volatility is modelled as a diffusion process with constant volatility. We use recent asymptotic results on densities of hypo-elliptic diffusions to calculate an expansion of the density of realized variance, that we integrate to obtain an expansion of option prices and their Black-Scholes implied volatility.The last chapter is dedicated to the pricing of interest rate derivatives in the Levy Libor market model, that generaliszes the classical (log-normal) Libor market model by introducing jumps. Writing the first model as a perturbation of the second and using the Feynman-Kac representation, we calculate explicit expansions of the prices of interest rate derivatives and, in particular, caplets and swaptions

We propose a new method to perform functional principal component analysis (FPCA) for discretely observed functional data by solving successive optimization problems. The new framework can be applied to both regularly and irregularly observed data, and to both dense and sparse data. Our method does not require estimates of the individual sample functions or the covariance functions. Hence, it can be used to analyze functional data with multidimensional arguments (e.g. random surfaces). Furthermore, it can be applied to many processes and models with complicated or nonsmooth covariance functions. In our method, smoothness of eigenfunctions is controlled by directly imposing roughness penalties on eigenfunctions, which makes it more efficient and flexible to tune the smoothness. Efficient algorithms for solving the successive optimization problems are proposed. We provide the existence and characterization of the solutions to the successive optimization problems. The consistency of our method is also proved. Through simulations, we demonstrate that our method performs well in the cases with smooth samples curves, with discontinuous sample curves and nonsmooth covariance and with sample functions having two dimensional arguments (random surfaces), repectively. We apply our method to classification problems of retinal pigment epithelial cells in eyes of mice and to longitudinal CD4 counts data. In the second part of this dissertation, we propose a sparse Fisher’s discriminant analysis method with thresholded linear constraints. Various regularized linear discriminant analysis (LDA) methods have been proposed to address the problems of the LDA in high-dimensional settings. Asymptotic optimality has been established for some of these methods when there are only two classes. A difficulty in the asymptotic study for the multiclass classification is that for the two-class classification, the classification boundary is a hyperplane and an explicit formula for the classification error exists, however, in the case of multiclass, the boundary is usually complicated and no explicit formula for the error generally exists. Another difficulty in proving the asymptotic consistency and optimality for sparse Fisher’s discriminant analysis is that the covariance matrix is involved in the constraints of the optimization problems for high order components. It is not easy to estimate a general high-dimensional covariance matrix. Thus, we propose a sparse Fisher’s discriminant analysis method which avoids the estimation of the covariance matrix, provide asymptotic consistency results and the corresponding convergence rates for all components. To prove the asymptotic optimality, we provide an asymptotic upper bound for a general linear classification rule in the case of muticlass which is applied to our method to obtain the asymptotic optimality and the corresponding convergence rate. In the special case of two classes, our method achieves the same as or better convergence rates compared to the existing method. The proposed method is applied to multivariate functional data with wavelet transformations.

Le présent travail est consacré à l’approximation numérique des solutions du modèle aux moments M1 pour le transfert radiatif. Il s’agit, ici, de développer des solveurs numériques performants et précis capables de prédire avec précision et robustesse des écoulements où le transfert radiatif joue un rôle essentiel. Dans ce sens, plusieurs méthodes numériques ont été envisagées pour la dérivation des schémas numériques de type solveur de Godunov. Une attention particulière a été portée sur les solveurs préservant les ondes de contact stationnaires. En particulier, un schéma de relaxation et un solveur HLLC sont présentés dans ce travail. Pour chacun de ces solveurs, la robustesse de la méthode a été établie (positivité de l’énergie radiative et limitation du flux radiatif). La validation et l’intérêt des méthodes abordées sont exhibés à travers de nombreuses expériences numériques mono et multidimensionelles. / The present work is dedicated to the numerical approximation of the M1 moments model solutions for radiative transfer. The objective is to develop efficient and accurate numerical solvers, able to provide with precise and robust computations of flows where radiative transfer effects are important. With this aim, several numerical methods have been considered in order to derive numerical schemes based on Godunov type solvers. A particular attention has been paid to solvers preserving the stationary contact waves. Namely, a relaxation scheme and a HLLC solver are presented in this thesis. The robustness of each of these solvers has been established (radiative energy positivity and radiative flux limitation). Several numerical experiments in one and two space dimensions validate the developed methods and outline their interest.

Este trabalho objetiva apresentar um estudo detalhado e sistemático de algumas condições de regularidade para inferências baseadas em máxima verossimilhança no modelo de regressão elíptico multivariado com parametrização geral proposto em Lemonte e Patriota (2011). O modelo em estudo tem vários modelos importantes como casos particulares, entre eles temos os modelos lineares e não lineares homocedásticos e heterocedásticos, modelos mistos, modelos heterocedásticos com erros nas variáveis e na equação, modelos multiníveis, entre outros. As condições de regularidade estudadas estão associadas à identificabilidade do modelo, à existência, à unicidade, à consistência e à normalidade assintótica dos estimadores de máxima verossimilhança (EMV) e à distribuição assintótica das estatísticas de testes. Para isso, são enunciadas as condições suficientes e formalizados os teoremas que garantem a existência, unicidade, consistência e normalidade assintótica dos EMV e a distribuição assintótica das estatísticas de teste usuais. Além disso, os resultados de cada teorema são comentados e as demonstrações são apresentadas com detalhes. Inicialmente, considerou-se o modelo sob a suposição de normalidade dos erros, para, na sequência, ser possível generalizar os resultados para o caso elíptico. A fim de exemplificar os resultados obtidos, foram verificadas, analiticamente, a validade de algumas condições e os resultados de alguns teoremas em casos particulares do modelo geral. Ademais, foi desenvolvido um estudo de simulação em que uma das condições é violada adotando o modelo heterocedástico com erros nas variáveis e na equação. Por meio de simulações de Monte Carlo foram avaliados os impactos sobre a consistência e normalidade assintótica dos EMV. / This work aims to present a detailed and systematic study of some regularity conditions for inferences based on maximum likelihood in the multivariate elliptic regression model with general parameterization proposed in Lemonte and Patriota (2011). The model under study has several important models as particular cases, among them we have the linear and non-linear homocedastic and heterocedastic models, mixed models, heterocedastic models with errors in the variables and in the equation, multilevel models, among others. The regularity conditions studied are associated with the identifiability of the model, existence, uniqueness, consistency and asymptotic normality of the maximum likelihood estimators (MLE) and the asymptotic distribution of some test statistics. Sufficient conditions are stated to guarantee the existence, unicity, consistency and asymptotic normality of the MLE and the asymptotic distribution of the usual test statistics. In addition, the results of each theorem are commented and the proof are presented in detail. Initially, the model was considered under the assumption of normality of the errors, and then the results were generalized for the elliptical case. In order to exemplify the attained results, some particular cases of the general model are analyzed analytically, the validity of some conditions and the results of some theorems are verified. In addition, a simulation study is developed with one of the conditions violated under the heterocedastic model with errors in the variables and in the equation. By means of Monte Carlo simulations, the impacts of this violation on the consistency and the asymptotic normality of the MLE are evaluated.

There is a growing interest in the dynamics of the co-infection of these two diseases. In this thesis, firstly we focus on studying the effect of a distributed delay representing the incubation period for the malaria parasite in the mosquito vector to possibly reduce the initial transmission and prevalence of malaria. This model can be regarded as a generalization of SEI models (with a class for the latently infected mosquitoes) and SI models with a discrete delay for the incubation period in mosquitoes. We study the possibility of occurrence of backward bifurcation. We then extend these ideas to study a full model of HIV and malaria co-infection. To get further inside into the dynamics of the model, we use the geometric singular perturbation theory to couple the fast and slow models from the full model. Finally, since the governing models are very complex, they cannot be solved analytically and hence we develop and analyze a special class of numerical methods to solve them.

In this thesis, we study these models and design and analyze robust numerical methods to solve them. To proceed in this direction, first we study the sub-models and then the full model. The first sub-model describes the transmission dynamics of HIV that accounts for behavior change. The impact of HIV educational campaigns is also studied. Further, we explore the effects of behavior change and different responses of individuals to educational campaigns in a situation where individuals may not react immediately to these campaigns. This is done by considering a distributed time delay in the HIV sub-model. This leads to Hopf bifurcations around the endemic equilibria of the model. These bifurcations correspond to the existence of periodic solutions that oscillate around the equilibria at given thresholds. Further, we show how the delay can result in more HIV infections causing more increase in the HIV prevalence. Part of this study is then extended to study a co-infection model of HIV-TB. A thorough bifurcation analysis is carried out for this model. Robust numerical methods are then designed and analyzed for these models. Comparative numerical results are also provided for each model.

In the development of digital communication and information theory, every channel decoding rule has resulted in a revolution at the time when it was invented. In the area of information theory, early channel coding theorems were established mainly by maximum likelihood decoding, while the arrival of typical sequence decoding signaled the era of multi-user information theory, in which achievability proof became simple and intuitive. Practical channel code design, on the other hand, was based on minimum distance decoding at the early stage. The invention of belief propagation decoding with soft input and soft output, leading to the birth of turbo codes and low-density-parity check (LDPC) codes which are indispensable coding techniques in current communication systems, changed the whole research area so dramatically that people started to use the term "modern coding theory'' to refer to the research based on this decoding rule. In this thesis, we propose a new decoding rule, dubbed jar decoding, which would be expected to bring some new thoughts to both the code performance analysis and the code design. Given any channel with input alphabet X and output alphabet Y, jar decoding rule can be simply expressed as follows: upon receiving the channel output y^n ∈ Y^n, the decoder first forms a set (called a jar) of sequences x^n ∈ X^n considered to be close to y^n and pick any codeword (if any) inside this jar as the decoding output. The way how the decoder forms the jar is defined independently with the actual channel code and even the channel statistics in certain cases. Under this jar decoding, various coding theorems are proved in this thesis. First of all, focusing on the word error probability, jar decoding is shown to be near optimal by the achievabilities proved via jar decoding and the converses proved via a proof technique, dubbed the outer mirror image of jar, which is also quite related to jar decoding. Then a Taylor-type expansion of optimal channel coding rate with finite block length is discovered by combining those achievability and converse theorems, and it is demonstrated that jar decoding is optimal up to the second order in this Taylor-type expansion. Flexibility of jar decoding is then illustrated by proving LDPC coding theorems via jar decoding, where the bit error probability is concerned. And finally, we consider a coding scenario, called interactive encoding and decoding, and show that jar decoding can be also used to prove coding theorems and guide the code design in the scenario of two-way communication.

<p>The global impact of the converging dual epidemics of tuberculosis (TB) and human immunodeficiency virus (HIV) is one of the major public health challenges of our time, because in many countries, human immunodeficiency virus (HIV) and mycobacterium tuberculosis (TB) are among the leading causes of morbidity and mortality. It is found that infection with HIV increases the risk of reactivating latent TB infection, and HIV-infected individuals who acquire new TB infections have high rates of disease progression. Research has shown that these two diseases are enormous public health burden, and unfortunately, not much has been done in terms of modeling the dynamics of HIV-TB co-infection at a population level. In this thesis, we study these models and design and analyze robust numerical methods to solve them. To proceed in this direction, first we study the sub-models and then the full model. The first sub-model describes the transmission dynamics of HIV that accounts for behavior change. The impact of HIV educational campaigns is also studied. Further, we explore the effects of behavior change and different responses of individuals to educational campaigns in a situation where individuals may not react immediately to these campaigns. This is done by considering a distributed time delay in the HIV sub-model. This leads to Hopf bifurcations around the endemic equilibria of the model. These bifurcations correspond to the existence of periodic solutions that oscillate around the equilibria at given thresholds. Further, we show how the delay can result in more HIV infections causing more increase in the HIV prevalence. Part of this study is then extended to study a co-infection model of HIV-TB. A thorough bifurcation analysis is carried out for this model. Robust numerical methods are then designed and analyzed for these models.&nbsp / Comparative numerical results are also provided for each model.</p>

The dissertation investigates asymptotic theory of decentralized sequential hypothesis testing problems as well as asymptotic behaviors of the Sequential Minimum Energy Design (SMED). The main results are summarized as follows. 1.We develop the first-order asymptotic optimality theory for decentralized sequential multi-hypothesis testing under a Bayes framework. Asymptotically optimal tests are obtained from the class of "two-stage" procedures and the optimal local quantizers are shown to be the "maximin" quantizers that are characterized as a randomization of at most M-1 Unambiguous Likelihood Quantizers (ULQ) when testing M >= 2 hypotheses. 2. We generalize the classical Kullback-Leibler inequality to investigate the quantization effects on the second-order and other general-order moments of log-likelihood ratios. It is shown that a quantization may increase these quantities, but such an increase is bounded by a universal constant that depends on the order of the moment. This result provides a simpler sufficient condition for asymptotic theory of decentralized sequential detection. 3. We propose a class of multi-stage tests for decentralized sequential multi-hypothesis testing problems, and show that with suitably chosen thresholds at different stages, it can hold the second-order asymptotic optimality properties when the hypotheses testing problem is "asymmetric." 4. We characterize the asymptotic behaviors of SMED algorithm, particularly the denseness and distributions of the design points. In addition, we propose a simplified version of SMED that is computationally more efficient.