En réponse à la demande croissante de performance par une grande variété d’applications (exemples : modélisation financière, simulation sub-atomique, bio-informatique, etc.), les systèmes informatiques se complexifient et augmentent en taille (nombre de composants de calcul, mémoire et capacité de stockage). L’accroissement de la complexité de ces systèmes se traduit par une évolution de leur architecture vers une hétérogénéité des technologies de calcul et des modèles de programmation. La gestion harmonieuse de cette hétérogénéité, l’optimisation des ressources et la minimisation de la consommation constituent des défis techniques majeurs dans la conception des futurs systèmes informatiques.Cette thèse s’adresse à un domaine de cette complexité en se focalisant sur les sous-systèmes à mémoire partagée où l’ensemble des processeurs partagent un espace d’adressage commun. Les travaux porteront essentiellement sur l’implémentation d’un protocole de cohérence de cache et de consistance mémoire, sur une architecture extensible et sur la méthodologie de validation de cette implémentation.Dans notre approche, nous avons retenu les processeurs 64-bits d’ARM et des co-processeurs génériques (GPU, DSP, etc.) comme composants de calcul, les protocoles de mémoire partagée AMBA/ACE et AMBA/ACE-Lite ainsi que l’architecture associée « CoreLink CCN » comme solution de départ. La généralisation et la paramètrisation de cette architecture ainsi que sa validation dans l’environnement de simulation Gem5 constituent l’épine dorsale de cette thèse.Les résultats obtenus à la fin de la thèse, tendent à démontrer l’atteinte des objectifs fixés / In response to the growing demand for performance by a wide variety of applications (eg, financial modeling, sub-atomic simulation, bioinformatics, etc.), computer systems become more complex and increase in size (number of computing components, memory and storage capacity). The increased complexity of these systems results in a change in their architecture towards a heterogeneous computing technologies and programming models. The harmonious management of this heterogeneity, resource optimization and minimization of consumption are major technical challenges in the design of future computer systems.This thesis addresses a field of this complexity by focusing on shared memory subsystems where all processors share a common address space. Work will focus on the implementation of a cache coherence and memory consistency on an extensible architecture and methodology for validation of this implementation.In our approach, we selected processors 64-bit ARM and generic co-processor (GPU, DSP, etc.) as components of computing, shared memory protocols AMBA / ACE and AMBA / ACE-Lite and associated architecture "CoreLink CCN" as a starting solution. Generalization and parameterization of this architecture and its validation in the simulation environment GEM5 are the backbone of this thesis.The results at the end of the thesis, tend to demonstrate the achievement of objectives

Globally, the justice system has set up courts to respond to complaints of a criminal and civil nature. Courts also respond to complaints which require swift relief by way of shortened procedures, in the form of motion proceedings. In all these complaints, courts have to respond in a manner that leaves litigants with a feeling of satisfaction that justice has been done. To the end of ensuring that there is legal certainty, justice systems in all jurisdictions have established a hierarchy of courts, with lower courts being bound by the decisions of higher courts in their jurisdiction. There has been no problem in the application of this principle called stare decisis, or judicial precedent, in disputes of law. However, in disputes of constitutional interpretation, courts have demonstrated a marked shift from observing the rule of judicial precedent. The disregard for this rule manifests itself particularly in the adjudication of cases surrounded by controversy. It is argued herein that constitutional interpretation is no different from legal interpretation, in that the rule of judicial precedent which characterises court decisions in legal disputes, should characterise court decisions in constitutional interpretation disputes. The Constitutional Court of South Africa itself, though it is the highest arbiter in constitutional matters, is bound by its own previous decisions, unless its previous decisions have become manifestly wrong. Three constitutional rights are analysed. The right to life in its three manifestations, namely, the right to life of the unborn child, the right to life of the convicted criminal not to be hanged, and the right of the terminally ill to continue living by receiving medical care at state expense. The other two rights are the right to privacy, and the right to culture. The right to privacy is the right that has been claimed in political controversies. In isolated instances, specifically mentioned herein, the Constitutional Assembly and the drafters of the Constitution have also contributed to the resultant inconsistency in constitutional interpretation. This is especially so with regard to the right to practise one‘s culture. / Constitutional, International & Indigenous Law / LL.D.

Globally, the justice system has set up courts to respond to complaints of a criminal and civil nature. Courts also respond to complaints which require swift relief by way of shortened procedures, in the form of motion proceedings. In all these complaints, courts have to respond in a manner that leaves litigants with a feeling of satisfaction that justice has been done. To the end of ensuring that there is legal certainty, justice systems in all jurisdictions have established a hierarchy of courts, with lower courts being bound by the decisions of higher courts in their jurisdiction. There has been no problem in the application of this principle called stare decisis, or judicial precedent, in disputes of law. However, in disputes of constitutional interpretation, courts have demonstrated a marked shift from observing the rule of judicial precedent. The disregard for this rule manifests itself particularly in the adjudication of cases surrounded by controversy. It is argued herein that constitutional interpretation is no different from legal interpretation, in that the rule of judicial precedent which characterises court decisions in legal disputes, should characterise court decisions in constitutional interpretation disputes. The Constitutional Court of South Africa itself, though it is the highest arbiter in constitutional matters, is bound by its own previous decisions, unless its previous decisions have become manifestly wrong. Three constitutional rights are analysed. The right to life in its three manifestations, namely, the right to life of the unborn child, the right to life of the convicted criminal not to be hanged, and the right of the terminally ill to continue living by receiving medical care at state expense. The other two rights are the right to privacy, and the right to culture. The right to privacy is the right that has been claimed in political controversies. In isolated instances, specifically mentioned herein, the Constitutional Assembly and the drafters of the Constitution have also contributed to the resultant inconsistency in constitutional interpretation. This is especially so with regard to the right to practise one‘s culture. / Constitutional, International and Indigenous Law / LL.D.

Les modèles linéaires généralisés sont une généralisation des modèles de régression linéaire, et sont très utilisés dans le domaine du vivant. Le modèle de régression logistique, l'un des modèles de cette classe, très souvent utilisé dans les études biomédicales demeure le modèle de régression le plus approprié quand il s'agit de modéliser une variable discrète de nature binaire. Dans cette thèse, nous nous intéressons au problème de l'inférence statistique dans le modèle de régression logistique, en présence d'individus immunes dans la population d'étude.Dans un premier temps, nous considérons le problème de l'estimation dans le modèle de régression logistique en présence d'individus immunes, qui entre dans le cadre des modèles de régression à excès de zéros (ou zéro-inflatés). Un individu est dit immune s'il n'est pas exposé à l'événement d'intérêt. Le statut d'immunité est inconnu sauf si l'événement d'intérêt a été observé. Nous développons une méthode d'estimation par maximum de vraisemblance en proposant une modélisation conjointe de l'immunité et des risques d'infection. Nous établissons d'abord l'identifiabilité du modèle proposé. Puis, nous montrons l'existence de l'estimateur du maximum de vraisemblance des paramètres de ce modèle. Nous montrons ensuite,la consistance de cet estimateur, et nous établissons sa normalité asymptotique. Enfin, nous étudions, au moyen de simulations, leur comportement sur des échantillons de taille finie.Dans un deuxième temps, nous nous intéressons à la construction de bandes de confiance simultanées pour la probabilité d'infection, dans le modèle de régression logistique avec fraction immune. Nous proposons trois méthodes de constructions de bandes de confiance pour la fonction de régression. La première méthode (méthodede Scheffé) utilise la propriété de normalité asymptotique de l'estimateur du maximum de vraisemblance, et une approximation par une loi du khi deux pour approcher le quantile nécessaire à la construction des bandes. La deuxième méthode utilise également la propriété de normalité asymptotique de l'estimateur du maximum de vraisemblance et est basée sur une égalité classique de (Landau & Sheep 1970). La troisième méthode (méthode bootstrap) repose sur des simulations, pour estimer le quantile approprié de la loi du supremum d'un processus gaussien. Enfin, nous évaluons, au moyen de simulations, leurs propriétés sur des échantillons de taille finie.Enfin, nous appliquons les résultats de modélisation à des données réelles surla dengue. Il s'agit d'une maladie vectorielle tropicale à transmission strictement inter-humaine. Les résultats montrent que les probabilités d'infection estimées à partir de notre approche de modélisation sont plus élevées que celles obtenues à partir d'un modèle de régression logistique standard qui ne tient pas compte d'une possible immunité. En particulier, les estimations fournies par notre approche suggèrent que le sous-poids constitue un facteur de risque majeur de l'infection par la dengue, indépendamment de l'âge. / Generalized linear models are a generalization of linear regression models, and are widely used in the field of life. The logistic regression model, one of this class of models, widely used in biomedical studies remains the most appropriate regression model when it comes to model discrete variable, binary in nature. In this thesis, we investigate the problem of statistical inference in the logistic regression model, in the presence of immune individuals in the study population.At first, we consider the problem of estimation in the logistic regression model in the presence of immune individuals that enters in the case of zero-inflated regression models. A subject is said to be immune if he cannot experience the outcome of interest. The immune status is unknown unless the event of interest has been observed. We develop a maximum like lihood estimation procedure for this problem, based on the joint modeling of the binary response of interest and the cure status. We investigate the identifiability of the resulting model. Then, we establish the existence, consistency and asymptotic normality of the proposed estimator, and we conduct a simulation study to investigate its finite-sample behavior. In a second time, we focus on the construction of simultaneous confidence bands for the probability of infection in the logistic regression model with immune fraction.We propose three methods of construction of confidence bands for the regression function. The first method (Scheffe's method) uses the asymptotic normality of the maximum like lihood estimator, and an approximation by the chi-squared distribution to approximate the necessary quantile for the construction of bands. The second method uses also the asymptotic normality of the maximum like lihood estimator and is based on a classical equality by (Landau & Sheep 1970). The third method (bootstrap method) is based on simulations, to estimate the appropriate quantile of the law of a supremum of a Gaussian process. Finally, we conduct a simulation study to investigate its finite-sample properties.Finally, we consider a study of dengue fever, which is a tropical mosquito-borneviral human disease, strictly inter-human. The results show that, the estimated probabilities of infection obtained from our approach are larger than the ones derived from a standard analysis that does not take account of the possible immunity. Inparticular, the estimates provided by our approach suggest that underweight constitutes a major risk factor for dengue infection, irrespectively of age.