Tesis sobre el tema "Simplification structurale"

Cette thèse concerne divers développements méthodologiques et leurs applications en RMN dans les solvants cristaux liquides chiraux. Nous avons d'abord montré que les expériences comprenant des impulsions sélectives permettaient d'extraire des informations des spectres proton et carbone-13 qui étaient inaccessibles sur les spectres monodimensionnels standards. Nous avons ensuite utilisé d'autres techniques pour mesurer un maximum de couplages dipolaires pour préciser de manière non-ambigue͏̈ les configurations relatives de centres stéréogènes dans des molécules rigides. Nous avons ainsi montré que pour la DHEA, un stéroi͏̈de, une seule géométrie relative des substituants était compatible avec l'ensemble des couplages dipolaires mesurés et conduisait à un écart type entre couplages dipolaires calculés et expérimentaux correct. Cette méthode est promise à un grand avenir car elle est beaucoup moins ambigue͏̈ que les méthodes classiques utilisées en milieu isotrope. Enfin, nous avons développé cette méthode dans le cas des molécules flexibles où le problème se complique car il existe une corrélation totale entre conformation moléculaire et ordre orientationnel. Le nombre de paramètres à itérer explose rapidement, alors que le nombre de mesurables reste constant, et donc un traitement général est impossible. En prenant pour exemple le 1,2-dibromopropane, nous avons montré que le problème est gérable, à condition d'utiliser un modèle paramétrique pour calculer la valeur moyenne des paramètres d'ordre pour chaque conformation. Les paramètres à itérer, sont alors beaucoup moins nombreux et le problème peut être traité. Ainsi, nous avons montré qu'avec quatre indices de liaison et la géométrie des différents conformères obtenue par calcul ab initio, nous pouvions obtenir un excellent accord entre les couplages dipolaires calculés et expérimentaux, et déterminer sans ambigui͏̈té la configuration relative des protons méthyléniques par rapport à la configuration du centre chiral
This thesis relates to various methodological developments and their applications in NMR in chiral liquid crystal solvents. Firstly, we have shown that experiments using selective pulses allowed us to extract information from proton and carbon-13 spectra which were inaccessible on the standard monodimensional spectra. We then used these techniques to measure a maximum of dipolar couplings in order to specify, in an unambiguous way, the relative configurations of stereogenic centres in rigid molecules. We showed that for the steroid DHEA, only one relative geometry of the substituents was compatible with all the measured dipolar couplings and led to a correct standard deviation between calculated and experimental dipolar couplings. This method is extremely promising because it is much less ambiguous than the more traditional methods used in isotropic media. Finally, we developed this method in the case of flexible molecules where the problem becomes more complicated because of the correlation between molecular conformation and orientational order. The number of parameters to be re-iterated explodes rapidly, whereas the number of measurables remain constant, and thus a general treatment is not possible. By studying for example, the 1,2-dibromopropane, we showed that the problem is manageable, under the condition of using a parametric model to calculate the average value of order parameters for each conformation. The parameters to be re-iterated, are then much fewer and the problem can be dealt with. Thus, we showed that, with four bond indices and the geometry of different conformers obtained by ab initio calculations, we could obtain an excellent agreement between the calculated and experimental dipolar couplings, and determine without ambiguity the relative configuration of the methylenic protons compared with the configuration of the chiral center

Le paludisme est la première parasitose en termes de mortalité à l’échelle mondiale. Les thérapies combinées à base d'artémisinine, traitement de première ligne du paludisme à Plasmodium falciparum, font face à des échecs dûs à l’apparition de résistances. Il est donc nécessaire de développer de nouvelles molécules antiplasmodiales possédant un mécanisme d’action novateur. Dans cet objectif, notre laboratoire a précédemment décrit la synthèse et les activités biologiques d'une chimiothèque de molécules azahétérocycliques α-trichlorométhylées, dont une molécule hit en série quinazoline qui présente le meilleur profil biologique.Une première partie de ce travail s’est intéressée à la pharmacomodulation en série 4-carboxamidoquinazoline. Afin de compléter l’étude RSA, la stratégie de scaffold hopping a permis l’obtention de nouvelles molécules en séries quinoxaline et phtalazine. Par simplification structurale, de nouveaux composés en séries pyrimidine, pyridazine et pyrazine ont été obtenus. Enfin, dans le but de moduler la partie benzénique des noyaux quinazoline et quinoxaline, des dérivés en série thiénopyrimidine et pyrido[2,3-b]pyrazine ont été synthétisés. Parmi plus de 110 nouvelles molécules originales synthétisées, plusieurs nouvelles molécules hit ont pu être identifiées. Leurs propriétés physicochimiques et pharmacocinétiques in vitro ont été déterminées en vue d’identifier une molécule candidate pour l’évaluation in vivo. De plus, afin d’élucider le mécanisme d’action de ces composés qui diffère de ceux des antipaludiques commerciaux, nous avons récemment identifié par immunofluorescence que ces molécules possèdent une action sur l’apicoplaste de P. falciparum
Malaria remains the leading cause of death among parasitic infections worldwide. Currently, there are major concerns about the spread of resistance to artemisinin derivatives that are the basis of first-line antimalarial treatment. Therefore, there is an urgent need to develop new antiplasmodial molecules with a novel mechanism of action. For this purpose, our laboratory has previously described the synthesis and biological activities of a chemical library of α-trichloromethylated azaheterocycles including a hit molecule in the quinazoline series which presents the best biological profile.The first part of this work focused on 4-carboxamide quinazoline pharmacomodulation. In order to complete the SARs, scaffold hopping strategies allowed us to obtain new compounds in the quinoxaline and phthalazine series. By structural simplification, new compounds in the pyrimidine, pyridazine and pyrazine series were obtained. Finally, in order to explore the benzene part of the quinazoline and quinoxaline rings, new thienopyrimidine and pyrido[2,3-b]pyrazine derivatives were also synthesized. More than 110 new original molecules were obtained, among them several new hit molecules were obtained. The physicochemical and in vitro pharmacokinetic properties were determined in order to initiate the study of their in vivo activity on Plasmodium berghei. In addition, in order to elucidate the mechanism of action of these compounds, which differs from those of commercial antimalarials, we have recently identified by immunofluorescence that these molecules target the apicoplast of P. falciparum, an organelle essential to parasite survival

Thesis: Ph. D., Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, May, 2020
Cataloged from the official PDF of thesis.
Includes bibliographical references (pages 169-187).
The typical machine learning algorithms looks for a pattern in data, and makes an assumption that the signal to noise ratio of the pattern is high. This approach depends strongly on the quality of the datasets these algorithms operate on, and many complex algorithms fail in spectacular fashion on simple tasks by overfitting noise or outlier examples. These algorithms have training procedures that scale poorly in the size of the dataset, and their out-puts are difficult to intepret. This thesis proposes solutions to both problems by leveraging the theory of optimal transport and proposing efficient algorithms to solve problems in: (1) quantization, with extensions to the Wasserstein barycenter problem, and a link to the classical coreset problem; (2) natural language processing where the hierarchical structure of text allows us to compare documents efficiently;(3) Bayesian inference where we can impose a hierarchy on the label switching problem to resolve ambiguities.
by Sebastian Claici.
Ph. D.
Ph.D. Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science

Since its inception, the notion of thin position has played an important role in low-dimensional topology. Thin position for knots in the 3-sphere was first introduced by David Gabai in order to prove the Property R Conjecture. In addition, this theory factored into Cameron Gordon and John Luecke's proof of the knot complement problem and revolutionized the study of Heegaard splittings upon its adaptation by Martin Scharlemann and Abigail Thompson. Let h be a Morse function from the 3-sphere to the real numbers with two critical points. Loosely, thin position of a knot K in the 3-sphere is a particular embedding of K which minimizes the total number of intersections with a maximal collection of regular level sets, where this number of intersections is called the width of the knot. Although not immediately obvious, it has been demonstrated that there is a close relationship between a thin position of a knot K and essential meridional planar surfaces in its exterior E(K). In this thesis, we study the nature of thin position under knot companionship; namely, for several families of knots we establish a lower bound for the width of a satellite knot based on the width of its companion and the wrapping or winding number of its pattern. For one such class of knots, cable knots, in addition to finding thin position for these knots, we establish a criterion under which non-minimal bridge positions of cable knots are stabilized. Finally, we exhibit an embedding of the unknot whose width must be increased before it can be simplified to thin position.

La validation des logiciels est une partie cruciale dans le cycle de leur développement. Deux techniques de vérification et de validation se sont démarquées au cours de ces dernières années : l'analyse statique et l'analyse dynamique. Les points forts et faibles des deux techniques sont complémentaires. Nous présentons dans cette thèse une combinaison originale de ces deux techniques. Dans cette combinaison, l'analyse statique signale les instructions risquant de provoquer des erreurs à l'exécution, par des alarmes dont certaines peuvent être de fausses alarmes, puis l'analyse dynamique (génération de tests) est utilisée pour confirmer ou rejeter ces alarmes. L'objectif de cette thèse est de rendre la recherche d'erreurs automatique, plus précise, et plus efficace en temps. Appliquée à des programmes de grande taille, la génération de tests, peut manquer de temps ou d'espace mémoire avant de confirmer certaines alarmes comme de vraies erreurs ou conclure qu'aucun chemin d'exécution ne peut atteindre l'état d'erreur de certaines alarmes et donc rejeter ces alarmes. Pour surmonter ce problème, nous proposons de réduire la taille du code source par le slicing avant de lancer la génération de tests. Le slicing transforme un programme en un autre programme plus simple, appelé slice, qui est équivalent au programme initial par rapport à certains critères. Quatre utilisations du slicing sont étudiées. La première utilisation est nommée all. Elle consiste à appliquer le slicing une seule fois, le critère de simplification étant l'ensemble de toutes les alarmes du programme qui ont été détectées par l'analyse statique. L'inconvénient de cette utilisation est que la génération de tests peut manquer de temps ou d'espace et les alarmes les plus faciles à classer sont pénalisées par l'analyse d'autres alarmes plus complexes. Dans la deuxième utilisation, nommée each, le slicing est effectué séparément par rapport à chaque alarme. Cependant, la génération de tests est exécutée pour chaque programme et il y a un risque de redondance d'analyse si des alarmes sont incluses dans d'autres slices. Pour pallier ces inconvénients, nous avons étudié les dépendances entre les alarmes et nous avons introduit deux utilisations avancées du slicing, nommées min et smart, qui exploitent ces dépendances. Dans l'utilisation min, le slicing est effectué par rapport à un ensemble minimal de sous-ensembles d'alarmes. Ces sous-ensembles sont choisis en fonction de dépendances entre les alarmes et l'union de ces sous-ensembles couvre l'ensemble de toutes les alarmes. Avec cette utilisation, on a moins de slices qu'avec each, et des slices plus simples qu'avec all. Cependant, l'analyse dynamique de certaines slices peut manquer de temps ou d'espace avant de classer certaines alarmes, tandis que l'analyse dynamique d'une slice éventuellement plus simple permettrait de les classer. L'utilisation smart consiste à appliquer l'utilisation précédente itérativement en réduisant la taille des sous-ensembles quand c'est nécessaire. Lorsqu'une alarme ne peut pas être classée par l'analyse dynamique d'une slice, des slices plus simples sont calculées. Nous prouvons la correction de la méthode proposée. Ces travaux sont implantés dans sante, notre outil qui relie l'outil de génération de tests PathCrawler et la plate-forme d'analyse statique Frama-C. Des expérimentations ont montré, d'une part, que notre combinaison est plus performante que chaque technique utilisée indépendamment et, d'autre part, que la vérification devient plus rapide avec l'utilisation du slicing. De plus, la simplification du programme par le slicing rend les erreurs détectées et les alarmes restantes plus faciles à analyser

La validation des logiciels est une partie cruciale dans le cycle de leur développement. Deux techniques de vérification et de validation se sont démarquées au cours de ces dernières années : l’analyse statique et l’analyse dynamique. Les points forts et faibles des deux techniques sont complémentaires. Nous présentons dans cette thèse une combinaison originale de ces deux techniques. Dans cette combinaison, l’analyse statique signale les instructions risquant de provoquer des erreurs à l’exécution, par des alarmes dont certaines peuvent être de fausses alarmes, puis l’analyse dynamique (génération de tests) est utilisée pour confirmer ou rejeter ces alarmes. L’objectif de cette thèse est de rendre la recherche d’erreurs automatique, plus précise, et plus efficace en temps. Appliquée à des programmes de grande taille, la génération de tests, peut manquer de temps ou d’espace mémoire avant de confirmer certaines alarmes comme de vraies erreurs ou conclure qu’aucun chemin d’exécution ne peut atteindre l’état d’erreur de certaines alarmes et donc rejeter ces alarmes. Pour surmonter ce problème, nous proposons de réduire la taille du code source par le slicing avant de lancer la génération de tests. Le slicing transforme un programme en un autre programme plus simple, appelé slice, qui est équivalent au programme initial par rapport à certains critères. Quatre utilisations du slicing sont étudiées. La première utilisation est nommée all. Elle consiste à appliquer le slicing une seule fois, le critère de simplification étant l’ensemble de toutes les alarmes du programme qui ont été détectées par l’analyse statique. L’inconvénient de cette utilisation est que la génération de tests peut manquer de temps ou d’espace et les alarmes les plus faciles à classer sont pénalisées par l’analyse d’autres alarmes plus complexes. Dans la deuxième utilisation, nommée each, le slicing est effectué séparément par rapport à chaque alarme. Cependant, la génération de tests est exécutée pour chaque programme et il y a un risque de redondance d’analyse si des alarmes sont incluses dans d’autres slices. Pour pallier ces inconvénients, nous avons étudié les dépendances entre les alarmes et nous avons introduit deux utilisations avancées du slicing, nommées min et smart, qui exploitent ces dépendances. Dans l’utilisation min, le slicing est effectué par rapport à un ensemble minimal de sous-ensembles d’alarmes. Ces sous-ensembles sont choisis en fonction de dépendances entre les alarmes et l’union de ces sous-ensembles couvre l’ensemble de toutes les alarmes. Avec cette utilisation, on a moins de slices qu’avec each, et des slices plus simples qu’avec all. Cependant, l’analyse dynamique de certaines slices peut manquer de temps ou d’espace avant de classer certaines alarmes, tandis que l’analyse dynamique d’une slice éventuellement plus simple permettrait de les classer. L’utilisation smart consiste à appliquer l’utilisation précédente itérativement en réduisant la taille des sous-ensembles quand c’est nécessaire. Lorsqu’une alarme ne peut pas être classée par l’analyse dynamique d’une slice, des slices plus simples sont calculées. Nous prouvons la correction de la méthode proposée. Ces travaux sont implantés dans sante, notre outil qui relie l’outil de génération de tests PathCrawler et la plate-forme d’analyse statique Frama-C. Des expérimentations ont montré, d’une part, que notre combinaison est plus performante que chaque technique utilisée indépendamment et, d’autre part, que la vérification devient plus rapide avec l’utilisation du slicing. De plus, la simplification du programme par le slicing rend les erreurs détectées et les alarmes restantes plus faciles à analyser
Software validation remains a crucial part in software development process. Two major techniques have improved in recent years, dynamic and static analysis. They have complementary strengths and weaknesses. We present in this thesis a new original combination of these methods to make the research of runtime errors more accurate, automatic and reduce the number of false alarms. We prove as well the correction of the method. In this combination, static analysis reports alarms of runtime errors some of which may be false alarms, and test generation is used to confirm or reject these alarms. When applied on large programs, test generation may lack time or space before confirming out certain alarms as real bugs or finding that some alarms are unreachable. To overcome this problem, we propose to reduce the source code by program slicing before running test generation. Program slicing transforms a program into another simpler program, which is equivalent to the original program with respect to certain criterion. Four usages of program slicing were studied. The first usage is called all. It applies the slicing only once, the simplification criterion is the set of all alarms in the program. The disadvantage of this usage is that test generation may lack time or space and alarms that are easier to classify are penalized by the analysis of other more complex alarms. In the second usage, called each, program slicing is performed with respect to each alarm separately. However, test generation is executed for each sliced program and there is a risk of redundancy if some alarms are included in many slices. To overcome these drawbacks, we studied dependencies between alarms on which we base to introduce two advanced usages of program slicing : min and smart. In the min usage, the slicing is performed with respect to subsets of alarms. These subsets are selected based on dependencies between alarms and the union of these subsets cover the whole set of alarms. With this usage, we analyze less slices than with each, and simpler slices than with all. However, the dynamic analysis of some slices may lack time or space before classifying some alarms, while the dynamic analysis of a simpler slice could possibly classify some. Usage smart applies previous usage iteratively by reducing the size of the subsets when necessary. When an alarm cannot be classified by the dynamic analysis of a slice, simpler slices are calculated. These works are implemented in sante, our tool that combines the test generation tool PathCrawler and the platform of static analysis Frama-C. Experiments have shown, firstly, that our combination is more effective than each technique used separately and, secondly, that the verification is faster after reducing the code with program slicing. Simplifying the program by program slicing also makes the detected errors and the remaining alarms easier to analyze

Grâce à l’évolution de la technologie à travers le Web, la documentation relative à la santé est de plus en plus abondante et accessible à tous, plus particulièrement aux patients, qui ont ainsi accès à une panoplie d’informations sanitaires. Malheureusement, la grande disponibilité de l’information médicale ne garantit pas sa bonne compréhension par le public visé, en l’occurrence les non-experts. Notre projet de thèse a pour objectif la création d’une ressource de simplification de textes médicaux, à partir d’une analyse syntaxico-sémantique des verbes dans quatre corpus médicaux en français qui se distinguent de par le degré d’expertise de leurs auteurs et celui des publics cibles. La ressource conçue contient 230 patrons syntaxicosémantiques des verbes (appelés pss), alignés avec leurs équivalents non spécialisés. La méthode semi-automatique d’analyse des verbes appliquée pour atteindre notre objectif est basée sur quatre tâches fondamentales : l’annotation syntaxique des corpus, réalisée grâce à l’analyseur syntaxique Cordial (Laurent, Dominique et al, 2009) ; l’annotation sémantique des arguments des verbes, à partir des catégories sémantiques de la version française de la terminologie médicale Snomed Internationale (Côté, 1996) ; l’acquisition des patrons syntactico-sémantiqueset l’analyse contrastive du fonctionnement des verbes dans les différents corpus. Les patrons syntaxico-sémantiques des verbes acquis au terme de ce processus subissent une évaluation (par trois équipes d’experts en médecine) qui débouche sur la sélection des candidats constituant la nomenclature de la ressource de simplification. Les pss sont ensuite alignés avec leurs correspondants non spécialisés, cet alignement débouche sur le création de la ressource de simplification, qui représente le résultat principal de notre travail de thèse. Une évaluation du rendement du contenu de la ressource a été effectuée avec deux groupes d’évaluateurs : des linguistes et des non-linguistes. Les résultats montrent que la simplification des pss permet de faciliter la compréhension du sens du verbe en emploi spécialisé, surtout lorsque un certains paramètres sont réunis
With the evolution of Web technology, healthcare documentation is becoming increasinglyabundant and accessible to all, especially to patients, who have access to a large amount ofhealth information. Unfortunately, the ease of access to medical information does not guaranteeits correct understanding by the intended audience, in this case non-experts. Our PhD work aimsat creating a resource for the simplification of medical texts, based on a syntactico-semanticanalysis of verbs in four French medical corpora, that are distinguished according to the levelof expertise of their authors and that of the target audiences. The resource created in thepresent thesis contains 230 syntactico-semantic patterns of verbs (called pss), aligned withtheir non-specialized equivalents. The semi-automatic method applied, for the analysis of verbs,in order to achieve our goal is based on four fundamental tasks : the syntactic annotation of thecorpora, carried out thanks to the Cordial parser (Laurent et al., 2009) ; the semantic annotationof verb arguments, based on semantic categories of the French version of a medical terminologyknown as Snomed International (Côté, 1996) ; the acquisition of syntactico-semantic patternsof verbs and the contrastive analysis of the verbs behaviors in the different corpora. Thepss, acquired at the end of this process, undergo an evaluation (by three teams of medicalexperts) which leads to the selection of candidates constituting the nomenclature of our textsimplification resource. These pss are then aligned with their non-specialized equivalents, thisalignment leads to the creation of the simplification resource, which is the main result of ourPhD study. The content of the resource was evaluated by two groups of people : linguists andnon-linguists. The results show that the simplification of pss makes it easier for non-expertsto understand the meaning of verbs used in a specialized way, especially when a certain set ofparameters is collected

Les modèles géologiques numériques 2D et 3D permettent de comprendre l'organisation spatiale des roches du sous-sol. Ils sont également conçus pour réaliser des simulations numériques afin d’étudier ou de prédire le comportement physique du sous-sol. Pour résoudre les équations qui gouvernent les phénomènes physiques, les structures internes des modèles géologiques peuvent être discrétisées spatialement à l’aide de maillages. Cependant, la qualité des maillages peut être considérablement altérée à cause de l’inadéquation entre, d’une part, la géométrie et la connectivité des objets géologiques à représenter et, d’autre part, les contraintes requises sur le nombre, la forme et la taille des éléments des maillages. Dans ce cas, il est souhaitable de modifier un modèle géologique afin de pouvoir générer des maillages de bonne qualité permettant la réalisation de simulations physiques fidèles en un temps raisonnable. Dans cette thèse, j’ai développé des stratégies de réparation et de simplification de modèles géologiques 2D dans le but de faciliter la génération de maillages et la simulation de processus physiques sur ces modèles. Je propose des outils permettant de détecter les éléments des modèles qui ne respectent pas le niveau de détail et les prérequis de validité spécifiés. Je présente une méthode pour réparer et simplifier des coupes géologiques de manière locale, limitant ainsi l’extension des modifications. Cette méthode fait appel à des opérations d’édition de la géométrie et de la connectivité des entités constitutives des modèles géologiques. Deux stratégies sont ainsi explorées : modifications géométriques (élargissements locaux de l'épaisseur des couches) et modifications topologiques (suppressions de petites composantes et fusions locales de couches fines). Ces opérations d’édition produisent un modèle sur lequel il est possible de générer un maillage et de réaliser des simulations numériques plus rapidement. Cependant, la simplification des modèles géologiques conduit inévitablement à la modification des résultats des simulations numériques. Afin de comparer les avantages et les inconvénients des simplifications de modèles sur la réalisation de simulations physiques, je présente trois exemples d'application de cette méthode : (1) la simulation de la propagation d'ondes sismiques sur une coupe au sein du bassin houiller lorrain, (2) l’évaluation des effets de site liés à l'amplification des ondes sismiques dans le bassin de la basse vallée du Var, et (3) la simulation d'écoulements fluides dans un milieu poreux fracturé. Je montre ainsi (1) qu'il est possible d’utiliser les paramètres physiques des simulations, la résolution sismique par exemple, pour contraindre la magnitude des simplifications et limiter leur impact sur les simulations numériques, (2) que ma méthode de simplification de modèles permet de réduire drastiquement le temps de calcul de simulations numériques (jusqu’à un facteur 55 sur une coupe 2D dans le cas de l’étude des effets de site) tout en conservant des réponses physiques équivalentes, et (3) que les résultats de simulations numériques peuvent être modifiés en fonction de la stratégie de simplification employée (en particulier, la modification de la connectivité d’un réseau de fractures peut modifier les écoulements fluides et ainsi surestimer ou sous-estimer la quantité des ressources produites)
Numerical geological models help to understand the spatial organization of the subsurface. They are also designed to perform numerical simulations to study or predict the rocks physical behavior. The internal structures of geological models are commonly discretized using meshes to solve the physical governing equations. The quality of the meshes can be, however, considerably degraded due to the mismatch between, on the one hand, the geometry and the connectivity of the geological objects to be discretized and, on the other hand, the constraints imposed on number, shape and size of the mesh elements. As a consequence, it may be desirable to modify a geological model in order to generate good quality meshes that allow realization of reliable physical simulations in a reasonable amount of time. In this thesis, I developed strategies for repairing and simplifying 2D geological models, with the goal of easing mesh generation and simulation of physical processes on these models. I propose tools to detect model elements that do not meet the specified validity and level of detail requirements. I present a method to repair and simplify geological cross-sections locally, thus limiting the extension of modifications. This method uses operations to edit both the geometry and the connectivity of the geological model features. Two strategies are thus explored: geometric modifications (local enlargements of the layer thickness) and topological modifications (deletions of small components and local fusions of thin layers). These editing operations produce a model on which it is possible to generate a mesh and to realize numerical simulations more efficiently. But the simplifications of geological models inevitably lead to the modification of the numerical simulation results. To compare the advantages and disadvantages of model simplifications on the physical simulations, I present three applications of the method: (1) the simulation of seismic wave propagation on a cross-section within the Lorraine coal basin, (2) the site effects evaluation related to the seismic wave amplifications in the basin of the lower Var river valley, and (3) the simulation of fluid flows in a fractured porous medium. I show that (1) it is possible to use the physical simulation parameters, like the seismic resolution, to constrain the magnitude of the simplifications and to limit their impact on the numerical simulations, (2) my method of model simplification is able to drastically reduce the computation time of numerical simulations (up to a factor of 55 in the site effects case study) while preserving an equivalent physical response, and (3) the results of numerical simulations can be changed depending on the simplification strategy employed (in particular, changing the connectivity of a fracture network can lead to a modification of fluid flow paths and overestimation or underestimation of the quantity of produced resources)

L’obiettivo del presente lavoro è verificare l’applicabilità di materiali innovativi, quali compositi (GFRP - Glass Fibre Reinforced Polymer) e colle strutturali, per la realizzazione di facciate continue ad alte prestazioni meccaniche e termiche e a basso impatto ambientale. Tale obiettivo è stato verificato anche tramite l’applicazione del principio della “Semplificazione tecnologica” che rappresenta il filo conduttore alla base dello studio e delle sperimentazioni svolte dal gruppo di ricerca, coordinato dal Prof. P.Munafò, che ha sviluppato il brevetto “Sistema per la realizzazione di facciate di edifici” (n.102015000087569) di cui il Professore è inventore. Con tale filosofia di approccio è possibile realizzare componenti edilizi altamente prestazionali e semplici nella loro concezione essendo costituiti con un numero limitato di pezzi implicando così un minor consumo di energia nella produzione, assemblaggio, manutenzione e smaltimento del prodotto, classificandolo quindi come eco-sostenibile. In questa tesi viene verificata la fattibilità di un sistema costruttivo per la realizzazione di facciate continue per edifici studiando preventivamente, con test sperimentali e analisi sul ciclo di vita dei componenti, le prestazioni meccaniche dei profili in GFRP e degli adesivi strutturali in condizioni di invecchiamento accelerato (durabilità) e non, e l’interazione del componente edilizio con l’ambiente, dalla produzione alla dismissione finale (LCA - Life Cycle Assessment). I metodi principalmente usati in questo studio sono di tipo sperimentale al fine di testare le proprietà meccaniche dei materiali, in condizioni ambientali e dopo invecchiamento (accelerato in camera climatica ad elevata umidità e temperatura (ISO 6270-2) e sotto esposizione ai raggi UV (ASTM D904–99)). In seguito ai singoli test di invecchiamento precedentemente citati, sono stati condotti ulteriori sperimentazioni riguardanti il trattamento di campioni a condizioni di invecchiamento combinato (camera climatica ed esposizione ai raggi UV - Tcc+Tuv - e viceversa - Tuv+Tcc -). Al fine di validare i risultati ottenuti dalle sperimentazioni effettuate sono stati eseguiti test numerici e analitici. Il risultato più significativo è dato proprio dalla validazione dell’idea brevettuale dimostrando la possibilità di industrializzare componenti (facciate continue) che utilizzano tale materiale composito (pultruso - GFRP), mediante l’accoppiamento a materiali come l’acciaio che possono conferire al componente alte prestazioni meccaniche, soprattutto per quanto riguarda il contenimento delle deformazioni sotto carico. Le soluzioni tecniche studiate inoltre evitano il problema della rottura fragile delle giunzioni bullonate che è uno dei problemi che riguardano le giunzioni di questo tipo su profili in pultruso. La deformabilità e la rottura fragile delle giunzioni bullonate dei profili in pultruso ne hanno limitato l’utilizzo nel settore dell’ingegneria edile per la realizzazione di facciate continue specie di grandi dimensioni. A tal fine l’attività di ricerca è stata prevalentemente incentrata a verificare la possibilità di inserire nei montati in pultruso di tali facciate, una lamina d’acciaio incollata per contenere la deformazione e per migliorare la qualità della giunzione bullonata in modo da evitare rotture di tipo fragile raggiunto il carico di collasso. Le risultanze dei test sperimentali condotti dimostrano le buone performance del sistema ibrido GFRP-acciaio anche in seguito all’esposizione a differenti condizioni di invecchiamento artificiale e verificano la fattibilità di realizzazione di una facciata continua ad alte prestazioni meccaniche e termiche.
The aim of this work is to demonstrate the applicability of innovative materials, such as Glass Fibre Reinforced Polymer (GFRP) industrialized components (profiles), structural adhesives, for the realization of curtain walls with high mechanical and thermal performances and low environmental impact. This objective with the “Technological Simplification” principle is verified. This latter is the guiding principle to the base of the search and experimental tests carried out by the research group. The teamwork coordinator and patent inventor is Prof P.Munafò, with him I developed a “System for the realization of building façade” (n. 102015000087569). The “Technological Simplification” principle allows the building components realization with high performance and easy to assemble, by using a limited number of pieces. All this involves lower energy consumption in the production, assembly, maintenance and disposal phases. For this reason, the construction element can be considered environmentally sustainable. In this thesis, the feasibility of the constructive system for the realization of building façade, through the experimental tests and component life cycle analysis, is verified. The components and materials properties both in laboratory conditions and after different types of ageing conditions (durability) are tested. The interaction between building components and environment, from the production to ultimate disposal (LCA - Life Cycle Assessment) are analysed. The methods used were mostly of the experimental type. The material mechanical properties both in environmental conditions and in different types of ageing conditions were analysed, such as continuous condensation (ISO 6270-2) and UV irradiation (ASTM D904–99). Additional test with combined artificial ageing (climatic chamber and exposure to UV radiation - Tcc+Tuv – and the other way around - Tuv+Tcc) were tested. The numerical and analytical studies were carried out, with the objective to check and validate the results obtained through experimental tests. The main outcome was the validation of the patents basic ideas, which is a key point in the industrialization process of the construction elements (Structural Member). The aim of this work is to demonstrate the feasibility of the use of pultruded Glass Fiber Reinforced Polymers (GFRP) profiles, adhesively joined with other materials (i.e. steel), in the construction sector. The objective is both to reduce the GFRP profiles deformation under loading conditions, and to avoid the brittle fractures that could occur in bolted joints. In the building engineering field, in fact, these issues (deformations and brittle fractures) prevent the use of pultruded materials. In the research activity, the possibility to adhesively join a steel laminate on the pultruded profile mullion for curtain walls was verified. The containment of the deformations and the prevention of brittle fractures in the bolted joint were checked, in order to verify the pultruded curtain wall feasibility, both constructively and for its structural and energy performances. Experimental results, in fact, demonstrated that the use of GFRP profiles, bonded with structural adhesives and combined with steel, is successful on curtain walls, even when they are exposed to adverse environmental conditions. The feasibility of the curtain wall implementation with high performance is verified.

In this thesis, we propose many developments in the context of Structural Similarity. We address both node (local) similarity and graph (global) similarity. Concerning node similarity, we focus on improving the diffusive process leading to compute this similarity (e.g. Commute Times) by means of modifying or rewiring the structure of the graph (Graph Densification), although some advances in Laplacian-based ranking are also included in this document. Graph Densification is a particular case of what we call graph rewiring, i.e. a novel field (similar to image processing) where input graphs are rewired to be better conditioned for the subsequent pattern recognition tasks (e.g. clustering). In the thesis, we contribute with an scalable an effective method driven by Dirichlet processes. We propose both a completely unsupervised and a semi-supervised approach for Dirichlet densification. We also contribute with new random walkers (Return Random Walks) that are useful structural filters as well as asymmetry detectors in directed brain networks used to make early predictions of Alzheimer's disease (AD). Graph similarity is addressed by means of designing structural information channels as a means of measuring the Mutual Information between graphs. To this end, we first embed the graphs by means of Commute Times. Commute times embeddings have good properties for Delaunay triangulations (the typical representation for Graph Matching in computer vision). This means that these embeddings can act as encoders in the channel as well as decoders (since they are invertible). Consequently, structural noise can be modelled by the deformation introduced in one of the manifolds to fit the other one. This methodology leads to a very high discriminative similarity measure, since the Mutual Information is measured on the manifolds (vectorial domain) through copulas and bypass entropy estimators. This is consistent with the methodology of decoupling the measurement of graph similarity in two steps: a) linearizing the Quadratic Assignment Problem (QAP) by means of the embedding trick, and b) measuring similarity in vector spaces. The QAP problem is also investigated in this thesis. More precisely, we analyze the behaviour of $m$-best Graph Matching methods. These methods usually start by a couple of best solutions and then expand locally the search space by excluding previous clamped variables. The next variable to clamp is usually selected randomly, but we show that this reduces the performance when structural noise arises (outliers). Alternatively, we propose several heuristics for spanning the search space and evaluate all of them, showing that they are usually better than random selection. These heuristics are particularly interesting because they exploit the structure of the affinity matrix. Efficiency is improved as well. Concerning the application domains explored in this thesis we focus on object recognition (graph similarity), clustering (rewiring), compression/decompression of graphs (links with Extremal Graph Theory), 3D shape simplification (sparsification) and early prediction of AD.
Ministerio de Economía, Industria y Competitividad (Referencia TIN2012-32839 BES-2013-064482)

Le but de cette these est de rapprocher deux mondes qui jusqu'a present ne communiquaient pas: celui de la cao et celui du calcul par la methode des elements finis. La conception assistee par ordinateur, cao, traite de tous les problemes lies a la description d'une piece a l'aide de moyens informatiques. La methode des elements finis nous permet de simuler le comportement mecanique, thermique, de cette piece. Pour ce faire, nous avons besoin du maillage de sa geometrie: decomposition en elements triangulaires, tetraedriques, etc. Cette these se presente en deux parties. Dans la premiere partie, en partant d'un objet complexe, nous commencerons par examiner les possibilites de le simplifier automatiquement dans le but de rendre son maillage facile. En effet, certains details n'ayant aucune influence sur les calculs, peuvent donner lieu a un grand nombre d'elements finis supplementaires. Meme pour une geometrie simplifiee, le maillage est une operation delicate a cause de son influence sur les resultats des calculs. Il existe deux types de maillages, le libre qui est entierement automatique et le regle qui est manuel. Le premier n'est pas apprecie par les industriels parce qu'il ne permet pas de maitriser l'erreur engendree par les calculs. Le second necessite un decoupage prealable de la geometrie en zones facilement maillables: les cellules maillables. La possibilite de controle de l'erreur qu'offre le maillage regle le rend beaucoup plus interessant que le maillage libre. Dans la deuxieme partie de cette these, nous partons d'un objet simplifie et nous presenterons une methode pour sa decomposition en cellules maillables. Nous proposerons aux ingenieurs d'analyse des structures un couplage cao-calcul: il permettra d'extraire du modele cao la geometrie d'une piece et de l'utiliser pour son maillage ou la decomposition en cellules maillables. Nous arrivons ainsi plus rapidement a la phase de calcul et d'interpretation des resultats

Une caisse automobile est un ensemble complexe formé de plusieurs éléments qui sont souvent constitués de matériaux différents et assemblés principalement par points soudés, généralement à plus de 80%. Au stade de la conception, plusieurs critères doivent être vérifiés numériquement et confirmés expérimentalement par le prototype de la caisse, dont sa tenue en endurance. Dans le contexte économique actuel, la politique de réduction des dépenses énergétiques ou autres a conduit les constructeurs automobiles à optimiser les performances des véhicules actuels, en particulier en réduisant de façon très conséquente la masse de leur caisse. Des problèmes liés à la tenue structurelle ou à la tenue en fatigue de la caisse sont alors apparus. Afin d'être validé, le prototype de caisse doit avoir une résistance suffisante pour supporter les essais de fatigue. Les tests de validation sur bancs d'essais réalisés en amont sur un prototype sont très coûteux pour l'industriel, en particulier lorsque les tests d'essais en fatigue sur la caisse ne permettent pas de confirmer les zones d'apparition des fissures identifiées par simulations numériques. Le sujet de la thèse se limitera à ce dernier point. Il porte sur l'ensemble des analyses à mettre en oeuvre afin d'étudier la tenue en endurance de caisses automobiles soumises à des profils de mission sévérisés. L'objectif principal est de mettre au point un processus d'analyse en simulation numérique permettant de garantir un bon niveau de prédictivité de tenue en endurance des caisses automobiles. On entend par bon niveau de prédictivité, le fait d'être en mesure de corréler correctement les résultats d'essais associés aux profils de missions sévérisés classiquement utilisés dans les plans de validation de la caisse. Cette thèse a conduit à :_ analyser le comportement mécanique de la caisse et les forces d'excitations appliquées au cours de l'essai de validation,_ établir une nouvelle méthode de réduction d'un chargement pour les calculs en endurance,_ mettre au point une nouvelle modélisation EF des liaisons soudées par points,_ améliorer les modèles de prédiction de durée de vie des PSR. Les études menées ont ainsi permis d'améliorer le niveau de prédiction des calculs en fatigue de la caisse afin :_ d'identifier la majorité des zones réellement critiques sur la caisse,_ d'évaluer de manière fiable de la criticité relative de chacune de ces zones,_ d'estimer de façon pertinente la durée de vie associée à chacune de ces zones
A body-in-white (biw) is a complex structure which consists of several elements that are made of different materials and assembled mainly by spot welds, generally above 80%. At the design stage, several criteria must be verified numerically and experimentally by the car prototype, as the biw durability. In the current economic context, the policy of reducing energy and other costs led automotive companies to optimize the vehicle performances, in particular by reducing very consistently the mass of the biw. As a consequences, some structural design problems appeared. In order to be validated, validation test benches are carried out upstream on a prototype vehicle. They are very costly to the manufacturer, especially when fatigue tests do not confirm the cracks areas identified by numerical simulations. The thesis is focused on numerical biw durability analysis. It covers all the numerical analysis to be implemented to study the biw durability behavior. The main objective is to develop a numerical simulation process to ensure a good level of durability prediction. It means to be able to have a good correlation level between test bench results and numerical fatigue life prediction. This thesis has led to:_ analyze the biw mechanical behavior and the excitation forces applied to the biw during the validation tests,_ establish a new fatigue data editing technique to simplify load signal,_ create a new finite element spot weld model,_ develop a new fatigue life prediction of spot welds. The studies have thus improved the level of biw fatigue life prediction by:_ identifying the majority of critical areas on the full biw,_ reliably assessing the relative criticality of each area,_ accurately estimating the lifetime associated with each of these areas

Dissertação de Mestrado em Administração Pública
As mudanças mundiais nos diferentes níveis durante as últimas décadas requerem de Estados que garantam respostas oportunas e adequadas às mesmas, e por tanto, a Administração pública deve ser mais eficiente e eficaz com o objetivo de adaptar às suas necessidades com a própria capacidade financeira. A doutrina da governança pública estimula e fomenta a interação dos diferentes atores do Estado, com o propósito de encontrar meios eficazes para atingir os objetivos que têm sido construídos socialmente num ambiente de inclusão, interdependência, administração responsável dos fundos públicos e eficiência. Em conformidade com informação oficial do Ministério de Justiça de Portugal, a modernização da Justiça constitui um dos desígnios do XXI Governo Constitucional, entre outras iniciativas, através do Plano de Ação “Justiça mais Próxima”. Este Plano tem por objetivo contribuir para o desenvolvimento de uma Justiça ágil, transparente, humana e mais próxima do Cidadão. O Plano refere uma estratégia potenciada pelo digital e focada nas reais necessidades dos diferentes públicos-alvo e ambiciona ser um motor de alteração do paradigma vigente no setor. Esta investigação resulta de uma descrição e análise de um processo de modernização dentro do setor de Justiça de Portugal à luz da doutrina da Governança pública, sob dois pilares: inovação e eficiência.
The global changes at different levels during the last decades require States to guarantee timely and adequate responses to this changes, and therefore, the Public Administration should be more efficient and effective aiming to adapt their needs with their own financial capacity. The doctrine of Public governance stimulates and encourages the interaction of different actors of the State, with the purpouse of finding effective means to achieve the objectives that have been socially constructed in an environment of inclusion, interdependence, accountability and efficiency. According to official information from the Ministry of Justice of Portugal, the modernization of Justice is one of the aims of the XXI Constitutional Government, among other initiatives, through the Action Plan "Justiça mais Próxima". The objective of this Plan is to contribute to the development of an agile, transparent, human and closer to the Citizen justice. The Plan refers a digital-enhanced strategy focused on the real needs of the different target audiences and aspire to be an engine for changing the current paradigm in the sector. This research stems from a description and analysis of a modernization process within the Justice sector of Portugal in the light of the doctrine of Public governance, under two bases: innovation and efficiency.
N/A

This thesis addresses the problem of the development of Natural Language Processing techniques for the extraction and generalization of compositional and functional relations from specialized written texts and, in particular, from patent claims. One of the most demanding tasks tackled in the thesis is, according to the state of the art, the semantic generalization of linguistic denominations of relations between object components and processes described in the texts. These denominations are usually verbal expressions or nominalizations that are too concrete to be used as standard labels in knowledge representation forms -as, for example, “A leads to B”, and “C provokes D”, where “leads to” and “provokes” both express, in abstract terms, a cause, such that in both cases “A CAUSE B” and “C CAUSE D” would be more appropriate. A semantic generalization of the relations allows us to achieve a higher degree of abstraction of the relationships between objects and processes described in the claims and reduces their number to a limited set that is oriented towards relations as commonly used in the generic field of knowledge representation.
Esta tesis se centra en el del desarrollo de tecnologías del Procesamiento del Lenguage Natural para la extracción y generalización de relaciones encontradas en textos especializados; concretamente en las reivindicaciones de patentes. Una de las tareas más demandadas de nuestro trabajo, desde el punto vista del estado de la cuestión, es la generalización de las denominaciones lingüísticas de las relaciones. Estas denominaciones, usualmente verbos, son demasiado concretas para ser usadas como etiquetas de relaciones en el contexto de la representación del conocimiento; por ejemplo, “A lleva a B”, “B es el resultado de A” están mejor representadas por “A causa B”. La generalización de relaciones permite reducir el n\'umero de relaciones a un conjunto limitado, orientado al tipo de relaciones utilizadas en el campo de la representación del conocimiento.

碩士
大葉大學
資訊管理研究所
87
Literature shows that excessive structures of systems and unacceptably heavy computation often restrain the applications of neuro-fuzzy systems. Thus, a novel structure simplification algorithm , SSA ,is proposed as the preprocess of all neuro-fuzzy systems. Two stages of processes comprise SSA. First, by combining the concepts of correlation matrix and Principal Component Analysis, the number of input variables can be significantly reduced without much loss of information content about training data. In order to minimizing the number of memberships used for fuzzy inference, the technique of sliding-window scanning is then applied to efficiently partition the domain space for each pairs of selected input and output signals. After the operation of SSA, the initial status of a neuro-fuzzy system is automatically and properly established without any human expertise. To illustrate the performance and generality of SSA, some benchmarks were respectively tested in both the well-known Adaptive Neuro-Fuzzy Inference Network (ANFIS) and the hybrid system of SSA and ANFIS by using MATLAB simulator. Those benchmarks include the approximation of a sinc function with two inputs, the prediction of Mackey-Glass chaotic time series with four inputs, and the prediction of Gas furnace time series with ten inputs. Experiments showed that, in problems of both function approximation and prediction, SSA can significantly reduce the nodes used in ANFIS, speed up the learning and inference processes, and largely improve the capabilities of identification and generalization.

碩士
國防管理學院
資源管理研究所
82
Automated generalization is a important and necessary process in computerp-assisted graphy. It is used to process the data of maps on a larger scale, and then produces a smaller one automatically. 　　Linear features are dominant in conventional maps, and own higher percentage in the vectorized digital maps. So the subject of this study is line, such as administrative boundary line, coast line etc.. There are many find of simplification aglorithm, while in order to process in computer, it is more important that data sructure of data simplitied. A movel simplification algorithm and data structure are introduced in this thesis, which improve the building of binary generalization tree made in Douglas algorithm. they process the coordinates effectively for the purpose of archieveing the goal of logical zooming. In accordance with the drawback of perpendicular distance algorithm (P D A), a new algorithm - E P D A is developed. It improve the destorsion of line simplitied in P D A.

碩士
國立暨南國際大學
資訊工程學系
93
For representing much more realistic visual effects, numerous polygons are often used in many applications of virtual reality and 3D computer graphics today. The number of polygons used in the complicated CG scene becomes more and more. Manipulating and rendering these polygons makes the rendering system a heavy burden. The main purpose of Levels of Detail（LOD）is to reduce the computing cost at the rendering stage by switching among the multi-resolution 3D models according to the actual needs of system and environments. However, the appearance of the model will become coarser certainly while the resolution is reduced. Therefore, simplifying the complexity of models and preserving the appearances of objects at the same time is a very important research issue for LOD. In this paper, to solve the problem mentioned above, we propose a perception-oriented LOD system by introducing the concept of 3D skeleton. The 3D skeleton is a simplified description to its original model. It provides important information to keep whole appearance while model is simplified. Hence, skeleton is taken into consideration first before we simplify models. Then model is divided appropriately in accordance with the alignment of the skeleton. The regions and vertices which represent the features of whole model could be found according to the relationship between the model and its 3D skeleton. Afterward the basic shape of the model could be preserved by reserving important feature regions and vertices while model is simplified. Furthermore, the properties about shape recognition are also taken into account to improve the result of simplification.

博士
國立成功大學
造船及船舶機械工程學系
88
Abstract For the free or forced vibration analysis of a uniform beam carrying any number of concentrated elements (such as fixed lumped masses, elastically-mounted lumped masses, linear springs, rotational springs, or spring-damper-mass systems), the analytical-and-numerical-combined method (ANCM) is one of the most simple and accurate approaches in addition to the conventional finite element method (FEM). From the existing literature one finds that the ANCM has been used in the "flexural vibration" analysis of the uniform beams, but use of the ANCM to the "torsional" vibration analysis of the rotating shafts carrying multiple concentrated elements (such as propellers, pulleys, or couplings) is not found yet. Hence, the first objective of this thesis is to investigate the free and forced torsional vibration characteristics of a propulsive shafting system by means of the ANCM. Besides, the literature relating to the dynamic analysis of the "non-uniform" beams without carrying any concentrated elements is rare and the study on the flexural vibration analysis of the "non-uniform" beams carrying any number of concentrated elements is not found yet. Therefore, the second objective of this thesis is to deal with the latter problem with the ANCM. For a multi-bay beam-like lattice girder, one of the main information required by the designers is the natural frequencies and mode shapes for the "global" (or whole) structure. But the dynamic analysis results obtained from the conventional FEM are composed of the natural frequencies and mode shapes for the "global" structure and those for the "local" structural members. These results often trouble the designers very much. Hence, a lot of researchers devoted themselves to the study of the structural-simplification method (SSM) of such kind of lattice girders to save computer time and to exclude the natural frequencies and mode shapes for the "local" structural members during the dynamic analysis. However, for the "three-dimensional" multi-bay beam-like lattice girders, the accuracy of the existing SSM is still not satisfactory. This is one of the reasons why the third objective of this thesis tries to present a lumped-mass model to incorporate with the stiffness matrix derived based on the rigid-plane assumption to improve the accuracy of the existing SSM.

This study investigates the possibilities and implications for the teaching of the solving of polynomial equations. It is historically directed and also focusses on the working procedures in algebra which target the cognitive and affective domains. The teaching implications of the development of representational styles of equations and their solving procedures are noted. Since concepts in algebra can be conceived as processes or objects this leads to cognitive obstacles, for example: a limited view of the equal sign, which result in learning and reasoning problems. The roles of sense-making, visual imagery, mental schemata and networks in promoting meaningful understanding are scrutinised. Questions and problems to solve are formulated to promote the processes associated with the solving of polynomial equations, and the solving procedures used by a group of college students are analysed. A teaching model/method, which targets the cognitive and affective domains, is presented.
Mathematics Education
M.A. (Mathematics Education)