Cette thèse est une partie d’un projet français qui s’appelle AEROSTRIP, un partenariat entre l’Institut Pascal, Sigma, SAPPI et Air-France industries, il est financé par le gouvernement français par le programme FUI (20 eme appel). Le projet AEROSTRIP consiste à développer le premier système automatique qui nettoie écologiquement les surfaces des avions et les pièces de rechange en utilisant un abrasif écologique projeté à grande vitesse sur la surface des avions (maïs). Ma thèse consiste à optimiser les trajectoires du système robotique total de telle façon que le décapage de l’avion soit optimal. Le déplacement des robots est nécessaire pour assurer une couverture totale de la surface à décaper parce que ces surfaces sont trop grandes et elles ne peuvent pas être décapées d’une seule position. Le but de mon travail est de trouver le nombre optimal de robots avec leur positions optimales pour décaper totalement l’avion. Une fois ce nombre est déterminé, on cherche les trajectoires des robots entre ces différentes positions. Alors, pour atteindre ce but, j’ai défini un cadre général composant de quatre étapes essentiels: l’étape pre-processing, l’étape optimization algorithm, l’étape generation of the end-effector trajectories et l’étape robot scheduling, assignment and control.Dans ma thèse, j’ai deux contributions dans deux différentes étapes du cadre général: l’étape pre-processing et l’étape optimization algorithm. Le calcul de l’espace de travail du robot est nécessairedans l’étape pre-processing: on a proposé l’Analyse par Intervalles pour trouver cet espace de travail parce qu’il garantie le fait de trouver des solutions dans un temps de calcul raisonnable. Alors, ma première contribution est une nouvelle fonction d’inclusion qui réduit le pessimisme, la surestimation des solutions qui est le principal inconvénient de l’Analyse par Intervalles. La nouvelle fonction d’inclusion est évaluée sur des problèmes de satisfaction de contraintes et des problèmes d’optimisation des contraintes. En plus, on a proposé un algorithme d’optimisation hybride pour trouver le nombre optimal de robots avec leur positions optimales: c’est notre deuxième contribution qui est dans l’étape optimization algorithm. Pour évaluer l’algorithme d’optimisation, on a testé cet algorithme sur des surfaces régulières, comme un cylindre et un hémisphère, et sur un surface complexe: une voiture. / This PHD is a part of a French project named AEROSTRIP, (a partnership between Pascal Institute,Sigma, SAPPI, and Air-France industries), it is funded by the French Government through the FUIProgram (20th call). The AEROSTRIP project aims at developing the first automated system thatecologically cleans the airplanes surfaces using a process of soft projection of ecological media onthe surface (corn). My PHD aims at optimizing the trajectory of the whole robotic systems in orderto optimally strip the airplane. Since a large surface can not be totally covered by a single robot base placement, repositioning of the robots is necessary to ensure a complete stripping of the surface. The goal in this work is to find the optimal number of robots with their optimal positions required to totally strip the air-plane. Once found, we search for the trajectories of the robots of the multi-robot system between those poses. Hence, we define a general framework to solve this problem having four main steps: the pre-processing step, the optimization algorithm step, the generation of the end-effector trajectories step and the robot scheduling, assignment and control step.In my thesis, I present two contributions in two different steps of the general framework: the pre-processing step, the optimization algorithm step. The computation of the robot workspace is required in the pre-processing step: we proposed Interval Analysis to find this workspace since it guarantees finding solutions in a reasonable computation time. Though, our first contribution is a new inclusion function that reduces the pessimism, the overestimation of the solution, which is the main disadvantage of Interval Analysis. The proposed inclusion function is assessed on some Constraints Satisfaction Problems and Constraints Optimization problems. Furthermore, we propose an hybrid optimization algorithm in order to find the optimal number of robots with their optimal poses: it is our second contribution in the optimization algorithm step. To assess our hybrid optimization algorithm, we test the algorithm on regular surfaces, such as a cylinder and a hemisphere, and on a complex surface: a car.

Chemically ordered FePt is one of the most promising materials to reach the ultimate limitations in storage density of future magnetic recording devices due to its high uniaxial magnetocrystalline anisotropy and a corrosion resistance superior to rare-earth based magnets. In this study, FePt and FePt/Cu bilayers have been sputter deposited at room temperature onto thermally oxidized silicon wafers, glass substrates and self-assembled arrays of spherical SiO2 particles with diameters down to 10 nm. Millisecond flash lamp annealing, as well as conventional rapid thermal annealing was employed to induce the phase transformation from the chemically disordered A1 phase into the chemically ordered L10 phase. The influence of the annealing temperature, annealing time and the film thickness on the ordering transformation and (001) texture evolution of FePt films with near equiatomic composition was studied. Whereas flash lamp annealed FePt films exhibit a polycrystalline morphology with high chemical L10 order, rapid thermal annealing can lead to the formation of chemically ordered FePt fifilms with (001) texture on amorphous SiO2/Si substrates. The resultant high perpendicular magnetic anisotropy and large coercivities up to 40 kOe are demonstrated. Simultaneuosly to the ordering transformation, rapid thermal annealing to temperatures exceeding 600 °C leads to a break up of the continuous FePt film into separated islands. This dewetting behavior was utilized to create regular arrays of FePt nanostructures on SiO2 particle templates with periods down to 50 nm. The addition of Cu improves the (001) texture formation and chemcial ordering for annealing temperatures T < 600 °C. In addition, the magnetic anisotropy and the coercivity of the ternary FePtCu alloy can be effectively tailored by adjusting the Cu content. The prospects of FePtCu based exchange spring media, as well as the magnetic properties of FePtCu nanostructures fabricated using e-beam and nanoimprint lithography have been investigated.

The primary goal of artificial intelligence research is to develop a machine capable of learning to solve disparate real-world tasks autonomously, without relying on specialized problem-specific inputs. This dissertation suggests that such machines are realistic: If No Free Lunch theorems were to apply to all real-world problems, then the world would be utterly unpredictable. In response, the dissertation proposes the information-maximization principle, which claims that the optimal optimization methods make the best use of the information available to them. This principle results in a new algorithm, evolutionary annealing, which is shown to perform well especially in challenging problems with irregular structure. / text

Ziel dieser Arbeit ist zunächst die Entwicklung einer Vergleichsgrundlage, auf Basis derer Algorithmen zur Berechnung der Zustandsdichte verglichen werden können. Darauf aufbauend wird ein bestehendes übergangsmatrixbasiertes Verfahren für das großkanonisch Ensemble um ein neues Auswerteverfahren erweitert. Dazu werden numerische Untersuchungen verschiedener Monte-Carlo-Algorithmen zur Berechnung der Zustandsdichte durchgeführt. Das Hauptaugenmerk liegt dabei auf Verfahren, die auf Übergangsmatrizen basieren, sowie auf dem Verfahren von Wang und Landau. Im ersten Teil der Forschungsarbeit wird ein umfassender Überblick über Monte-Carlo-Methoden und Auswerteverfahren zur Bestimmung der Zustandsdichte sowie über verwandte Verfahren gegeben. Außerdem werden verschiedene Methoden zur Berechnung der Zustandsdichte aus Übergangsmatrizen vorgestellt und diskutiert. Im zweiten Teil der Arbeit wird eine neue Vergleichsgrundlage für Algorithmen zur Bestimmung der Zustandsdichte erarbeitet. Dazu wird ein neues Modellsystem entwickelt, an dem verschiedene Parameter frei gewählt werden können und für das die exakte Zustandsdichte sowie die exakte Übergangsmatrix bekannt sind. Anschließend werden zwei weitere Systeme diskutiert für welche zumindest die exakte Zustandsdichte bekannt ist: das Ising Modell und das Lennard-Jones System. Der dritte Teil der Arbeit beschäftigt sich mit numerischen Untersuchungen an einer Auswahl der vorgestellten Verfahren. Auf Basis der entwickelten Vergleichsgrundlage wird der Einfluss verschiedener Parameter auf die Qualität der berechneten Zustandsdichte quantitativ bestimmt. Es wird gezeigt, dass Übergangsmatrizen in Simulationen mit Wang-Landau-Verfahren eine wesentlich bessere Zustandsdichte liefern als das Verfahren selbst. Anschließend werden die gewonnenen Erkenntnisse genutzt um ein neues Verfahren zu entwickeln mit welchem die Zustandsdichte mittels Minimierung der Abweichungen des detaillierten Gleichgewichts aus großen, dünnbesetzten Übergangsmatrizen gewonnen werden kann. Im Anschluss wird ein Lennard-Jones-System im großkanonischen Ensemble untersucht. Es wird gezeigt, dass durch das neue Verfahren Zustandsdichte und Dampfdruckkurve bestimmt werden können, welche qualitativ mit Referenzdaten übereinstimmen.

Chemically ordered FePt is one of the most promising materials to reach the ultimate limitations in storage density of future magnetic recording devices due to its high uniaxial magnetocrystalline anisotropy and a corrosion resistance superior to rare-earth based magnets. In this study, FePt and FePt/Cu bilayers have been sputter deposited at room temperature onto thermally oxidized silicon wafers, glass substrates and self-assembled arrays of spherical SiO2 particles with diameters down to 10 nm. Millisecond flash lamp annealing, as well as conventional rapid thermal annealing was employed to induce the phase transformation from the chemically disordered A1 phase into the chemically ordered L10 phase. The influence of the annealing temperature, annealing time and the film thickness on the ordering transformation and (001) texture evolution of FePt films with near equiatomic composition was studied. Whereas flash lamp annealed FePt films exhibit a polycrystalline morphology with high chemical L10 order, rapid thermal annealing can lead to the formation of chemically ordered FePt fifilms with (001) texture on amorphous SiO2/Si substrates. The resultant high perpendicular magnetic anisotropy and large coercivities up to 40 kOe are demonstrated. Simultaneuosly to the ordering transformation, rapid thermal annealing to temperatures exceeding 600 °C leads to a break up of the continuous FePt film into separated islands. This dewetting behavior was utilized to create regular arrays of FePt nanostructures on SiO2 particle templates with periods down to 50 nm. The addition of Cu improves the (001) texture formation and chemcial ordering for annealing temperatures T < 600 °C. In addition, the magnetic anisotropy and the coercivity of the ternary FePtCu alloy can be effectively tailored by adjusting the Cu content. The prospects of FePtCu based exchange spring media, as well as the magnetic properties of FePtCu nanostructures fabricated using e-beam and nanoimprint lithography have been investigated.

Im Fokus der vorliegenden Arbeit lag die Untersuchung polykristalliner TiO2:Ta-Schichten, hergestellt mittels Gleichstrom-Magnetron-Sputtern durch Verwendung reduzierter keramischer Targets und anschließender thermischer Nachbehandlung im Vakuum der zunächst nichtleitfähigen amorphen Precursorschichten. Es wurden die physikalischen Zusammenhänge, welche die strukturellen, elektrischen und optischen Eigenschaften der kristallinen TiO2:Ta-Schichten beeinflussen analysiert und dabei eine empfindliche Abhängigkeit vom Sauerstofffluss während der Abscheidung festgestellt. Es zeigte sich, dass die Verringerung der kinetischen Energie der Plasmateilchen beim Magnetron-Sputtern durch die Erhöhung des Gesamtdruckes vorteilhaft ist, um das Wachstum des gegenüber Rutil besser leitfähigen Anatas in Verbindung mit dem für niedrige Widerstände notwendigen Sauerstoffdefizit zu realisieren. Bei einem Gesamtdruck von 2 Pa abgeschiedene polykristalline TiO2:Ta-Schichten haben einen spezifischen Widerstand von 1,5·10-3 Ωcm, eine hohe Ladungsträgermobilität (≈8 cm2V-1s-1) und einen geringen Extinktionskoeffizienten von 0,006. Die Abhängigkeit des elektrischen Widerstandes vom Sauerstoffdefizit in der TiO2:Ta-Schicht wurde unter dem Gesichtspunkt der Ladungsträgeraktivierung sowie der Bildung von Ti-Fehlstellen diskutiert, welche vermutlich zur Kompensation und Lokalisierung von freien Elektronen beitragen. Darüber hinaus wurde zur effizienteren Gestaltung der thermischen Nachbehandlung die konventionelle Vakuumtemperung erstmalig erfolgreich durch die Blitzlampentemperung ersetzt. / The work is focused on understanding the physical processes responsible for the modification of the structural, electrical and optical properties of polycrystalline TiO2:Ta films formed by vacuum annealing of initially not conductive amorphous films deposited by direct current magnetron sputtering. It is shown that the oxygen deficiency of amorphous and annealed TiO2:Ta films, respectively, is critical to achieve low resistivity and high optical transmittance of the crystalline films. Increasing the total pressure during magnetron sputter deposition is shown to be beneficial to achieve the desired oxygen deficient anatase growth, which is discussed in terms of energetic particle bombardment. Polycrystalline anatase TiO2:Ta films of low electrical resistivity (1,5·10-3 Ωcm), high free electron mobility (≈8 cm2V-1s-1), and low extinction (0,006) are obtained in this way at a total pressure of 2 Pa. The dependence of the polycrystalline film electrical properties on the oxygen content is discussed in terms of Ta dopant electrical activation as well as transport limiting processes taking into account the formation of Ti-vacancies. In addition, the conventional vacuum annealing has been successfully substituted by the flash lamp annealing in the millisecond range.

Thesis (M.Com.(quantitative Management)) / Satellite Scheduling Problems (SSP) are NP-hard and constraint programming and metaheuristics solution methods yield mixed results. This study investigates a new version of the SSP, the Satellite Constellation Time-Window Optimization Problem (SCoTWOP), involving commercial satellite constellations that provide frequent earth coverage. The SCoTWOP is related to the dual of the Vehicle Routing Problem with Multiple Timewindows, suggesting binary solution vectors representing an activation of time-windows. This representation fitted well with the MatLab® Genetic Algorithm and Direct Search Toolbox subsequently used to experiment with genetic algorithms, tabu search, and simulated annealing as SCoTWOP solution methods. The genetic algorithm was most successful and in some instances activated all 250 imaging time-windows, a number that is typical for a constellation of six satellites. / Quantitative Management

Les résultats présentés dans cette thèse précisent certains aspects de la fonction du cotransporteur Na+/glucose (SGLT1), une protéine transmembranaire qui utilise le gradient électrochimique favorable des ions Na+ afin d’accumuler le glucose à l’intérieur des cellules épithéliales de l’intestin grêle et du rein. Nous avons tout d’abord utilisé l’électrophysiologie à deux microélectrodes sur des ovocytes de xénope afin d’identifier les ions qui constituaient le courant de fuite de SGLT1, un courant mesuré en absence de glucose qui est découplé de la stoechiométrie stricte de 2 Na+/1 glucose caractérisant le cotransport. Nos résultats ont démontré que des cations comme le Li+, le K+ et le Cs+, qui n’interagissent que faiblement avec les sites de liaison de SGLT1 et ne permettent pas les conformations engendrées par la liaison du Na+, pouvaient néanmoins générer un courant de fuite d’amplitude comparable à celui mesuré en présence de Na+. Ceci suggère que le courant de fuite traverse SGLT1 en utilisant une voie de perméation différente de celle définie par les changements de conformation propres au cotransport Na+/glucose, possiblement similaire à celle empruntée par la perméabilité à l’eau passive. Dans un deuxième temps, nous avons cherché à estimer la vitesse des cycles de cotransport de SGLT1 à l’aide de la technique de la trappe ionique, selon laquelle le large bout d’une électrode sélective (~100 μm) est pressé contre la membrane plasmique d’un ovocyte et circonscrit ainsi un petit volume de solution extracellulaire que l’on nomme la trappe. Les variations de concentration ionique se produisant dans la trappe en conséquence de l’activité de SGLT1 nous ont permis de déduire que le cotransport Na+/glucose s’effectuait à un rythme d’environ 13 s-1 lorsque le potentiel membranaire était fixé à -155 mV. Suite à cela, nous nous sommes intéressés au développement d’un modèle cinétique de SGLT1. En se servant de l’algorithme du recuit simulé, nous avons construit un schéma cinétique à 7 états reproduisant de façon précise les courants du cotransporteur en fonction du Na+ et du glucose extracellulaire. Notre modèle prédit qu’en présence d’une concentration saturante de glucose, la réorientation dans la membrane de SGLT1 suivant le relâchement intracellulaire de ses substrats est l’étape qui limite la vitesse de cotransport. / The results presented in this thesis clarify certain functional aspects of the Na+/glucose cotransporter (SGLT1), a membrane protein which uses the downhill electrochemical gradient of Na+ ions to drive the accumulation of glucose in epithelial cells of the small intestine and the kidney. We first used two microelectrodes electrophysiology on Xenopus oocytes to indentify the ionic species mediating the leak current of SGLT1, a current measured in the absence of glucose that is uncoupled from the strict 2 Na+/1 glucose stoichiometry characterising cotransport. Our results showed that cations such as Li+, K+ and Cs+, which interact weakly with SGLT1 binding sites and are unable to generate the conformational changes that are triggered by Na+ binding, were however able to generate leak currents similar in amplitude to the one measured in the presence of Na+. This suggests that the leak current permeating through SGLT1 does so using a pathway that differs from the conformational changes associated with Na+/glucose cotransport. Moreover, it was found that the cationic leak and the passive water permeability could share a common pathway. We then sought to estimate the turnover rate of SGLT1 using the ion-trap technique, where a large tip ion-selective electrode (~100 μm) is pushed against the oocyte plasma membrane, thus enclosing a small volume of extracellular solution referred to as the trap. The variations in ionic concentration occurring in the trap as a consequence of SGLT1 activity made it possible to assess that the turnover rate of Na+/glucose cotransport was 13 s-1 when the membrane potential was clamped to -155 mV. As a last project, we focused our interest on the development of a kinetic model for SGLT1. Taking advantage of the simulated annealing algorithm, we constructed a 7-state kinetic scheme whose predictions accurately reproduced the currents of the cotransporter as a function of extracellular Na+ and glucose. According to our model, the rate limiting step of cotransport under a saturating glucose concentration is the reorientation of the empty carrier that follows the intracellular release of substrates.

Thesis (M.Com.(quantitative Management)) / Satellite Scheduling Problems (SSP) are NP-hard and constraint programming and metaheuristics solution methods yield mixed results. This study investigates a new version of the SSP, the Satellite Constellation Time-Window Optimization Problem (SCoTWOP), involving commercial satellite constellations that provide frequent earth coverage. The SCoTWOP is related to the dual of the Vehicle Routing Problem with Multiple Timewindows, suggesting binary solution vectors representing an activation of time-windows. This representation fitted well with the MatLab® Genetic Algorithm and Direct Search Toolbox subsequently used to experiment with genetic algorithms, tabu search, and simulated annealing as SCoTWOP solution methods. The genetic algorithm was most successful and in some instances activated all 250 imaging time-windows, a number that is typical for a constellation of six satellites. / Quantitative Management

Résumé indisponible. / A healthcare network or health system consists of all organizations, actions and people who participate to promote, restore or maintain people’s health. The health care systems in many developed countries are facing increasing costs. The major reason is the changing age distribution of the population with more elderly people in need of support. Increasing healthcare costs has created new alternatives to traditional hospitalization in which one is Home Health Care (HHC). Home health care or domiciliary care is the provision of health care and assistance to people in their own homes, according to a formal assessment of their needs. HHC has attained a specific place in healthcare network. HHC programs have now been successfully implemented in many countries. The purpose of HHC is to provide the care and support needed to assist patients to live independently in their own homes. HHC is primarily performed by means of personal visitations of healthcare workers to patients in their homes, where they provide care assistance according to patients’ needs. In this thesis we have considered different aspects of planning problems for home health care services. The efficient use of resources is necessary in continuous healthcare services. To meet the increased demand of HHC, operation research specialist can play an important role by solving the various combinatorial optimization problems arising in HHC. These problems can be tactical, strategic or operational with respect to planning horizon. Strategic problems are those which help in attaining long term goals or objectives, e.g. higher level of quality for HHC patients and efficient use of resources. These strategic objectives can be achieved through tactical i.e. medium term panning and operational planning i.e. short term planning. The main purpose of our thesis is to identify these potential optimization problems and solve them via recent metaheuristics. HHC is an alternative to traditional hospitalization and has got a significant share in the organization of healthcare in developed countries. The change in aging demographics, recent development in technology and the increase in the demand of healthcare services are major reasons for this rapid growth. Some studies show HHC as a tool to reduce costs of care, which is a major preoccupation in developed countries. Some others reveal that it leads to the improvement of patients’ satisfaction without increasing the resources. Home health care, i.e. visiting and nursing patients in their homes, is a flourishing realm in the medical industry. The number of companies has grown largely both in public and private sectors. The staffing needs for HHC companies have been expanded as well. Also they face the problem of assigning geographically dispersed patients to home healthcare workers and preparing daily schedules for these workers. The challenge of this problem is to combine aspects of vehicle routing and staff rostering. Both of them are well known NP- hard combinatorial optimization problems, it means the amount of computational time required to find solution increases exponentially with problem size. Home healthcare workers scheduling problem is difficult to solve optimally due to presence of large number of constraints. These are two types of constraints: hard constraints and soft constraints. The hard constraints are the restrictions to be fulfilled for the schedules to be applicable and soft constraints are preferences to improve the quality of these schedules. (...)