In this thesis, we focus on integration issue of manufacturing and sales functions from the perspective of aggregate production planning. The manufacturing function and sales function are performed by separate affiliated companies of the same business group, which operate as an integrated supplier-buyer system. In particular, this study provides theoretical and practical insight into the use of forecast volatility measure to better match supply with demand so as to reduce the costs of inventory and stock-outs in the manufacturer-buyer relationship under described master production-scheduling environment. Nature of forecast modifications provided by the buyer lays the foundation for the study. We modify the existing aggregate production planning model to accommodate a measure of historical forecast evolution. The overall objective of the thesis is to provide management with aforecast evolution-modeling framework to examine performance characteristics of the manufacturer-buyer interaction.

Nous étudions la mesure spectrale des transformations stationnaires, puis nous l’utilisons pour étudier le théorème ergodique et le théorème limite central. Nous étudions également les martingales avec une nouvelle preuve du théorème central limite, sans analyse de Fourier. Pour le théorème limite central pour marches aléatoires dans un environnement aléatoire sur la dimension 1, on donne deux méthodes pour l’obtenir: approximation pour une martingale et méthode des moments. La méthode des martingales fait résoudre l’équation de Dirichlet (I - P)h = 0, alors que celle des moments résoudre l’équation de Poisson (I - P)h = f. Enfin, nous pouvons utiliser la deuxième méthode pour prouver la relation d’Einstein pour des diffusions réversibles dans un environnement aléatoire dans une dimension. / We study the spectral measure for stationary transformations, and then apply to Ergodic theorem and Central limit theorem. We study also martingale process with a new proof of the central limit theorem without Fourier analysis. For the central limit theorem for random walks in random environment, we give two methods to obtain it: martingale approximation and moments. The method of martingales solves Dirichlet’s equation (I - P)h = 0, and the method of moments solves Poisson’s equation (I - P)h = f. Finally, we can use the second method to prove the Einstein relation for reversible diffusions in random environment in one dimension.

This thesis concerns applications of some probabilistic tools to phylogeny reconstruction and population genetics. Modelling the evolution of species by continuous-time random walks on the signed permutation groups, we study the asymptotic medians of a set of random permutations sampled from simple random walks at time 0.25cn, for c> 0. Running k independent random walks all starting at identity, we prove that the medians approximate the ancestor (identity permutation) up to time 0.25n, while there exists a constant c>1 after which the medians loose credibility as an estimator. We study the median of a set of random permutations on the symmetric group endowed with different metrics. In particular, for a special metric of dissimilarity, called breakpoint, where the space is not geodesic, we find a large group of medians of random permutations using the concept of partial geodesics (or geodesic patches). Also, we study the Fleming-Viot process in random environment (FVRE) via martingale and duality methods. We develop the duality method to the case of time-dependent and quenched martingale problems. Using a family of dual processes we prove the convergence of the Moran processes in random environments to FVRE in Skorokhod topology. We also study the long-time behaviour of FVRE and prove the existence of equilibrium for the joint annealed-environment process and prove an ergodic theorem for the latter.

Mémoire numérisé par la Direction des bibliothèques de l'Université de Montréal.

In this dissertation, we have studied diffusion models and their applications in risk theory and insurance. Let Xt be a d-dimensional diffusion process satisfying a system of Stochastic Differential Equations defined on an open set G Rd, and let Ut be a utility function of Xt with U0 = u0. Let T be the first time that Ut reaches a level u^*. We study the Laplace transform of the distribution of T, as well as the probability of ruin, psileft(u_{0}right)=Prleft{ T<inftyright} , and other important probabilities. A class of exponential martingales is constructed to analyze the asymptotic properties of all probabilities. In addition, we prove that the expected discounted penalty function, a generalization of the probability of ultimate ruin, satisfies an elliptic partial differential equation, subject to some initial boundary conditions. Two examples from areas of actuarial work to which martingales have been applied are given to illustrate our methods and results: 1. Insurer's insolvency. 2. Terrorism risk. In particular, we study insurer's insolvency for the Cram'{e}r-Lundberg model with investments whose price follows a geometric Brownian motion. We prove the conjecture proposed by Constantinescu and Thommann. / Mathematics

Arithmetic Asian options are difficult to price and hedge, since, at the present, no closed-form analytical solution exists to price them. This difficulty, moreover, has led to the development of various methods and models used to price these instruments. The purpose of this thesis is two-fold. First, we present an overview of the literature. Secondly, we develop a pseudo-analytical method proposed by Geman and Yor and present an accurate and relatively quick algorithm which can be used to price European-style arithmetic Asian options and their hedge parameters. / Master of Science

Dans cette thèse, on s'intéresse à la notion de temps fort de stationnarité et à celle, étroitement liée, de dual de stationnarité forte. Ces outils permettent d'étu- dier la convergence de processus ergodiques, en déterminant un instant aléatoire où l'équilibre est atteint. Les espaces d'état des processus considérés ici sont des graphes continus ou discrets. Dans la première partie, on considère le cas discret, et on dégage une condition nécessaire et suffisante à l'existence, pour n'importe quelle loi initiale, d'un temps fort de stationnarité fini. Pour cela, on construit explicitement un dual de station- narité forte, à valeurs dans l'ensemble des parties connexes du graphe, qui évolue à chaque étape en ajoutant ou en enlevant des points de sa frontière. Lorsque cette opération sépare l'ensemble dual en plusieurs parties, afin de ne pas le déconnecter, une de ces parties est choisie au hasard, avec une probabilité proportionnelle à son poids par la mesure invariante. On s'intéresse également au comportement général d'un processus dual, et on donne quelques exemples différents de celui construit précédemment. Dans la deuxième partie, on traite le cas continu, et le processus étudié est alors une diffusion. On caractérise notamment sa mesure invariante, et on explicite un générateur infinitésimal qui devrait être celui d'un processus dual. Néanmoins, ce cas s'avère plus compliqué que le cas discret. Le processus dual n'est donc construit que pour un mouvement brownien sur un graphe particulier, comme l'unique so- lution d'un problème de martingale. Des pistes sont présentées pour traiter des diffusions sur des graphes plus généraux, notamment en utilisant la convergence d'une suite de processus de saut tels que ceux présentés dans la première partie. / In this thesis, we are interested in the notion of strong stationary time, and in that, strongly connected, of strong stationary dual. These tools allow to study the convergence of ergodic processes, by determining a random time when the equilibrium is reached. The state space of the considered processes are discrete or continuous graphs. In the first part, we consider the discrete case, and we explicit a necessary and sufficient condition to the existence, for any initial distribution, of a finite strong stationary time. To do so, we construct explicitly a strong stationary dual, with values in the set of connected subsets of the graph, which evolves at each step by adding or removing some points at its border. Whenever this operation separates the dual set in several parts, in order not to disconnect it, one of these parts is chosen randomly, with a probability proportionnal to its weight relative to the invariant distribution. We also study the general behaviour of any dual process,2 and we give some other examples. In the second part, we deal with the continuous case, and the studied process is then a diffuion. We caracterize its invariant distribution, and we explicit an infinitesimal generator, which is expected to be that of a dual process. Nevertheless, this case turns out to be a little more involved that the discrete one. The dual process is thus constructed only for a brownian motion on a particular graph, as the unique solution of a martingale problem. Some leads are given to solve the case of diffusions on more general graphs, especially by using the convergence of a sequence of jump processes such as those presented in the first part.

Nous étudions la mesure spectrale des transformations stationnaires, puis nous l'utilisons pour étudier le théorème ergodique et le théorème limite central. Nous étudions également les martingales avec une nouvelle preuve du théorème central limite, sans analyse de Fourier. Pour le théorème limite central pour marches aléatoires dans un environnement aléatoire sur la dimension 1, on donne deux méthodes pour l'obtenir: approximation pour une martingale et méthode des moments. La méthode des martingales fait résoudre l'equation de Dirichlet (I −P )h = 0, alors que celle des moments résoudre l'equation de Poisson (I − P )h = f . Enfin, nous pouvons utiliser la deuxième méthode pour prouver la relation d'Einstein pour des diffusions réversibles dans un environnement aléatoire dans une dimension.

In this study, general geometric Levy market models are considered. Since these models are, in general, incomplete, that is, all contingent claims cannot be replicated by a self-financing portfolio consisting of investments in a risk-free bond and in the stock, it is suggested that the market should be enlarged by artificial assets based on the power-jump processes of the underlying Levy process. Then it is shown that the enlarged market is complete and the explicit hedging portfolios for claims whose payoff function depends on the prices of the stock and the artificial assets at maturity are derived. Furthermore, the portfolio optimization problem is considered in the enlarged market. The problem consists of choosing an optimal portfolio in such a way that the largest expected utility of the terminal wealth is obtained. It is shown that for particular choices of the equivalent martingale measure in the market, the optimal portfolio only consists of bonds and stocks. This corresponds to completing the market with additional assets in such a way that they are superfluous in the sense that the terminal expected utility is not improved by including these assets in the portfolio.

Cette thèse porte sur l'étude de la concentration autour de la moyenne de fonctions de variables aléatoires indépendantes à l'aide de techniques de martingales et d'inégalités de comparaison.Dans une première partie, nous prouvons des inégalités de comparaison pour des fonctions générales séparément convexes de variables aléatoires indépendantes non nécessairement bornées. Ces résultats sont établis à partir de nouvelles inégalités de comparaison dans des classes de fonctions convexes (contenant, en particulier, les fonctions exponentielles croissantes) pour des variables aléatoires réelles uniquement dominées stochastiquement.Dans la seconde partie, nous nous intéressons aux suprema de processus empiriques associés à des observations i.i.d. Le point clé de cette partie est un résultat d'échangeabilité des variables. Nous montrons d'abord des inégalités de type Fuk-Nagaev avec constantes explicites lorsque les fonctions de la classe ne sont pas bornées. Ensuite, nous prouvons de nouvelles inégalités de déviation avec une meilleure fonction de taux dans les bandes de grandes déviations dans le cas des classes de fonctions uniformément bornées. Nous donnons également des inégalités de comparaison de moments généralisés dans les cas uniformément borné et uniformément majoré. Enfin, les résultats de la première partie nous permettent d'obtenir une inégalité de concentration lorsque les fonctions de la classe ont une variance infinie. / This thesis deals with concentration properties around the mean of functions of independent random variables using martingale techniques and comparison inequalities.In the first part, we prove comparison inequalities for general separately convex functions of independent and non necessarily bounded random variables. These results are based on new comparison inequalities in convex classes of functions (including, in particular, the increasing exponential functions) for real-valued random variables which are only stochastically dominated.In the second part, we are interested in suprema of empirical processes associated to i.i.d. random variables. The key point of this part is a result of exchangeability of variables. We first give Fuk-Nagaev type inequalities with explicit constants when the functions of the considered class are unbounded. Next, we provide new deviation inequalities with an improved rate function in the large deviations bandwidth in the case of classes of uniformly bounded functions. We also provide generalized moment comparison inequalities in uniformly bounded and uniformly bounded from above cases. Finally, results from the first part allow us to prove a concentration inequality when the functions of the class have an infinite variance.