We study minimization problems with respect to a one-parameter family of generalized relative entropies. These relative entropies, which we call relative -entropies (denoted I (P; Q)), arise as redundancies under mismatched compression when cumulants of compression lengths are considered instead of expected compression lengths. These parametric relative entropies are a generalization of the usual relative entropy (Kullback-Leibler divergence). Just like relative entropy, these relative -entropies behave like squared Euclidean distance and satisfy the Pythagorean property. We explore the geometry underlying various statistical models and its relevance to information theory and to robust statistics. The thesis consists of three parts. In the first part, we study minimization of I (P; Q) as the first argument varies over a convex set E of probability distributions. We show the existence of a unique minimizer when the set E is closed in an appropriate topology. We then study minimization of I on a particular convex set, a linear family, which is one that arises from linear statistical constraints. This minimization problem generalizes the maximum Renyi or Tsallis entropy principle of statistical physics. The structure of the minimizing probability distribution naturally suggests a statistical model of power-law probability distributions, which we call an -power-law family. Such a family is analogous to the exponential family that arises when relative entropy is minimized subject to the same linear statistical constraints. In the second part, we study minimization of I (P; Q) over the second argument. This minimization is generally on parametric families such as the exponential family or the - power-law family, and is of interest in robust statistics ( > 1) and in constrained compression settings ( < 1). In the third part, we show an orthogonality relationship between the -power-law family and an associated linear family. As a consequence of this, the minimization of I (P; ), when the second argument comes from an -power-law family, can be shown to be equivalent to a minimization of I ( ; R), for a suitable R, where the first argument comes from a linear family. The latter turns out to be a simpler problem of minimization of a quasi convex objective function subject to linear constraints. Standard techniques are available to solve such problems, for example, via a sequence of convex feasibility problems, or via a sequence of such problems but on simpler single-constraint linear families.

This study consisted of a qualitative analysis of articles from two Canadian newspapers related to North America’s only safe injection facility for drug users, Vancouver’s Insite, and examined the texts for latent themes derived from a review of harm reduction and governmentality literature. The investigation asked “In what ways are Insite and its clients represented in the media and what implications do those portrayals have in terms of Insite’s operation as a harm reduction practice as well as a governmental strategy designed to direct the conduct of drug users who visit the site?” The analysis revealed conflicting representations, some which have positive potential in terms of Insite’s adherence to the fundamental principles of harm reduction and others that undermined those principles and suggested that the site may have traditional governmental functions, perhaps indicating less distance between the harm reduction and governmentality philosophies in the discourse surrounding the SIS than expected.

Les problèmes d’optimisation combinatoire sont généralement réputés NP-difficiles, donc il n’y a pas d’algorithmes efficaces pour les résoudre. Afin de trouver des solutions optimales locales ou réalisables, on utilise souvent des heuristiques ou des algorithmes approchés. Les dernières décennies ont vu naitre des méthodes approchées connues sous le nom de métaheuristiques, et qui permettent de trouver une solution approchées. Cette thèse propose de résoudre des problèmes d’optimisation déterministe et stochastique à l’aide de métaheuristiques. Nous avons particulièrement étudié la méthode de voisinage variable connue sous le nom de VNS. Nous avons choisi cet algorithme pour résoudre nos problèmes d’optimisation dans la mesure où VNS permet de trouver des solutions de bonne qualité dans un temps CPU raisonnable. Le premier problème que nous avons étudié dans le cadre de cette thèse est le problème déterministe de largeur de bande de matrices creuses. Il s’agit d’un problème combinatoire difficile, notre VNS a permis de trouver des solutions comparables à celles de la littérature en termes de qualité des résultats mais avec temps de calcul plus compétitif. Nous nous sommes intéressés dans un deuxième temps aux problèmes de réseaux mobiles appelés OFDMA-TDMA. Nous avons étudié le problème d’affectation de ressources dans ce type de réseaux, nous avons proposé deux modèles : Le premier modèle est un modèle déterministe qui permet de maximiser la bande passante du canal pour un réseau OFDMA à débit monodirectionnel appelé Uplink sous contraintes d’énergie utilisée par les utilisateurs et des contraintes d’affectation de porteuses. Pour ce problème, VNS donne de très bons résultats et des bornes de bonne qualité. Le deuxième modèle est un problème stochastique de réseaux OFDMA d’affectation de ressources multi-cellules. Pour résoudre ce problème, on utilise le problème déterministe équivalent auquel on applique la méthode VNS qui dans ce cas permet de trouver des solutions avec un saut de dualité très faible. Les problèmes d’allocation de ressources aussi bien dans les réseaux OFDMA ou dans d’autres domaines peuvent aussi être modélisés sous forme de problèmes d’optimisation bi-niveaux appelés aussi problèmes d’optimisation hiérarchique. Le dernier problème étudié dans le cadre de cette thèse porte sur les problèmes bi-niveaux stochastiques. Pour résoudre le problème lié à l’incertitude dans ce problème, nous avons utilisé l’optimisation robuste plus précisément l’approche appelée « distributionnellement robuste ». Cette approche donne de très bons résultats légèrement conservateurs notamment lorsque le nombre de variables du leader est très supérieur à celui du suiveur. Nos expérimentations ont confirmé l’efficacité de nos méthodes pour l’ensemble des problèmes étudiés. / Combinatorial optimization problems are generally NP-hard problems, so they can only rely on heuristic or approximation algorithms to find a local optimum or a feasible solution. During the last decades, more general solving techniques have been proposed, namely metaheuristics which can be applied to many types of combinatorial optimization problems. This PhD thesis proposed to solve the deterministic and stochastic optimization problems with metaheuristics. We studied especially Variable Neighborhood Search (VNS) and choose this algorithm to solve our optimization problems since it is able to find satisfying approximated optimal solutions within a reasonable computation time. Our thesis starts with a relatively simple deterministic combinatorial optimization problem: Bandwidth Minimization Problem. The proposed VNS procedure offers an advantage in terms of CPU time compared to the literature. Then, we focus on resource allocation problems in OFDMA systems, and present two models. The first model aims at maximizing the total bandwidth channel capacity of an uplink OFDMA-TDMA network subject to user power and subcarrier assignment constraints while simultaneously scheduling users in time. For this problem, VNS gives tight bounds. The second model is stochastic resource allocation model for uplink wireless multi-cell OFDMA Networks. After transforming the original model into a deterministic one, the proposed VNS is applied on the deterministic model, and find near optimal solutions. Subsequently, several problems either in OFDMA systems or in many other topics in resource allocation can be modeled as hierarchy problems, e.g., bi-level optimization problems. Thus, we also study stochastic bi-level optimization problems, and use robust optimization framework to deal with uncertainty. The distributionally robust approach can obtain slight conservative solutions when the number of binary variables in the upper level is larger than the number of variables in the lower level. Our numerical results for all the problems studied in this thesis show the performance of our approaches.

La première partie de cette thèse (articles I et II) est consacrée à l'étude du comportement asymptotique des solutions d'un système dynamique du second ordre avec dissipation évanescente. Le système dynamique est étudié dans sa version continue et dans sa version discrète via un algorithme.La deuxième partie de cette thèse (articles III à VI) est consacrée à l'étude de plusieurs algorithmes de type proximal. Nous montrons que ces algorithmes convergent vers des solutions de certains problèmes de minimisation. Dans chaque cas, une application est donnée dans le cadre de la décomposition de domaine pour les EDP. / The first part of this thesis is devoted to the study of the asymptotic behavior of solutions of a second order dynamic system with vanishing dissipation. The dynamic system is studied in its continuous version and in its discrete version via an algorithm.The second part is about the study of several proximal-type algorithms. We show that these algorithms converge to solutions of some minimization problems. In each case, an application is given in the area of domain decomposition for PDE's.

The result of this project is a sustainability assessment questionnaire for the high school setting. The questionnaire fills a gap in the literature pertaining to assessing direct and indirect environmental impacts on high schools.

Slepá dekonvoluce je metoda, při které je rozptylová funkce a skutečný obraz rekonstruován zároveň. Cílem této práce je představit různé metody slepé dekonvoluce a najít optimální metodu rekonstrukce původního obrazu a rozptylové funkce. Jako nejvhodnější metoda slepé dekonvoluce byl zvolen algoritmus střídavé minimalizace, který byl upraven a testován. Vlastnosti navrženého algoritmu byly testovány na uměle degradovaných datech a na reálných datech pořízených skenovacím transmisním elektronovým mikroskopem. Účinnost algoritmu byla hodnocena hned několika hodnotícími kritérii. Byla zjištěna omezení algoritmu a tím specifikováno jeho využití.

This thesis deals with the analysis of micromotion of a single charged calcium ion trapped inside the linear Paul's ion trap and the influence of residual micromotion on the systematic frequency shifts of the clock transition of calcium ion. The fundamental properties of the motion of an ion confined within linear Paul's ion trap are shown in general using a theoretical description. The micromotion component of the overall motion is especially emphasized. A model expressing micromotion in the axial direction of the trap is introduced on the basis of the results of the numerical calculation of electric fields inside the trap. The model is compared to the reality experimentally. Then, the photon-correlation method of detection of micromotion is introduced and subsequently used to minimize and to estimate a measure of residual micromotion in all spacial directions. According to the achievable measure of residual micromotion, the systematic frequency shifts caused by this micromotion are estimated. It can be seen that we are able to reach uncertainties of the relative frequency shifts due to micromotion below 10^20. We expect that uncertainty of total motional systematic frequency shift is in our case limited by thermal motion.

This diploma thesis deals with an optimization of hydrostatic slide of the large-portal gantry machine FRU (TOS Kuřim). First part of the thesis concerns with an analysis of the current practice in hydrostatic guides in the field of heavy gantry machines, an usage of alternative materials for a construction of carrying parts of machines, and possibilities of current simulation methods. Aim of the second part of this thesis is a brief clarification of the principle of mathematical optimization, and creation of general methodology of construction optimization by means of system analysis. Optimization calculations were processed by ANSYS Workbench 11.0. There is achieved adjusted and optimized design of the hydrostatic slide as the result of the thesis. Moreover, the comparison of key deformations was added. Created methodology is applicable in both an area of construction of gantry machines, and a wide spectrum of construction fields.

Nondeterministic finite automaton is an important tool, which is used to process strings in many different areas of programming. It is important to try to reduce its size for increasing programs' effectiveness. However, this problem is computationally hard, so we need to search for new techniques. Basics of finite automata are described in this work. Some methods for their reduction are then introduced. Usable reduction algorithms are described in greater detail. Then they are implemented and tested. The test results are finally evaluated.

Confidentiel / Confidential