Tesi sul tema "Assisted evolution"

The field of evolutionary computation is inspired by the achievements of natural evolution, in which there is no final objective. Yet the pursuit of objectives is ubiquitous in simulated evolution because evolutionary algorithms that can consistently achieve established benchmarks are lauded as successful, thus reinforcing this paradigm. A significant problem is that such objective approaches assume that intermediate stepping stones will increasingly resemble the final objective when in fact they often do not. The consequence is that while solutions may exist, searching for such objectives may not discover them. This problem with objectives is demonstrated through an experiment in this dissertation that compares how images discovered serendipitously during interactive evolution in an online system called Picbreeder cannot be rediscovered when they become the final objective of the very same algorithm that originally evolved them. This negative result demonstrates that pursuing an objective limits evolution by selecting offspring only based on the final objective. Furthermore, even when high fitness is achieved, the experimental results suggest that the resulting solutions are typically brittle, piecewise representations that only perform well by exploiting idiosyncratic features in the target. In response to this problem, the dissertation next highlights the importance of leveraging human insight during search as an alternative to articulating explicit objectives. In particular, a new approach called novelty-assisted interactive evolutionary computation (NA-IEC) combines human intuition with a method called novelty search for the first time to facilitate the serendipitous discovery of agent behaviors. In this approach, the human user directs evolution by selecting what is interesting from the on-screen population of behaviors. However, unlike in typical IEC, the user can then request that the next generation be filled with novel descendants, as opposed to only the direct descendants of typical IEC. The result of such an approach, unconstrained by a priori objectives, is that it traverses key stepping stones that ultimately accumulate meaningful domain knowledge. To establishes this new evolutionary approach based on the serendipitous discovery of key stepping stones during evolution, this dissertation consists of four key contributions: (1) The first contribution establishes the deleterious effects of a priori objectives on evolution. The second (2) introduces the NA-IEC approach as an alternative to traditional objective-based approaches. The third (3) is a proof-of-concept that demonstrates how combining human insight with novelty search finds solutions significantly faster and at lower genomic complexities than fully-automated processes, including pure novelty search, suggesting an important role for human users in the search for solutions. Finally, (4) the NA-IEC approach is applied in a challenge domain wherein leveraging human intuition and domain knowledge accelerates the evolution of solutions for the nontrivial octopus-arm control task. The culmination of these contributions demonstrates the importance of incorporating human insights into simulated evolution as a means to discovering better solutions more rapidly than traditional approaches.
ID: 031001574; System requirements: World Wide Web browser and PDF reader.; Mode of access: World Wide Web.; Adviser: Kenneth O. Stanley.; Title from PDF title page (viewed August 26, 2013).; Thesis (Ph.D.)--University of Central Florida, 2012.; Includes bibliographical references (p. 129-138).
Ph.D.
Doctorate
Electrical Engineering and Computing
Engineering and Computer Science
Computer Engineering

The work presented in this thesis focused on the synthesis of gold nanoparticles, exploring new ways to synthesise them and also using new tools to improve the study and discovery of new nanomaterials. Whilst there has been a special concern in understanding how they are organised and which are the best conditions to achieve specific shapes, there is gap in finding new approaches that can allow fast synthesis of nanoparticles and fast screening of the chemical space and real time observation of the reaction under different reaction conditions. In the first chapter of this thesis we are going to present a new synthesis method to prepare gold nanoparticles-POM hybrids. Also, we will discuss how POMs can influence the aggregation of nanoparticles, depending on the size and charge of the POM. For example, gold nanoparticles will aggregate more easily if they are surrounded by small and less negatively charged POMs. In the next chapters of the thesis, we will aim to demonstrate that an automated system is able to evolve gold nanomaterials, this means that an automated system will use raw materials (simple chemicals, in this specific case HAuCl4, CTAB and NaBH4) to synthesise very simple nanostructures, such as spheres, then reuse those spheres and other chemicals to produce even more complex structures. In chapter 2 we will go through the process of building an automated system, in this case, the system will be designed to synthesise gold nanoparticles. We will start by designing the automated system and testing it, we will see the flaws that those different systems had and how we overcame them by doing some improvements on them, such as more control over the temperature of the reaction, keeping a constant temperature of the reagents, improving the cleaning process, trying different concentrations of the reagents, trying different algorithms and different ways to calculate the fitness factor, etc. until we found a system that was very reliable, which was able to provide reproducible results. In the last chapter of this thesis, we will focus on the results obtained in the different versions of the automated system. First, we will explain our first objective, which was obtaining gold nanorods of very short aspect ratio. We will analyse the results we obtained for that objective, trying different fitness factor calculations and how the different ways to calculate the fitness factor affected the process to obtain the desired product. We will discuss why we changed the fitness factor calculation and how this helped to achieve our objective. Then, we will jump to the next level of the project, which was to synthesise very simple gold nanoparticles from raw chemicals and reuse these simple structures to obtain more complex structures. This demonstrates that we have an automated system able to evolve gold nanomaterials; the first of this kind. In this chapter, we will also talk about the different techniques used to characterise the product, where we used in-line analytical techniques such as UV-Vis or image analysis (which are the ones used to calculate the fitness factor that the algorithm is going to use) and other techniques to fully characterise the final product such as TEM.

Afin de réduire le coût de calcul pour des problèmes d'optimisation coûteuse, cette thèse a été consacrée à la Stratégie d'Evolution avec Adaptation de Matrice de Covariance assistée par modèle de Krigeage (KA-CMA-ES). Plusieurs algorithmes de KA-CMA-ES ont été développés et étudiés. Une application de ces algorithmes KA-CMA-ES développés est réalisée par l'identification des paramètres matériels avec un modèle constitutif d'endommagement élastoplastique. Les résultats expérimentaux démontrent que les algorithmes KA-CMA-ES développés sont plus efficaces que le CMA-ES standard. Ils justifient autant que le KA-CMA-ES couplé avec ARP-EI est le plus performant par rapport aux autres algorithmes étudiés dans ce travail. Les résultats obtenus par l'algorithme ARP-EI dans l'identification des paramètres matériels montrent que le modèle d'endommagement élastoplastique utilisé est suffisant pour décrire le comportement d'endommage plastique et ductile. Ils prouvent également que la KA-CMA-ES proposée améliore l'efficace de la CMA-ES. Par conséquent, le KA-CMA-ES est plus puissant et efficace que CMA-ES pour des problèmes d'optimisation coûteuse
In order to reduce the cost of solving expensive optimization problems, this thesis devoted to Kriging-Assisted Covariance Matrix Adaptation Evolution Strategy (KA-CMA-ES). Several algorithms of KA-CMA-ES were developed and a comprehensive investigation on KA-CMA-ES was performed. Then applications of the developed KA-CMA-ES algorithm were carried out in material parameter identification of an elastic-plastic damage constitutive model. The results of experimental studies demonstrated that the developed KA-CMA-ES algorithms generally are more efficient than the standard CMA-ES and that the KA-CMA-ES using ARP-EI has the best performance among all the investigated KA-CMA-ES algorithms in this work. The results of engineering applications of the algorithm ARP-EI in material parameter identification show that the presented elastic-plastic damage model is adequate to describe the plastic and ductile damage behavior and also prove that the proposed KA-CMA-ES algorithm apparently improve the efficiency of the standard CMA-ES. Therefore, the KA-CMA-ES is more powerful and efficient than CMA-ES for expensive optimization problems

The Bacillus thuringiensis δ-endotoxins (Bt toxins) are widely used insecticidal proteins in engineered crops that provide agricultural, economic, and environmental benefits, constituting a substantial and increasingly large portion of the total global production of various crops, including cotton, corn, maize and soybeans. Bt toxins are exquisitely selective for targeted pests, typically do not affect off-target insects, and are completely orthogonal to human biology. However, the development of insect resistance to Bt toxins endangers their long-term effectiveness. In this thesis, I describe the development of methodology for the systematic directed evolution of novel Bt toxins to selectively affect resistant insects. Using Phage-Assisted Continuous Evolution (PACE), a previously developed general platform for the continuous directed evolution of biomolecules, I developed a highly sensitized selection for novel protein-protein interactions. This system robustly reported on interactions spanning affinities from low micromolar to picomolar. However, attempts at using this system for the directed evolution of novel protein-protein interactions were largely unsuccessful, presumably as a consequence of low mutagenesis efficiency. To increase the utility of the platform, I sought to enhance the mutagenesis levels afforded by PACE, but current in vivo methods suffer from a lack of control, genomic instability, low efficiency, and narrow mutational spectra. I used a systematic, mechanism-driven approach to create a potent, inducible, broad-spectrum, and vector-based mutagenesis system in E. coli that enhances mutation rates by 322,000-fold over basal levels, surpassing the mutational efficiency and spectra of widely used in vivo and in vitro mutagenesis methods. This system enabled the high-frequency, broad-spectrum mutagenesis of chromosomal, episomal, and viral nucleic acids in vivo, and dramatically enhanced the success of PACE experiments, highlighting the importance of mutagenesis efficiency on selection outcome. Using this enhanced mutagenesis approach and the previously described sensitized selection platform, I was able to evolve variants of the commonly used Bt toxin Cry1Ac that bind toxin binding region of a cadherin-like receptor from the insect pest Trichoplusia ni (TnCAD) that is not targeted by wild-type Cry1Ac. The resulting evolved Cry1Ac variants bind TnCAD with high affinity (Kd = 11-41 nM), kill TnCAD-expressing insect cells that are not susceptible to wild-type Cry1Ac, and kill Cry1Ac-resistant T. ni insects up to 335-fold more potently than wild-type Cry1Ac. Our findings establish that the evolution of Bt toxins with novel insect cell receptor affinity can overcome Bt toxin resistance in insects and confer lethality approaching that of the wild-type Bt toxin against non-resistant insects. In conclusion, these finding offer a novel mechanism of overcoming what is quickly becoming among the largest issue overshadowing the continued success of Bt toxins for pest control and management, and establish a platform for the detection and evolution of a wide array of protein-protein interactions.
Chemical Biology

By screening a low number of Pseudomonas fluorescens (PFE) mutants, directed evolution identified a mutant with modestly improved enantioselectivity (from Ewild-type = 12 to EThr230Ile = 19) toward the useful chiral synthon methyl 3-bromo-2-methylpropionate (MBMP). A homology model of HE revealed that the mutation was far away from the enzyme active site and may contribute to increased enzyme flexibility.
In a second approach to improving enantioselectivity of PFE toward MBMP, the homology model was used to select amino acid residues near the stereocenter of the docked tetrahedral intermediate of the substrate. Randomization of these residues yielded a Trp29Leu mutant with E increased to 58, as well as a Phe199Trp mutant with E decreased to 2.

In this work, a resin infusion process used by Flight Dynamics Corp. was investigated for the fabrication of low-cost composite sandwich structures. Trials were performed to establish a complete set of potential defects that could develop during the manufacture of E-glass epoxy vinyl ester sandwich panels. A series of non-destructive evaluation techniques were examined for their potential to identify two representative defects. It was found that thermography and bondline analysis were effective at detecting defects such as dry spots and incomplete resin infiltration. However, due to technical limitations of these methods visual inspection of high quality digital images proved to be the most accurate method for evaluating the specimens. The effect of these defects was also evaluated through a selection of compression tests. These tests showed that the sensitivity of this test to the presence of defects is low. A parametric study of the infusion process was performed to evaluate the potential of defect evolution under controlled conditions. The evolution of defects was most affected by the vacuum pressure level and the method for distributing resin through the thickness of the sandwich panel. As part of this study, the resin flow front was evaluated to determine the effect on the evolution of defects. A series of defect metrics were developed for a variety of resin distribution methods. In general, it was found that defect evolution due to a single variable was limited and that further evaluation of the interaction of process parameters would be useful to optimize the process.

As a class, protein-based therapeutics offer tremendous advantages over traditional small molecule drugs. Due to their sizes and folding energies, proteins are ideal for catalyzing chemical reactions, and can bind tightly and selectively to extended target surfaces. However, due to their large size, virtually all proteins are unable to spontaneously enter cells, and as a result protein therapeutics are restricted to extracellular targets. We developed a platform for delivery of proteins to intracellular target sites by engineering the surface chemistry of a model protein, green fluorescent protein (GFP). We found that 'supercharged' cationic GFP variants (scGFPs) bind to anionic cell surface molecules and initiate endocytosis, resulting in the efficient delivery of translationally fused cargo to intracellular targets. We discovered that scGFPs, and cationic delivery reagents in general, alter endosomal trafficking in a manner proportional to both their charge and their delivery efficiency, suggesting that avoidance of endosomal maturation is a key step in the endosomal escape of delivered protein cargos. We also developed a method for encapsulation of recombinant proteins by cationic lipid delivery reagents using negatively supercharged GFP. Genetic modification technologies have matured rapidly following the discovery of protein classes with programmable DNA-binding specificities. While site-directed genetic knockout technologies are highly effective, targeted integration and repair remain comparatively inefficient. Site-specific recombinases directly catalyze strand exchange and ligation between DNA molecules, offering an approach to efficient genomic integration. However, most site-specific recombinases are not easily reprogrammable. To address this problem, we developed a genetic selection technique based on the Phage-Assisted Continuous Evolution (PACE) system, to enable the rapid evolution of recombinase proteins towards targets of interest. Using Cre recombinase as a model, the PACE system was optimized, validated, and used to evolve Cre variants with higher activity on their native loxP target site, as well as altered specificity towards a human genomic sequence within the hROSA26 locus. Finally, we developed a method for enhancing the specificity of RNA-guided nucleases by restricting activity to sites of obligate dimeric nuclease assembly. We engineered a FokI nuclease fusion to a catalytically inactivated Cas9 protein that mediates efficient modification with significantly reduced off-target activity.

La biodiversité est confrontée à une crise d’extinction sans précédent en raison de pressions anthropiques, telles que le changement climatique. La lenteur des processus évolutifs spontanés, par rapport à la rapidité des changements environnementaux, suscite de grandes inquiétudes quant à la persistance de nombreuses espèces. En réponse à ces menaces, les flux de gènes assistés sont présentés comme la translocation gérée d’individus au sein de leur aire de répartition historique. Ces pratiques ont pour but d’introduire ou d’augmenter la fréquence de génotypes susceptibles de conférer un avantage dans des conditions climatiques nouvelles ou futures, ainsi que d’accroître la variation génétique dans les populations ciblées, dans l’espoir de stimuler la démographie et, en augmentant la variation génétique sur laquelle la sélection peut agir, la capacité à s’adapter à un environnement changeant. Les risques associés aux flux de gènes assistés peuvent inclure l’introduction de pathogènes, la dépression hybride, l’envahissement génétique ou encore l’augmentation de la maladaptation. Il existe un besoin urgent de lignes directrices et d’études des stratégies optimales et des incertitudes associées à ces pratiques. Nous développons des modèles démo-évolutifs pour aider à identifier les paramètres critiques et les sources d’incertitude lors de la mise en œuvre de ces stratégies. Ces modèles peuvent aider à orienter les pratiques de flux de gènes assistés et alimenter les débats sur la pertinence de leur utilisation. Nous montrons tout d’abord, à l’aide d’un modèle structuré en stade, l’importance de la prise en compte du cycle de vie des espèces ciblées, car le choix optimal des individus à transférer chez les espèces à longue durée de vie confrontées à un climat changeant dépend d’un compromis entre une bonne adaptation au début et à la fin de la vie. Nous proposons également une nouvelle méthode pour intégrer l’évolution dans un modèle de projection intégrale (IPM) et montrons comment seule une petite gamme de diversité génétique introduite pour des traits adaptatifs permettrait à une plante rare et menacée, Centaurea corymbosa, d’échapper à l’extinction dans un climat qui se réchauffe. Enfin, nous montrons à l’aide d’un modèle analytique que si des allèles bénéfiques sont introduits entre des populations génétiquement distinctes ou des espèces étroitement apparentées, de petites introductions précoces de matériel génétique non local minimisent l’envahissement du génome de la population locale et maximisent la probabilité de sauvetage de petites populations en déclin
Biodiversity is facing an unprecedented extinction crisis due to anthropogenic pressures, such as climate change. There is growing concern about the slow spontaneous pace of evolutionary processes relative to the high speed of environmental changes. Assisted gene flow has been introduced as the managed translocation of individuals within the historical range of a species. Its purpose is to introduce or increase the frequency of genotypes expected to confer an advantage under new or future climatic conditions, but also to increase genetic variation in targeted pop- ulations, in the hope that it can stimulate demography through heterosis and, by increasing the genetic variation on which selection can act on, the ability to adapt to a changing environment. Risks associated with assisted gene flow can include pathogen introduction, outbreeding depres- sion, genomic swamping, and increased maladaptation. There is, therefore, a strong need for guidelines and information on the optimal strategies and uncertainty associated with these prac- tices. Using both analytical predictions and simulations, we have developed demo-evolutionary models to to help identify critical parameters and sources of uncertainty when implementing these strategies. These models can help guiding assisted gene flow strategies and inform debates on the relevance of their use and optimal implementation. We first show how important taking into account the life cycle of targeted species is, as the optimal choice of individuals to translocate in long-lived species facing a changing climate is subject to a trade-off between being well adapted at the beginning and end of life. We also propose a new method for integrating evo- lution into an integral projection model (IPM) and show how only a small range of introduced genetic diversity for adaptive traits would allow a rare and endangered monocarpic perennial plant, Centaurea corymbosa, to escape extinction in a warming climate. Lastly, we show with an analytical model that if beneficial alleles are introgressed through managed gene flow between genetically distinct populations or closely related species, small and early introductions of non local genetic material minimize swamping of the local population’ genome and maximize the probability of rescue in small declining populations

Cette thèse est consacrée au développement d'un système d'amortissement actif pour la réduction des vibrations du mât d'éoliennes en cas de vent avec rafales et de vent avec turbulence. La présence de vibrations entraîne souvent soit une déflexion ultime au sommet du mât d'éolienne, soit une défaillance due à la fatigue du matériau près du bas du mât d'éolienne. De plus, étant donné la nature aléatoire de l'état du vent, il est indispensable d'examiner ce problème d'un point de vue probabiliste. Dans ce travail, un cadre probabiliste d'analyse de la fatigue est développé et amélioré en utilisant le réseau de neurones résiduels. Un système d'amortissement utilisant un amortisseur actif, le Twin Rotor Damper, est conçu pour l'éolienne référentielle NREL 5MW. La conception est optimisée par un algorithme évolutionniste avec une méthode de réglage automatique des paramètres basée sur l'exploitation et l'exploration
This dissertation is devoted to the development of an active damping system for vibration reduction of wind turbine tower under gusty wind and turbulent wind. The presence of vibrations often leads to either an ultimate deflection on the top of wind tower or a failure due to the material’s fatigue near the bottom of wind tower. Furthermore, given the random nature of wind conditions, it is indispensable to look at this problem from a probabilistic point of view. In this work, a probabilistic framework of fatigue analysis is developed and improved by using a residual neural network. A damping system employing an active damper, Twin Rotor Damper, is designed for NREL 5MW reference wind turbine. The design is optimized by an evolutionary algorithm with automatic parameter tuning method based on exploitation and exploration

Le vieillissement de la population fait naître des problématiques de santé publique telles que l’augmentation du nombre de fractures dues à la fragilisation des os et la nécessité d’adapter les traitements. Aujourd’hui, un certain nombre de stratégies sont adoptées pour prévenir ou ralentir la perte de masse osseuse et pour traiter les fractures. Elles présentent toutes des limitations qui poussent la recherche vers l’identification de nouvelles cibles de traitements. Les ostéocytes représentent 95 % des cellules de l’os et vivent plusieurs dizaines d’années à l’intérieur de la matrice minéralisée. Ils ont une forme caractéristique avec des prolongements partant du corps cellulaire vers les autres ostéocytes, les cellules à la surface de l’os et les vaisseaux sanguins. Pendant longtemps, ils ont été considérés comme passifs parce qu’emmurés au cœur de la matrice. Cependant, le développement d’outils in vitro et in vivo a permis d’identifier leur rôle central dans le maintien de la masse osseuse. En effet, les ostéocytes sont les cellules de l’os les plus sensibles aux variations de sollicitations mécaniques provoquées par l’exercice physique ou l’immobilisation. En réponse, ils envoient des signaux aux ostéoblastes et ostéoclastes pour renforcer la matrice ou l’éliminer. Le vieillissement provoque des changements métaboliques et hormonaux systémiques qui affectent le réseau ostéocytaire. Cependant, comme il est aujourd’hui encore difficile d’étudier les ostéocytes dans leur environnement d’origine, beaucoup reste à explorer. Notamment, leur rapport au réseau vasculaire qui subit également des changements lors du vieillissement ainsi que l’impact de ce dernier sur leur métabolisme énergétique. De plus, ils ont été très peu considérés en tant qu’acteurs de la régénération osseuse pouvant potentiellement avoir un impact sur la qualité de la réparation. Les fractures problématiques, qui ne se réparent pas spontanément sont appelées défauts critiques. Pour les réparer, des solutions sous forme de substituts osseux sont depuis des années en développement. Parmi ceux-ci, les biocéramiques bénéficient d’un intérêt particulier car elles sont capables de libérer des ions Ca2+ et PO43- dans leur environnement. Leur effet sur les ostéocytes a été très peu étudié bien qu’ils régulent le métabolisme du calcium et du phosphate. Pour aborder ces différents aspects, le projet de thèse a d’abord consisté en l’optimisation d’une méthode de microdissection laser pour spécifiquement collecter les ostéocytes dans leur environnement. Ensuite, cette méthode a été utilisée pour analyser leur expression génique dans différents contextes : d'une part la maturation et le vieillissement physiologique, d'autre part la réparation d’un défaut critique avec ou sans biocéramique chez la souris mâle. Lors de la première phase du projet, en supplément des données d’expression génique pour les ostéocytes, l’évolution morphologique des réseaux ostéocytaire et vasculaire a été décrite lors de la maturation et du vieillissement grâce à des techniques d’imagerie en fluorescence. Les changements opposés observés sur la morphologie de l’os étaient accompagnés de changements spécifiques à chaque réseau, sans lien apparent. L’objectif de la deuxième phase (effectuée dans un laboratoire aux Etats-Unis) était d’établir des techniques d’analyse du métabolisme énergétique des ostéocytes en s’intéressant aux acides gras à longue chaîne comme source d’énergie. Cela a mené à l’établissement de protocoles de mesure bioénergétique in vitro et d’imagerie ex vivo. Finalement, l’étude de l’expression génique des ostéocytes dans les phases précoces du processus de réparation a suggéré une contribution de leur part et un effet des biocéramiques via les gènes Il6 et Dmp1. Les outils mis au point et les résultats produits serviront de base pour initier des études visant à mieux comprendre le fonctionnement des ostéocytes dans des contextes encore peu explorés
Populations live longer raising public health concerns related to aging, such as the increase in fracture number due to bone frailty and the necessity to adapt treatments. Nowadays, multiple strategies are followed to prevent or slow down the loss of bone mass, and to treat fractures. They all present limitations forcing researchers to look for new treatment targets. Osteocytes represent 95 % of the cells in bone and live decades embedded inside their mineralized matrix. They have a specific shape with dendrites extending from their body towards other osteocytes, cells at the bone surface, and towards blood vessels. For a long time, they have been considered passive because of their location. However, the development of in vitro and in vivo tools enabled to identify their central role in bone mass maintenance. This is due to the fact that osteocytes are the most mechanosensitive cells in bone, meaning that they react to variations in mechanical loading coming from exercise or disuse. They are able to send signals to osteoblasts and osteoclasts to form and resorb the matrix where it is needed. Aging causes systemic hormonal and metabolic changes affecting the osteocyte network. However, a lot remains to be explored because it is still difficult to study them in their environment. In particular, the nature of their interactions with the vascular network and the changes in energy metabolism with aging need to be investigated. Moreover, very few studies considered osteocytes as having a role in the bone healing process, or an impact on the quality of the repair. Difficult fractures do not repair spontaneously and are called critical. To repair them, bone substitutes have been under development for years. Among them, bioceramics benefit from a specific interest because they are able to release Ca2+ et PO43- in their environment. Their impact on osteocytes has not been well studied, although these cells regulate calcium and phosphate metabolism. To address these different aspects, the first task of the Ph.D. work was to optimize a laser-assisted microdissection protocol to specifically collect osteocytes in their environment. Then, this method was applied to the analysis of osteocyte gene expression during maturation, aging, and during the repair of a critical-size defect in male mice. For the first part of the project, in addition to the osteocyte gene expression analysis, the evolution of the osteocyte and blood vessel network morphologies was described during maturation and aging, with the help of fluorescent imaging techniques. The opposite changes in bone morphology observed during maturation and aging were characterized by distinct, network-specific changes. The second part of the project was elaborated within a lab in the USA, the goal was to establish different techniques to analyze osteocyte energy metabolism using long-chain fatty acids as a fuel source. This led to the optimization and use of in vitro bioenergetics assays and ex vivo imaging. In the last part of the project, the osteocyte gene expression during the early phases of bone repair was analyzed. Among the genes tested, a contribution of osteocytes was identified through the genes Il6 and Dmp1, as well as an impact of the presence of the bioceramics. The different tools and techniques optimized, and the results produced during this PhD project will enable the initiation of new research studies to better understand osteocyte function in contexts still underexplored

Les défaillances des systèmes mécatroniques sont souvent causées par la rupture par fatigue des joints de brasure de ses composants électroniques. Avec la miniaturisation croissante des boîtiers électroniques, la sollicitation des joints de brasure, qui assurent la liaison entre les sorties du composant et la carte, peut devenir problématique. En effet, les joints de brasures peuvent accepter des taux de déformation importants, mais une accumulation des sollicitations répétées provoquent leur vieillissement prématuré pouvant conduire à la rupture de joints brasés (phénomène de fatigue thermomécanique). Ainsi, les études basées sur la simulation par les méthodes des éléments finis sont menées pour étudier numériquement la durée de vie des boîtiers montés sur circuit imprimé (fiabilité de deuxième niveau). Le coût de calcul élevé de la résolution de ce type de problèmes peut rendre la procédure d’optimisation et d’évaluation de la fiabilité irréalisable en raison de la grande consommation de temps de calcul de la simulation multiphysique par la méthode des éléments finis. Cette thèse propose d’une part, une adaptation de la méthode CMA-ES Assisté par le métamodèle de krigeage pour l’optimisation globale d’un problème. La mise en œuvre de cet algorithme proposé a été réalisée dans l’optimisation globale des joints de brasure d’un boîtier de type PQFP. Les résultats des études numériques montrent que l’algorithme KA-CMA-ES proposé est plus efficace et plus performant que l’algorithme CMA-ES standard. D’autre part, une méthodologie générale pour l’analyse de la fiabilité en fatigue est proposée dans ce manuscrit. Elle s’appuie sur la modélisation des incertitudes (chargement, propriétés du matériau, géométrie) et vise à quantifier le niveau de fiabilité du système étudié pour un scénario de défaillance en fatigue. Une méthode basée sur les techniques de métamodélisation est précisément proposée pour remédier à la complexité des systèmes mécatroniques dans la résolution du problème de fiabilité. Le métamodèle est utilisé pour imiter le comportement du modèle mécanique tout en satisfaisant en même temps l’efficacité et la précision de ce dernier. La mise en œuvre de cette méthodologie est appliquée pour l’analyse de fiabilité d’un boîtier de type T-CSP
Mechatronic system failures are often caused by fatigue failure of the solder joints of its electronic devices. With the increasing miniaturization of electronic devices, the stress on solder joints, which provide the connection between the component outputs and the PCB, has become a major challenge. Indeed, solder joints can accept high deformation rates, but an accumulation of repeated stresses causes their premature ageing which can lead to the rupture of solder joints (thermomechanical fatigue phenomenon). Thus, the studies based on finite element simulation methods are performed to numerically investigate the lifetime of PCB-mounted devices (secondlevel reliability). The high computational costs required to solve such problems may make the optimization and reliability assessment procedure impracticable due to the high computation time of multi-physics finite element simulation. This thesis proposes on the one side, an adaptation of the CMA-ES method Assisted by the kriging metamodel for the global optimization of a given problem. The implementation of this proposed algorithm was performed in the global optimization of the solder joints in a PQFP package. The results of the numerical studies show that the proposed KA-CMA-ES algorithm is more efficient and more efficient than the standard CMA-ES algorithm. On the other side, a general methodology for reliability analysis in fatigue is proposed in this manuscript. It is based on the uncertainty modeling (loading, material properties, geometry) and aims to quantify the reliability level of the system studied for a fatigue failure scenario. A method based on metamodelling techniques is precisely proposed to address the complexity of mechatronic systems in solving the reliability problem. The metamodel is used to emulate the mechanical model behaviour while satisfying both the its efficiency and accuracy. The implementation of the proposed methodology is applied for the reliability analysis of a T-CSP package

Muitos problemas de otimização reais apresentam mais de uma função-objetivo. Quando os objetivos são conflitantes, estratégias especializadas são necessárias, como é o caso dos algoritmos evolutivos multiobjetivo (MOEAs, do inglês Multi-objective Optimization Evolutionary Algorithms). Entretanto, se a avaliação das funções-objetivo é custosa (alto custo computacional ou econômico) muitos MOEAs propostos são impraticáveis. Uma alternativa pode ser a utilização de um modelo de aprendizado de máquina que aproxima o cálculo do fitness (surrogate) no algoritmo de otimização. Este trabalho propõe e investiga uma plataforma chamada ELMOEA/D que agrega MOEAs do estado da arte baseados em decomposição de objetivos (MOEA/D) e máquinas de aprendizado extremo (ELMs, do inglês Extreme Learning Machines) como modelos surrogate. A plataforma proposta é testada com diferentes variantes do algoritmo MOEA/D e apresenta bons resultados em problemas benchmark, comparada a um algoritmo da literatura que também utiliza MOEA/D mas modelos surrogates baseados em redes com função de base radial. A plataforma ELMOEA/D também é testada no Problema de Predição de Estrutura de Proteínas (PPEP). Apesar dos resultados alcançados pela proposta não serem tão animadores quanto aqueles obtidos nos benchmarks (quando comparados os algoritmos com e sem surrogates), diversos aspectos da proposta e do problema são explorados. Por fim, a plataforma ELMOEA/D é aplicada a uma formulação alternativa do PPEP com sete objetivos e, com estes resultados, várias direções para trabalhos futuros são apontadas.
Many real optimization problems have more than one objective function. When the objectives are in conflict, there is a need for specialized strategies, as is the case of the Multi-objective Optimization Evolutionary Algorithms (MOEAs). However, if the functions evaluation is expensive (high computational or economical costs) many proposed MOEAs are impractical. An alternative might be the use of a machine learning model to approximate the fitness function (surrogates) in the optimization algorithm. This work proposes and investigates a framework called ELMOEA/D that aggregates state-of-the-art MOEAs based on decomposition of objectives (MOEA/D) and extreme learning machines as surrogate models. The proposed framework is tested with different MOEA/D variants and show good results in benchmark problems, compared to a literature algorithm that also encompasses MOEA/D but uses surrogate models based on radial basis function networks. The ELMOEA/D framework is also applied to the protein structure prediction problem (PSPP). Despite the fact that the results achieved by the proposed approach were not as encouraging as the ones achieved in the benchmarks (when the algorithms with and without surrogates are compared), many aspects of both algorithm and problem are explored. Finally, the ELMOEA/D framework is applied to an alternative formulation of the PSPP and the results lead to various directions for future works.

El present treball de recerca es centra en l'estudi de l'aprenentatge col·laboratiu mediat per ordinador
en entorns d'ensenyament i aprenentatge virtuals i en l'àmbit de la docència universitària. Es descriu
i s'analitza l'ús que fa el professorat d'entorns virtuals d'ensenyament/aprenentatge (E/A) de
caràcter general. Així mateix, s'exploren les possibilitats de noves formes d'ensenyar i aprendre a
través de l'ús d'entorns d'E/A dissenyats específicament per afavorir l'aprenentatge col·laboratiu
mediat per l'ordinador (CSCL).
El punt de partença del treball consisteix, d'una banda, en un anàlisi d'un recull ampli
d'investigacions realitzades sobre el tema que es presenta sistematitzat en una taula. D'altra banda
s'explora l'ús que es fa dels entorns virtuals i/o de suport virtual (WebCT i Sakai) a la universitat de
Lleida (UdL).
La part central del treball consisteix en investigar si les característiques del software utilitzat poden
ser decisives per a la millora de l'aprenentatge. Seguint un disseny quasi-experimental es
seleccionen fòrums de debat de diverses assignatures, pertanyents a entorns virtuals generals (Sakai
i WebCT) i es desenvolupen assignatures on s'empren plataformes específiques dissenyades per
afavorir el CSCL (Knowledge Forum, Synergeia, Fle3). Es comparen les aportacions fetes entre
entorns generals i específics, i entre entorns específics que requereixen catalogar les aportacions
(scaffolds de Fle3 i Synergeia) i els que no ho requereixen.
A més a més, s'analitzen les aportacions que fan els subjectes, segons diferents paràmetres
quantitatius. S'observa que els estudiants i les estudiantes universitaris fan un ús de les bastides
limitat i imprecís. Així mateix, es fa un anàlisi qualitatiu de les aportacions d'acord a un sistema de
categories fonamentat en els processos socials d'aprenentatge col·laboratiu (Stahl, 2000) i els
scaffolds de Knowledge Forum, seguint un procés de fiabilitat en la categorització.
Els resultats mostren que les diferències entre entorns generals i específics són molt poques.
S'ha constatat que en els 60 fòrums analitzats, independentment de l'entorn, la participació i la
motivació disminueixen si els missatges no són valorats pel tutor. Les aportacions són bàsicament
d'opinió i d'explicitació sobre el tema. S'observen pocs missatges on els estudiants entrin en debat
amb idees aportades anteriorment pels altres participants en el fòrum.
Finalment, es proposa estimular la participació, oferir pautes de construcció del discurs, i l'ús guiat
de les bastides, com a recursos per optimitzar l'aprenentatge virtual (o en suport virtual) per a que
sigui més col·laboratiu.
La presente investigación está centrada en el estudio del aprendizaje colaborativo mediado por
ordenador en entornos de enseñanza y aprendizaje virtuales en el ámbito de la docencia
universitaria. Se describe y analiza el uso que hace el profesorado de entornos virtuales de
enseñanza y aprendizaje (E/A) de carácter general. También se exploran las posibilidades de nuevas
formas de enseñar y aprender a través del uso de entornos (E/A) diseñados específicamente para
favorecer el aprendizaje colaborativo mediado por ordenador (CSCL).
El punto de partida del trabajo consiste, por un lado, en un análisis de una amplia selección de
investigaciones realizadas sobre el tema que se presenta sistematizado en una tabla. Por otro lado,
se explora el uso que se hace de los entornos virtuales y/o con apoyo virtual (WebCT y Sakai) en la
universidad de Lleida (UdL).
La parte central del trabajo consiste en investigar si las características del software utilizado pueden
ser decisivas para la mejora del aprendizaje. Siguiendo un diseño cuasi experimental se seleccionan
foros de diferentes asignaturas de entornos virtuales generales (Sakai y WebCT) y de entornos
específicos diseñados para favorecer el CSCL (Knowledge Forum, Synergeia, Fle3). Se comparan
las aportaciones hechas entre entornos generales y específicos, y entre entornos específicos que
requieren catalogar las aportaciones (scaffolds de Fle3 y Synergeia) y los que no lo requieren.
Además, se analizan las aportaciones que hacen los sujetos, según diferentes parámetros
cuantitativos. Se observa que los estudiantes universitarios hacen un uso de los andamiajes limitado
e impreciso. También, se hace un análisis cualitativo de las aportaciones de acuerdo a un sistema de
categorías fundamentado en los procesos sociales de aprendizaje colaborativo (Stahl, 2000) y los
scaffolds de Knowledge Forum, siguiendo un proceso de fiabilidad en la categorización.
Los resultados muestran que las diferencias entre entornos generales y específicos son muy pocas.
Se ha constatado que en los 60 foros analizados, independientemente del entorno, la participación y
la motivación disminuyen si los mensajes no son valorados por el tutor. Las aportaciones son
básicamente de opinión y explicitación sobre el tema. Se observan pocos mensajes donde los
estudiantes entren en debate con ideas aportadas anteriormente por otros participantes en el foro.
Finalmente, se propone estimular la participación, ofrecer pautas de construcción del discurso, y el
uso guiado de los andamiajes (scaffolds), como recursos para optimizar el aprendizaje virtual para
que sea más colaborativo
This research is based on computer supported collaborative learning (CSCL), which takes place in
the university teaching environment. This study describes and analyses the usage of teaching
-learning virtual environments by professors. Moreover, the advantages of new ways of teaching
and learning in CSCL environments are explored.
The starting point of this work, on the one hand, consists of the analysis of a broad collection of
researches on this topic, which is shown in a specific classification table. On the other hand, the
usage of the virtual environments and/ or virtual support (WebCT and Sakai) in the university of
Lleida (UdL) is investigated.
The core subject of the project deals with the need of sorting out whether the software used can
strongly influence the improvement on learning or not. Then, forums from different subjects are
selected, following a quasi-experimental methodology. These forums can be divided into two
groups: those which were found in general virtual environments (Sakai and WebCT) and those in
specific virtual environments (Knowledge Forum, Synergeia, Fle3). The latter were designed in
order to enhance CSCL. Later on messages from one group are compared to messages of the other
group. Besides, messages which belong to virtual specific environments (scaffolds from Fle3 and
Synergeia) and which require being catalogued are also compared to messages from a specific
environment (Knowledge Forum) which doesn't require this.
Furthermore, messages are analysed from different quantitative parameters. It is observed that
university students use scaffolds in an uncertain and limited way. Likewise, messages are analysed
qualitatively following a category system based on social processes of collaborative learning (Stahl,
2000) and the scaffolds from Knowledge Forum. In addition, this analysis has undergone a process
of reliability in the categorising stage.
Results show that there are few differences between general and specific environments. Since
firstly, it has been proved that in all of the 60 analysed forums, whatever the environment was,
participation and motivation decrease if messages aren't assessed by a tutor. Secondly, we have
come to the conclusion that students either express their own opinions, or their actual knowledge on
the topic with or without mentioning the sources of the information they are giving. Nevertheless,
there are few messages in which students comment on their classmates' ideas.
In conclusion, there are different ways to optimize virtual learning so that it becomes more
collaborative, which are the following: the participation, the scripts and the scaffolds.

Spatial frequency reuse is a long-established approach for enhancing the capacity of wireless systems through increased spectral efficiency (bits per second per unit bandwidth). The future 5th generation of wireless systems is expected to incorporate various forms of frequency reuse. This includes multiple-input multiple-output (MIMO) communication enabling frequency reuse across antennas, device-to-device (D2D)-based densification allowing spectrum reuse across direct communication links, and full duplexing utilizing the same spectrum for communication in the transmit and receive directions. This dissertation aims at determining the performance limits of emerging wireless systems underpinned by dense spatial frequency reuse and interference suppression, and to glean key system design insights. Stochastic geometry is the toolbox invoked to conduct the analysis, with network locations modeled as points of a Poisson process. A new framework is developed by introducing a Gaussian fit to the interference and variable degrees of spatial averaging, which enable more meaningful results and compact expressions compared to those of existing analyses. Within this framework, we first consider MIMO spatial multiplexing and interference alignment (IA). The former scheme utilizes all available spatial dimensions for signaling and the latter minimizes interference at the expense of knowing the instantaneous fading states at both transmitters and receivers and of a reduction in spatial signaling dimensions. Despite the intense work on IA and spatial multiplexing, there is limited work aimed at understanding their engineering tradeoff in the context of practically relevant cellular settings such as propagation losses, fading dynamics due to user mobility and imperfect knowledge of the fading states. We have studied this problem in depth through both system- and link-level analyses. Even under perfect knowledge of the fading, IA is seen to be beneficial over spatial multiplexing only in very specific and relatively infrequent network situations, and IA loses all its advantages at vehicular speeds when the fading knowledge is imperfect. Second, we focus on ITLinQ and FlashLinQ, the two principal channelization schemes proposed to date for controlling the interference in D2D networks. An analytical characterization of ITLinQ scheme is provided, opening the door to optimizing its controllable parameters. It is shown that both channelization schemes outperform the unchannelized baseline, with a slight edge for ITLinQ. On the most unfavorable network geometries, ITLinQ yields multiple-fold improvements in spectral efficiency with respect to an unchannelized network. Finally, we introduce full-duplex transceivers in cellular networks and characterize the impact of increased interference on their performance. It is established through analysis and complemented by simulations on a Vodafone LTE field test network that additional user-to-user interference only has a minor impact while base-to-base interference would render full-duplex operation unfeasible in dense microcellular networks without any interference management. In summary, MIMO spatial multiplexing and D2D-based densification are seen to play a vital role in improving wireless system capacity while IA and full duplexing are found to be ineffective.
La reutilització de la freqüència espacial és l’aproximació més acceptada per tal de millorar la capacitat dels sistemes wireless mitjançant l'increment de l'eficiència espectral (bits per segon per unitat d'ample de banda). S'espera que la futura cinquena generació de sistemes wireless incorpori diverses formes de reutilització de freqüència. Això inclou la comunicació multi-input multi-output (MIMO) que permet la reutilització a través d'antenes, densificació dispositiu-a-dispositiu (D2D) que permet reutilitzar l'espectre a través d’enllaços de comunicació directa, així com un full-dúplex emprant el mateix espectre per a la comunicació en la transmissió i recepció de direccions. Aquest treball pretén determinar els límits de rendiment dels sistemes wireless emergents, basats en una densa reutilització de la freqüència espacial i en la supressió d'interferències, així com espigolar coneixement clau per al disseny de sistemes d'aquest tipus. La geometria estocàstica és l'eina que s'aplicarà a l'anàlisi que es durà a terme, modelitzant les localitzacions dins la xarxa com a punts d'un procés de Poisson. La introducció d'un ajust Gaussià a la interferència, conjuntament amb la consideració de nivells variables d'expectació espacial, han permès definir un nou marc matemàtic que fa possible unes expressions més compactes i uns resultats més significatius en comparació amb els anàlisis existents. Dins d'aquest marc, en primer lloc es prenen en consideració la multiplexació MIMO i l'aliniament d'interferència (IA, en les seves sigles en anglès). El primer esquema empra totes les dimensions espacials disponibles per a la senyalització i el segon minimitza la interferència a costa de conèixer els estats de esvaïment instantani dels transmissors i receptors, i d'una reducció en les dimensions de senyalització espacial. Malgrat l'intens treball en l'IA i la multiplexació espacial, s'ha prestat escassa atenció a tractar de comprendre el seu balanç d’enginyeria en el context d'xarxes cel.lulars de rellevància pràctica, com els de propagació de pèrdues, o les dinàmiques de esvaïment degudes a la mobilitat de l'usuari i al coneixement imperfecte dels estats de esvaïment. En aquest treball s'ha estudiat en profunditat aquest problema a través d'anàlisis tant a nivell dels enllaços com del sistema. Fins i tot en condicions de coneixement perfecte del esvaïment, l'IA resulta beneficiós sobre la multiplexació només en situacions de xarxa molt específiques i relativament infreqüents, mentre que perd tots els seus avantatges a velocitats vehiculars quan el coneixement del esvaïment és imperfecte. En segon lloc, el treball es centra en el ITLinQ i el FlashLinQ, els dos principals esquemes de canalització proposats fins al moment per controlar la interferència en xarxes D2D. S'ofereix una caracterització analítica de l'esquema ITLinQ, obrint així la porta a l'optimització dels seus paràmetres controlables. Es mostra que tots dos esquemes de canalització aconsegueixen millors resultats que l'esquema no canalitzat, amb un lleuger avantatge per al ITLinQ. Considerant la geometria de xarxa més desfavorable, el ITLinQ produeix millores múltiples en l'eficiència espectral en comparació amb la xarxa no canalitzada. Finalment, el treball introdueix els transreceptors full-dúplex en xarxes cel.lulars i caracteritza l'impacte de la interferència incrementada en el seu funcionament. A través d'anàlisis i de simulacions complementàries en una xarxa de test de Vodafone LTE, s'estableix que la interferència d'usuari a usuari té un impacte poc significatiu mentre que la interferència de base a base faria inviable l'operació full-dúplex en xarxes microcell.lulars sense gestió de les interferències. En resum, aquest tesis doctoral aporta evidència de que el multiplexat MIMO i la densificació basada en D2D juguen un paper vital en la millora de la capacitat dels sistemes wireless mentre que el IA i el full-dúplex resulten inefectius.

Thesis (Ph. D.)--University of Georgia, 2008.
Directed by Khaled Rasheed. Includes an article published in The proceedings of the Genetic and Evolutionary Computation Conference. Includes bibliographical references (leaves 100-108).

碩士
明志科技大學
化學工程系碩士班
106
This study focused on alkaline water electrolysis to produce hydrogen gas in a varying magnetic field. The experiment used several electrodes, including platinum, FTO and TiO2/FTO. We used AFM and SEM to obtain the cathode surface morphology. It was found that the surface through blade coating were rough while that through spin coating were relatively flat. The electrolysis experiment shows that the impedance increased with applying a fixed magnetic field or a varying magnetic field. The production of hydrogen gas could be enhanced with increasing the frequency of the magnetic field. Furthermore, the change in current density corresponded to the desorption of hydrogen gas from the working electrode.

碩士
國立中央大學
化學工程與材料工程學系
104
Up to date, more and more evidences show that global warming and extreme weather conditions are associated with the CO2 level in atmosphere. Current energy supply and use do not be seen to reduce energy-related green-house-gas emission. Therefore, only changes in production and consumption path can decrease the dependence on fossil fuels. Developing low-carbon energy technologies is critical. Many agencies projected that the advances in hydrogen and fuel cell technologies can support climate change and energy security goals. In particular, hydrogen from renewable energies provides flexibility and sustainability for future low-carbon energy systems. ZnIn2S4 (ZIS) is a visible-light-driven photocatalyst with energy band gap of ~2.4 eV. In our previous work, we developed a microwave-assisted hydrothermal method to generate ZIS particles. The gold-silver nanoshells (GS-NS) with tunable absorption were embedded in ZIS matrix for plasmonic-enhanced photocatalytic hydrogen production. However, the coverage and thickness of ZIS on top of GS-NS were not precisely controlled. In this work, we focused on preparing SiO2@ZIS core-shell nanoparticles with tunable thickness of ZIS shells. Control over the core-shell particles enables us to study structure-property relations. Our experimental findings showed that the surface modification on SiO2 surfaces promoted nucleation of ZIS, leading to a homogeneous coverage. In addition, the thickness of ZIS shell can be easily tuned using microwave-assisted hydrothermal synthesis. Thus, our facile procedure paves the way to generate a more complex structure, GS-NS@dielectric@photocatalyst, for optimization of solar hydrogen production.

碩士
國立中央大學
化學工程與材料工程學系
102
Energy crisis and environmental protection are big challenges of this century. Hydrogen is the most promising replacement for fossil fuels. Therefore, the development of visible-light-driven photocatalysts for water splitting is critical. The purpose of this study is to effectively use photocatalysts to change solar energy into hydrogen energy. We used ZIS (ZnmIn2S3+m) as visible-light-driven photocatalyst. Its water splitting reaction rate increased with the temperature. The absorption of the nanoshells in the coreshell nanoparticles can be adjusted systematically from visible light to IR range making the solar energy into heat and resulted in local thermal effect which can effectively enhance hydrogen evolution to 74%. Because the nanoshell was formed by silver and gold nanoparticles, it had the surface plasmon resonance. Using nanoshells absorbing at 700 nm can transfer enengy to photocatalysts and separated the combination of electrons and holes in photocatalysts making the enhancement of hydrogen evolution to 1.62 times. We also changed the thickness of SiO2 on the nanoshells to observe the interaction between nanoshells and coreshells which might influence the enhancement of hydrogen evolution. When there was no SiO2, electron would transfer between nanoshells and photocatalysts. Thicker thickness of SiO2 might hinder the translation of energy from nanoshells decreasing the enhancement of hydrogen evolution. Making photocatalysts directly into coreshell structures might cause thicker shell or uncovered nanoshells. So we try to mdify the surface of ZIS or mdify the surface of nanoshells and formed coreshell, making uniform distribution of ZIS on nanoshells.

Glow discharge atomic emission spectroscopy is a useful analytical method for the direct analysis of conducting solids thereby obviating the need for time-consuming and hazardous dissolution procedures common with other methods. Detection limits for analytical glow discharges, however, are restricted to relatively high analyte concentrations when compared to other methods. One aspect of glow discharge sampling which proves adverse to analytical performance is through significant analyte loss before excitation by the re-deposition of sputtered species back onto the sample surface. Sputtered atoms are typically ejected from the sample surface with a range of energies that extends to 20 eV, however, this ejection energy is quickly thermalized by collisions with support gas species at pressures typically used for analyses. As a consequence, sputtered atoms are readily re-deposited back onto the sample surface, primarily due to diffusion. For a glow discharge using a planar diode electrode geometry, operating at pressures typically used for analytical purposes, up to 95 % of sputtered species re-deposit on the sample surface. Therefore, any method that retards re-deposition would significantly increase the atomization efficiency of glow discharges and increase the sensitivity of the technique. This work addresses the re-deposition problem using a jet assisted source that relies on a directed support gas flow that not only aids sample transport to the excitation region, but impedes re deposition. The original design has gone through a three-stage evolution: each stage correcting certain imbalances found for the previous model which culminates in an emission source capable of sub-ppm level limits of detection and a precision of less than 0.3 % for certain elements. A comprehensive study for the jet flow effects on the sample surface, using Scanning Electron Microscopy and Energy Dispersive X-ray Fluorescence, and the emitting plasma, using atomic emission and absorption spectroscopies, has been conducted. In addition, excitation processes have been studied in the jet-assisted plasma plume as it issues from the anode housing. Results indicate that the dominant atomic excitation process is through electron excitation. The electrons originate from the collision of two argon atoms which reside in metastable states.

The principal observational contribution of this thesis is an innovative technique, using spatially resolved spectroscopy of highly magnified, gravitationally lensed galaxies, to study their internal structure and kinematics at redshift, z≥1 on sub-galactic scales. The scientific objective is to measure the important, but poorly understood, role of star formation and associated feedback on galaxy evolution. With Gemini GMOS-IFU observations of CFRS03+1077, a lensed galaxy at z=2.94, we determined surface brightness and integration time requirements for spatially resolved kinematics with spectra in the visible region (< 1 micron). For reasonable exposure times the presence of a strong emission line is key, limiting the redshift range to < 1.5 for [OII]3727Å. To tackle the lack of suitable lenses for such studies, we designed a lens search algorithm suitable for multi-color photometric data (with a minimum of 2 colors). Our method uses a two-step approach, first automatically identifying galaxy clusters and groups as high likelihood lensing regions, followed by a dedicated visual search for lensed arcs in pseudo-color images of sub-regions centered on these candidates. By using the color-position clustering of elliptical galaxies in high density environments, the algorithm efficiently isolates candidates with a completeness ≥ 80% for z ≤ 0.6 in Monte-Carlo simulations. Implemented on the CFHT Legacy Survey-Wide fields with available g, r and i photometry, the present yield is 9 lenses (8 new and 1 previously known) from 104 deg². With Gemini GMOS, we confirmed two lensed galaxies with strong [OII]3727Å emission suitable for IFU spectroscopy. The follow-up of both systems, the confirmation of remaining lenses and the application of the lens detector to the remaining 91 square degrees of CFHTLS-Wide are ongoing. In a complementary project, we aim to understand non-linear structure formation within the Λ-CDM framework by characterizing the mass distributions and mass/light ratios of galaxy groups; these structures (where 60% of all galaxies reside), have masses representative of the critical break between cluster and field galaxy mass scales. We use strong gravitational lensing to constrain the mass in the inner core, with velocity dispersion measurements from MOS spectroscopy to map the mass distribution up to the scale of the virial radius. The formalism supporting this approach as well as the tools for analysis (including an efficient B-spline based method for flat fielding and sky subtraction of sky limited spectra) are presented in this thesis. The deflectors of 6 lenses in our catalog resemble galaxy groups suitable for this study. One group, for which the observations are complete, is compatible with either NFW or Hernquist profile; these results will be corroborated with observations of other candidates in forthcoming observing programs. The objective is to amalgamate our results with mass measurements from weak lensing and X-ray observations from our Strong Lensing Legacy Survey (SL2S) collaborators to build a comprehensive picture of the dark matter profile and thus constrain theoretical predictions of mass assembly in galaxy groups.

碩士
國立中央大學
化學工程與材料工程學系
106
Our society is facing growing challenges of energy and environment. In order to find clean and renewable energy resources instead of fossil fuels, many researchers worked hard and tried to find a way to solve the problem. Among these renewable energy resources, the biggest potential to develop is solar energy due to the endless of sun irradiation. To store the solar energy is another problem. One of the solutions is converting solar energy to chemical energy. So we use water-splitting photocatlyst to produce hydrogen under sun irradiation for energy storage. Hydrogen is a clean energy resource because it only produces water and energy after combustion. ZnIn2S4 (ZIS) is a visible-light-driven photocatalyst with the energy band gap of 2.4 eV. We developed a microwave-assisted hydrothermal method to generate ZIS particles. In particular, our studies showed that the gold-silver nanoshells with SiO2 shell(GSNS@SiO2) embedded in ZIS matrix exhibited a unique plasmonic-enhanced photocatalytic hydrogen production. However, the coverage and thickness of ZIS on top of GS-NS were not precisely controlled. If we improve the crystallinity and coverage of ZIS to control shell thickness of ZIS, we can find out the factor which could affect hydrogen production efficiency. Because GSNS@SiO2 is hard to synthesize, our research is focusing on SiO2 core instead of GSNS@SiO2. We found that SiO2@ZIS synthesized in pure water solution has better crystallinity, though synthesized in pure ethanol solution has better coverage. Adding HCl into both water and ethanol solution to lower pH condition would increase crystallinity, better crystallinity related to better hydrogen production efficiency. SiO2@ZIS synthesized in ethanol-water mixed solution would combine the advantage of two solutions, but the properties of the samples were too complex to analyze. Though ZIS shell could get thicker by repeating coating ZIS synthesized in different ratio of ethanol-water solution, not precisely controlling ZIS shell and only outer ZIS shell participating water-splitting reaction limited the use of repeating coating method. Finally we could increase crystallinity and coverage of SiO2@ZIS by adding HCl into pure ethanol solution, ZIS shell thickness could also be controlled by different concentration of ZIS precursor. However, due to the different properties between SiO2 core and GSNS@SiO2, it takes time to study SiO2@ZIS synthesis process applied to GSNS@SiO2. Thus, our facile procedure paves the way to generate a more complex structure GS-NS@dielectric@ photocatalyst, for optimization of solar hydrogen production.

Differentiation of the lithospheric mantle occurred principally through partial melting and extraction of melts. Harzburgites are generally considered as melting residues whereas lherzolites are regarded as pristine mantle weakly affected by melting. However, some orogenic peridotites show evidence of igneous refertilization. In this context, this work re-investigates the nature of the Lherz lherzolites (Pyrenees), type-locality of lherzolites, described as a piece of preserved fertile mantle. Structural and geochemical data show that these lherzolites are not pristine but formed through a refertilization reaction between MORB-like melts and refractory lithosphere. Moreover, the Lherz peridotites were partly used to infer the composition of the primitive upper mantle and these results may have important implications for the nature of the late veneer. Additionally, crystal-preferred orientations of minerals (CPO) highlight a strong feedback between melt percolation and finite strain in the percolated rocks. CPO variations are ruled by a subtle balance between instantaneous melt fraction and local strain rate. This work also investigated the effect of melt percolation on Hf, Nd and Sr isotopes. Isotope systematics in Lherz shows that strong isotopic decoupling may arise in a percolation front. The modelling suggests that decoupled isotopic signatures are generated during porous flow and governed by the melt/matrix elements concentrations, chemical diffusivities or efficiency of isotopic homogenization. Melt-rock interactions can generate “intraplate-like” isotopic signatures. This suggests that a part of isotopic signatures of mantle-related rocks could be generated by diffusional processes associated with melt transport.

Cotton has been a world-wide economic staple in textiles and oil production. There has been a concerted effort for cotton improvement to increase yield and quality to compete with non-natural man-made fibers. Unfortunately, cultivated cotton has limited genetic diversity; therefore finding new marketable traits within cultivated cotton has reached a plateau. To alleviate this problem, traditional breeding programs have been attempting to incorporate practical traits from wild relatives into cultivated lines. This incorporation has presented a new problem: uncultivated cotton hampered by photoperiodism. Traditionally, due to differing floral times, wild and cultivated cotton species were unable to be bred together in many commercial production areas world-wide. This worldwide breeding problem has inhibited new trait incorporation. Before favorable traits from undomesticated cotton could be integrated into cultivated elite lines using marker-assisted selection breeding, the markers associated with photoperiod independence needed to be discovered. In order to increase information about this debilitating trait, we set out to identify informative markers associated with photoperiodism. This study was segmented into four areas. First, we reviewed the history of cotton to highlight current problems in production. Next, we explored cotton’s floral development through a study of floral transition candidate genes. The third area was an in-depth analysis of Phytochrome C (previously linked to photoperiod independence in other crops). In the final area of study, we used Genotype-By-Sequencing (GBS), in a segregating population, was used to determine photoperiod independence associated with single nucleotide polymorphisms (SNPs). In short, this research reported SNP differences in thirty-eight candidate gene homologs within the flowering time network, including photoreceptors, light dependent transcripts, circadian clock regulators, and floral integrators. Also, our research linked other discrete SNP differences, in addition to those contained within candidate genes, to photoperiodicity within cotton. In conclusion, the SNP markers that our study found may be used in future marker assisted selection (MAS) breeding schemas to incorporate desirable traits into elite lines without the introgression of photoperiod sensitivity.

Additive manufacturing of gun propellants is an emerging and promising field which addresses the limitations of conventional manufacturing techniques. Gun propellants are manufactured using wetted extrusion, which uses volatile solvents and dies of limited and constant geometries. On the other hand, additive techniques are faced with the challenges of maintaining the gun propellant’s energetic content as well as its structural integrity during high pressure combustion. The work presented in this thesis demonstrates the feasibility of producing functioning gun propellant grains using vibration-assisted 3D printing, a novel method which has been shown to extrude extremely viscous materials such as clays and propellant pastes. At first, the technique is compared to screw-driven additive methods which have been used in printing gun propellant pastes with slightly lower energetic content. In chapter two, diethylene glycol dinitrate (DEGDN), a highly energetic plasticizer, was investigated due to its potential to replace nitroglycerin in double base propellants with high nitroglycerin content. A novel isoconversional method was applied to analyze its decomposition kinetics. The ignition and lifetime values of diethylene glycol dinitrate were obtained using the new isoconversional method, in order to assess the safety of using the plasticizer in a modified double base propellant. In chapter three, a modified double base propellant (M8D) containing DEGDN was additively manufactured using VAP. The printed strands had little to no porosity, and their density was nearly equal to the theoretical maximum density of the mixture. The strands were burned at high pressures in a Crawford bomb and the burning was visualized using high speed cameras. The burning rate equation as a function of the M8D propellant as a function of pressure was obtained. Overall, this work shows that VAP is capable of printing highly energetic gun propellants with low solvent content, low porosity, with high printing speeds, and which have consistent burning characteristics at high pressures.

Introduction : La dysfonction diaphragmatique est très fréquente chez des patients adultes aux soins intensifs et elle est associée à des évolutions cliniques défavorables. Il n’y a pour l’instant aucune méthode reconnue pour évaluer la fonction du diaphragme chez l’enfant sous ventilation mécanique (VM), et aucune étude décrivant son évolution dans le temps chez cette population. Méthodes : Dans ce travail, nous avons évalué la fonction contractile du diaphragme chez des enfants sous ventilation invasive aux soins intensifs pédiatriques (SIP) et en salle d’opération (SOP). Pour ce faire, la pression au tube endotrachéal (Paw) et l’activité électrique du diaphragme (EAdi) étaient enregistrées simultanément lors de respirations spontanées pendant une brève manœuvre d’occlusion des voies respiratoires. Afin de prendre en compte la commande respiratoire, un ratio d’efficience neuro-mécanique (NME, Paw/EAdi) a d’abord été calculé puis validé par une analyse de variabilité. La fonction du diaphragme a ensuite été comparée entre les deux populations, et son évolution dans le temps au sein du groupe SIP décrite. Résultats : Le NME médian était la mesure de fonction diaphragmatique la plus fiable, avec un coefficient de variation de 23.7% et 21.1% dans les groups SIP et SOP, respectivement. Le NME dans le groupe SIP après 21 heures de VM (1.80 cmH2O/μV, IQR 1.25–2.39) était significativement inférieur à celui du groupe SOP (3.65 cmH2O/μV, IQR 3.45–4.24, p = 0.015). Dans le groupe SIP, le NME n’a pas diminué de façon significative pendant la VM (coefficient de corrélation -0.011, p = 0.133). Conclusion : La fonction diaphragmatique peut être mesurée au chevet des enfants sous VM par de brèves manœuvres d’occlusion. L’efficience du diaphragme était significativement plus élevée dans un groupe sain que dans une cohorte d’enfants critiquement malades, mais elle était stable dans ce groupe avec une commande respiratoire préservée. Dans le futur, les contributions relatives de la maladie critique et de la ventilation mécanique sur la fonction diaphragmatique devront être mieux caractérisées avant de procéder à l’évaluation de potentielles interventions visant à protéger le diaphragme.
Introduction : Diaphragmatic dysfunction is highly prevalent in adult critical care and is associated with worse outcomes. There is at present no recognized method to assess diaphragmatic function in children under mechanical ventilation (MV) and no study describing its evolution over time in this population. Methods : In this work, we have assessed the contractile function of the diaphragm in children under invasive MV in the pediatric intensive care unit (PICU) and in the operating room (OR). This was done by simultaneously recording airway pressure at the endotracheal tube (Paw) and electrical activity of the diaphragm (EAdi) over consecutive spontaneous breaths during brief airway occlusion maneuvers. In order to account for central respiratory drive, a neuro-mechanical efficiency ratio (NME, Paw/EAdi) was first computed and then validated using variability analysis. Diaphragmatic function was then compared between the two populations and its evolution over time in the PICU group described. Results : Median NME was the most reliable measure of diaphragmatic function with a coefficient of variation of 23.7% and 21.1% in the PICU and OR groups, respectively. NME in the PICU group after 21 hours of MV (1.80 cmH2O/μV, IQR 1.25–2.39) was significantly lower than in the OR group (3.65 cmH2O/μV, IQR 3.45–4.24, p = 0.015). In the PICU group, NME did not decrease significantly over time under MV (correlation coefficient -0.011, p = 0.133). Conclusion : Diaphragmatic function can be measured at the bedside of children under MV using brief airway occlusions. Diaphragm efficiency was significantly higher in healthy controls than in a cohort of critically ill children, but it was stable over time under MV in this group with preserved respiratory drive. In the future, the relative contributions of critical illness and mechanical ventilation on diaphragmatic function should be better characterized before evaluating potential interventions aimed at protecting the diaphragm.

Dissertação de mestrado em Industrial Engineering
The work related to the project of this dissertation will consist of an analysis of the technological evolution of the machining processes with CNC (Computer Numerical Control) machines regarding the new concept of Industry 4.0. The concept fits into the current transformation process for the fourth industrial revolution, such as integrated Cyber-Physical Systems (CPS) within the manufacturing processes using the Internet of Things (IoT) in industrial processes. Faced with technological advances, the processes of Industrial Engineering in machining using CNC machines must undergo adaptations, aiming at substantial increases in the operational effectiveness. Thus, an approach will be made to understand how current processes can adapt to the concept under study when analyzing the evolution of the machining tools for CNC machines in the face of new processes. A thorough study will be done to adapt the methodology of Industry 4.0 applying it to the machining processes in CNC Machines. Thereby, a proposal for future applications will be given on the topics studied. The methodology will be based entirely on a documental analysis research strategy. The virtual technology in machining tools is still a subject in development, being one of the main factors to be understood in this dissertation. In this study, it will be possible to analyze the main factors that can influence directly or indirectly the production processes of a factory with CNC machines. It will be explored and studied the types of machining processes for CNC machines and the types of machining tools developed with virtual technology. When we are talking about virtual technology, we are usually addressing the need for software. In CNC machining operations, there is a CAM (Computer Aided Manufacturing) software that performs machining simulations for CNC machines. Thus, a study and analysis of a production system involving a CAM software, a tool with virtual technology and CNC machines will be done to verify how this set can work encompassed and what changes this production model introduces. In the sequence of this study, an idea of a new production system will be proposed, allowing for a better understanding of the possibilities for application of new approaches in the future.
O trabalho relacionado ao projeto desta dissertação de mestrado consistirá de uma análise da evolução tecnológica dos processos de usinagem com as máquinas CNC (Comando Numérico Computacional) em relação ao novo conceito da Indústria 4.0. O conceito se enquadra no atual processo de transformação da quarta revolução industrial, com os Sistemas Ciber-Físicos integrados (CPS) dentro dos processos de fabricação que utilizam a Internet das Coisas (IoT) em processos industriais. Diante dos avanços tecnológicos, os processos de Engenharia Industrial em usinagem utilizando máquinas CNC devem sofrer adaptações, visando um aumento substancial na eficácia operacional. Assim, uma abordagem será feita para entender como os processos atuais podem se adaptar ao conceito em estudo, visando também uma análise da evolução das ferramentas de usinagem para máquinas CNC em face de novos processos. Um estudo minucioso será feito para adaptar a metodologia da Indústria 4.0, aplicando-a aos processos de usinagem em máquinas CNC. Com isso, algumas proposta para aplicações futuras serão apresentadas para os tópicos estudados. A metodologia será totalmente baseada em uma estratégia de investigação documental. A tecnologia virtual em ferramentas de usinagem ainda é um assunto em desenvolvimento, sendo um dos principais fatores a serem compreendidos na realização deste trabalho. Neste estudo, será possível analisar os principais fatores que podem influenciar direta ou indiretamente nos processos de produção de uma fábrica com máquinas CNC. Serão explorados e estudados os tipos de processos de usinagem para máquinas CNC e os tipos de ferramentas de usinagem desenvolvidas com tecnologia virtual. Quando abordamos o assunto sobre tecnologia virtual, geralmente estamos a lidar com a necessidade de um software. Nas operações de usinagem CNC, existe um software CAM (Manufatura Assistida por Computador) que realiza simulações de usinagem para máquinas CNC. Assim, um estudo e análise do sistema de produção envolvendo um software CAM, uma ferramenta com tecnologia virtual e máquinas CNC será feito para verificar como este conjunto pode trabalhar englobado e quais as mudanças para esse modelo de produção. Na sequência dessa análise, será proposta uma ideia de um novo sistema de produção, que permite uma melhor compreensão das possibilidades de aplicação no futuro das novas abordagens.