The principal focus of this thesis is the exploration of sparse structures in a variety of statistical modelling problems. While more comprehensive models can be useful to solve a larger number of problems, its calculation may be ill-posed in most practical instances because of the sparsity of informative features in the data. If this sparse structure can be exploited, the models can often be solved very efficiently. The thesis is composed of four projects. Firstly, feature sparsity is incorporated to improve the performance of support vector machines when there are a lot of noise features present. The second project is about an empirical study on how to construct an optimal cascade structure. The third project involves the design of a progressive, rate-distortionoptimized shape coder by combining zero-tree algorithm with beamlet structure. Finally, the longest run statistics is applied for the detection of a filamentary structure in twodimensional rectangular region. The fundamental ideas of the above projects are common — extract an efficient summary from a large amount of data. The main contributions of this work are to develop and implement novel techniques for the efficient solutions of several dicult problems that arise in statistical signal/image processing.

This thesis presents a new cascaded iterative VLSI sorting architecture that can accelerate data sorting of variable-length sequences. The proposed sorter mainly consists of a central data memory block, a core comparison unit, and a special address generation module. Many fast sorting algorithms can be represented by a network of compare-and-swap (C&S) operations which can be divided into several processing steps. Instead of using parallel C&S functional units to perform C&S operations of the same sorting step, our comparison unit is composed of cascaded C&S units connected through data commutator such that different sorting steps can be processed simultaneously. The advantage of cascaded architecture is that the number of data memory accesses can be reduced by a factor equal to the number of cascaded stages. However, how to reduce the overhead of the data commutator becomes the most critical design issue. This thesis has explored the feature of C&S operation order in Bitonic sorting such that much simpler and more regular data commutator module can be achieved compared with the previous cascade design derived based on Batcher sorting. Therefore, the cascade level of our sorter architecture can be more than 2. A sample of 4-level cascade sorter has been implemented in our thesis. To generate the address sequence suitable for the proposed cascaded comparison unit, this paper proposes a low-cost address generator design based on the bit-permutation technique. Although high cascade level can lead to significant reduction of memory access which can help reducing the power dissipation, the issues of low hardware utilization for short data sequences and the increasing commutator overhead cannot be neglected. Therefore, to achieve further speed-up, this paper also adopts another parallelism approach for data sorter design by utilizing block-level C&S units which can compare a block of data at the same time. The block-level C&S units can be designed based on traditional Batcher¡¦s sorting network. Based on the proposed Bitonic cascade and Batcher¡¦s block sorting approaches, very fast and low-power sorter hardware can be achieved.

No description available.

Two topics involving irradiation damage in alpha-iron have been considered. First, damage cascades representative of those that would be induced by radiation have been simulated using molecular dynamics (MD). The number and type of defects produced are compared for pure iron and iron with a small hydrogen concentration. Second, the inter- action energy between point defects and line dislocations has been calculated for a number of configurations, using both molecular statics methods and calculations based on linear elastic continuum theory and the dipole force tensor. Results from both methods are com- pared. Results from these two topics are relevant for predicting macroscopic behaviors such as creep and plasticity in reactor structural materials.

Dans ce travail de thèse, on aborde la manifestation des effets d'Électrodynamique Quantique en Cavité (EDQC) dans le domaine THz. Le domaine THz est définie comme la région des fréquences entre 300 GHz et 30 THz (longueurs d'onde entre 10 µm et 1000 µm). Les sources que nous employons pour la génération rayonnement THz sont des cascades quantiques électroluminescentes<br />GaAs/GaAlAs.<br /><br />Les longueurs d'ondes rayonnées (l ~100 µm) sont ainsi beaucoup plus grandes que l'épaisseur typique des émetteurs (~1 µm). Dans ces conditions, lorsque la source est mise dans une cavité métallique planaire, le taux de l'émission spontanée croît comme l'inverse de l'épaisseur de la cavité.<br /><br />Le travail de thèse comporte d'abord un analyse théorique des dispositifs. Le taux d'émission spontanée est étudié à la fois dans le formalisme classique du champ rétro-réfléchi, et dans le<br />formalisme quantique du règle d'or de Fermi. Pour la dernière approche on a développé une méthode numérique pour le calcul et la normalisation des modes optiques supportés par un système<br />multicouches quelconque, qui peut être appliqué à l'étude des dispositifs lasers. L'analyse des cavités planaires sub-longueur d'onde permet de distinguer un effet "utile", qui porte sur le mode TM_0 guidé dans la cavité, et un effet d'absorption, associé aux plasmons de surface supporté par les<br />couches de contact dopées qui sont nécessaires pour l'injection électrique dans la cascade quantique.<br /><br />Le problème d'extraction de l'émission de la cavité est ensuite analysé. Pour cela on étudie un dispositif photonique de nouveau type, la "cavité complexe", dans lequel on a remplacé le miroir<br />métallique supérieur par un réseau métallique. Un modèle de diffraction conique général est développé pour cette étude. Ce modèle a été vérifié dans le domaine THz par mesures de transmission. Il est appliqué dans le cas de rayonnement dipolaire à l'intérieur de la cavité complexe. Le comportement de la puissance extraite de la cavité en fonction de la période du réseau et de l'épaisseur de la cavité permet de reconnaître les effets d'Électrodynamique Quantique en Cavité dans le domaine THz.<br /><br />Pour la mise en évidence expérimentale des effets EDQC, nous avons fabriqué et étudié des cavités complexes avec une cascades quantique. La croissance de type MOCVD, peu exploitée jusqu'à<br />maintenant pour la fabrication des dispositifs à cascade quantique, a été mis au point avec la perspective de réalisation des dispositifs lasers. Notre étude expérimentale confirme les<br />prévisions théoriques. Un facteur de renforcement de l'émission spontanée de ~ 50 est démontré pour la structure de l'épaisseur la plus faible 0.4 µm. Nous avons ainsi mis en évidence, pour la première fois, l'effet de réduction de l'épaisseur du dispositif rayonnant sur l'émission spontanée dans<br />le domaine THz.<br /><br />Parmi les perspectives sur lesquelles cette étude débouche, on peut citer la réduction de la taille latérale des dispositifs pour obtenir un effet Purcell 3D, et également l'exploitation du plasmon des couches des contact dopées pour la conception et la fabrication des dispositifs photoniques THz compacts de nouveau type.

A novel application of the Pauson-Khand reaction was applied to the synthesis of a series of bridged azatricyclic piperazines. This method represents the first application of the Pauson-Khand reaction to synthesize azabridged scaffolds. The ubiquity of bridged azabicyclic ring systems in biologically active natural product skeletons has provided the synthetic chemist with a wealth of opportunity for development over the last century. To this day, the development of new methodologies to tackle these structurally challenging systems remains at the forefront of synthetic chemistry. During our efforts to achieve a total synthesis of the stemofoline alkaloids, we have thus far developed a novel and scalable synthetic strategy to access the fully functionalized caged azatricyclic core of these challenging alkaloids. The overall synthetic strategy we have implemented began with the commercially available and affordable 2-deoxy-D-ribose as a chiral starting material. Furthermore, we have developed a novel 1,3-dipole cascade cycloaddition, which was successfully employed as the key step in the construction of the bridged azatricyclic core of the stemofoline alkaloids. / text

Since their initial demonstration in 1994, quantum cascade lasers (QCLs) have become prominent sources of mid-infrared radiation. Over the years, a large scientific and engineering effort has led to a dramatic improvement in their efficiency and power output, with continuous wave operation at room temperature and Watt-level output power now standard. However, beyond this progress, new functionalities and capabilities need to be added to this compact source to enable its integration into consumer-ready systems. Two main areas of development are particularly relevant from an application standpoint and were pursued during the course of this thesis: wavelength control and wavefront engineering of QCLs. The first research direction, wavelength control, is mainly driven by spectroscopic applications of QCLs, such as trace gas sensing, process monitoring or explosive detection. We demonstrated three different capabilities, corresponding to different potential spectroscopic measurement techniques: widely tunable single longitudinal mode lasing, simultaneous lasing on multiple well-defined longitudinal modes, and simultaneous lasing over a broad and continuous range of the spectrum. The second research direction, wavefront engineering of QCLs, i.e. the improvement of their beam quality, is relevant for applications necessitating transmission of the QCL output over a large distance, for example for remote sensing or military countermeasures. To address this issue, we developed plasmonic lenses directly integrated on the facets of QCLs. The plasmonic structures designed are analogous to antenna arrays imparting directionality to the QCLs, as well as providing means for polarization control. Finally, a research interest in plasmonics led us to design passive flat optical elements using plasmonic antennas. All these projects are tied together by the involvement of Fourier analysis as an essential design tool to predict the interaction of light with various gratings and periodic arrays of grooves and scatterers. / Engineering and Applied Sciences

Evolution of neural networks, or neuroevolution, has been a successful approach to many low-level control problems such as pole balancing, vehicle control, and collision warning. However, certain types of problems — such as those involving strategic decision-making — have remained difficult to solve. This dissertation proposes the hypothesis that such problems are difficult because they are fractured: The correct action varies discontinuously as the agent moves from state to state. To evaluate this hypothesis, a method for measuring fracture using the concept of function variation of optimal policies is proposed. This metric is used to evaluate a popular neuroevolution algorithm, NEAT, empirically on a set of fractured problems. The results show that (1) NEAT does not usually perform well on such problems, and (2) the reason is that NEAT does not usually generate local decision regions, which would be useful in constructing a fractured decision boundary. To address this issue, two neuroevolution algorithms that model local decision regions are proposed: RBF-NEAT, which biases structural search by adding basis-function nodes, and Cascade-NEAT, which constrains structural search by constructing cascaded topologies. These algorithms are compared to NEAT on a set of fractured problems, demonstrating that this approach can improve performance significantly. A meta-level algorithm, SNAP-NEAT, is then developed to combine the strengths of NEAT, RBF-NEAT, and Cascade-NEAT. An evaluation in a set of benchmark problems shows that it is possible to achieve good performance even when it is not known a priori whether a problem is fractured or not. A final empirical comparison of these methods demonstrates that they can scale up to real-world tasks like keepaway and half-field soccer. These results shed new light on why constructive neuroevolution algorithms have difficulty in certain domains and illustrate how bias and constraint can be used to improve performance. Thus, this dissertation shows how neuroevolution can be scaled up from learning low-level control to learning strategic decision-making problems. / text

Η χορήγηση φαρμάκων μέσω της αναπνευστικής οδού αποτελεί τον πιο αποτελεσματικό τρόπο αντιμετώπισης νόσων του αναπνευστικού συστήματος όπως το άσθμα. Τα τελευταία χρόνια, η αύξηση στη συχνότητα εμφάνισης του άσθματος δημιούργησε την ανάγκη βελτίωσης της θεραπείας με το σχεδιασμό νέων συσκευών εισπνοής και την ανάπτυξη καλύτερης μορφοποίησης των φαρμάκων ώστε να εξασφαλίζεται ικανοποιητική κατανομή τους στο αναπνευστικό δένδρο. Στην παρούσα μελέτη, λαμβάνοντας υπόψη τις δυσκολίες ανάπτυξης ενός φαρμάκου για εισπνεόμενη χορήγηση, αποφασίσαμε να ασχοληθούμε με τη μορφοποίηση της φουμαρικής φορμοτερόλης, ενός ευρέως χρησιμοποιούμενου βρογχοδιασταλτικού φαρμάκου στη θεραπεία του βρογχικού άσθματος. Στόχος είναι η ανάπτυξη της βέλτιστης μορφοποίησης με την οποία θα επιτυγχάνεται το μέγιστο θεραπευτικό κλάσμα FPD. Διερευνήθηκε η επίδραση διαφόρων παραγόντων στη μεταφορά και εναπόθεση των λεπτόκοκκων σωματιδίων της φουμαρικής φορμοτερόλης στο αναπνευστικό δένδρο. Συγκεκριμένα, μελετήθηκε η επίδραση της κατανομής μεγέθους των σωματιδίων της λακτόζης στη μορφοποίηση και προσδιορίστηκε η κατανομή που επιτρέπει ικανοποιητική προσκόλληση της φουμαρικής φορμοτερόλης στη λακτόζη αλλά και εύκολο διαχωρισμό της απ’ αυτήν κατά την εισπνοή. Στη συνέχεια, διεξήχθησαν πειράματα προσδιορισμού του ιδανικού χρόνου ανάμιξης και των συνθηκών παραγωγής που εξασφαλίζουν ικανοποιητική ροή των κόνεων και πλήρωση των κυαθίων και δεν επηρεάζουν τη χημική σταθερότητα του φαρμάκου. Ακολούθησαν πειράματα για την αξιολόγηση της επίδρασης της μεταβολής στο βάρος πλήρωσης του κυαθίου στη μορφοποίηση και κατ’ επέκταση στο θεραπευτικό κλάσμα FPD. Τέλος, η μελέτη ολοκληρώθηκε μ’ ένα πείραμα ελέγχου της επίδρασης των εργαστηριακών συνθηκών στο Andersen Cascade Impactor και συνεπώς στη σταθερότητα του προϊόντος αλλά και την ικανότητα διασποράς του. / Drug delivery through the respiratory tract constitutes the most effective way of treating respiratory diseases such as asthma. Over the last years, the increased frequency of asthma occurrence challenged the improvement of therapy with the design of new inhalers and a better formulation development of drugs so as to achieve effective drug distribution to the respiratory tree. In the current study, having taken into consideration the difficulties of drug development for inhalation delivery, we have decided to work on the formulation of formoterol fumarate, a bronchodilator broadly used to the treatment of bronchial asthma. The aim is the optimization of formulation development so as to achieve the optimum therapeutic fraction FPD. The main idea was the examination of the effect of certain parameters on the delivery and deposition of formoterol fumarate fine particles to the respiratory tree. Specifically, the effect of lactose size distribution on the formulation was studied and the size distribution that allows not only effective adherence of formoterol fumarate to lactose but also efficient release during inhalation was determined. As a next step, experiments were carried out for the determination of the ideal mixing time and manufacturing conditions for satisfactory powder flowability, blister filling and chemical stability of drug. Experiments for the evaluation of the effect of blister filling variation on the formulation and consequently on the therapeutic fraction FPD followed. Finally, this study was completed by controlling the effect of laboratory conditions on the Andersen Cascade Impactor and consequently on the product stability and its dispersion ability.

Les transitions intersousbandes dans les hétérostructures de nitrure d'éléments III ont été intensément étudiées dans le proche infrarouge pour des applications télécoms. L'accordabilité dans le proche infrarouge est rendu possible grâce à la discontinuité de potentiel en bande de conduction qui peut atteindre 1.75 eV pour le système GaN/AlN. Les matériaux nitrures suscitent actuellement un grand intérêt à plus grande longueur d'onde infrarouge. C'est par exemple le développement de détecteurs et d'imageurs rapides à cascade quantique dans la gamme 2-5 µm. C'est aussi l'extension des dispositifs intersousbandes dans le domaine de fréquences THz. Ce travail de thèse porte sur l'étude des transitions intersousbandes dans les puits quantiques GaN/Al(Ga)N épitaxiés par jets moléculaires. Le but est d'accorder ces transitions dans une gamme spectrale très large allant du proche au lointain infrarouge. Je montre que les transitions ISB peuvent être accordées dans la gamme 1-12 µm dans les puits quantiques GaN/AlGaN en phase hexagonale synthétisés selon l'axe polaire c [0001]. Ceci impose l'ingénierie du champ électrique interne, dont la valeur peut atteindre dans le GaN 10 MV/cm. Une solution alternative consiste à utiliser une orientation particulière, dite semipolaire, qui conduit à une réduction du champ électrique interne le long de l'axe de croissance [11-22]. J'ai montré que cette réduction du champ interne permet d'accorder les résonances intersousbandes des puits quantiques GaN/AlN dans le proche infrarouge et j'ai pu estimer le champ en comparant les résultats de spectroscopie et simulations. J'ai d'autre part étudié les propriétés interbandes et intersousbandes des puits quantiques de symétrie cubique, qui par raison de symétrie, ne présentent pas de champ électrique interne. Finalement j'ai mis en évidence les premières transitions intersousbandes aux fréquences THz dans les puits quantiques GaN/AlGaN polaires mais aussi cubiques.