Hydrolysis products of aluminum that exist in aqueous solutions play an important role in controlling the fate and transport of contaminants and are also used for coagulants to purification of wastewater streams. Adsorption of contaminants such as heavy metals and organics are widely recognized, but the molecular level understanding of the mechanism of action has not been clearly defined. In this research we present the crystallization, structural characterization and chemical characterization of three novel Keggin-type aluminum polycations including ((Al(IDA)H2O)2(Al30O8(OH)60(H2O)22)(2,6 NDS)4(SO4)2Cl4(H2O)40) (Al32-IDA),[(Cu(H2O)2(µ2-OH)2)2(Al2(µ4-O)8(Al28(µ2-OH)50(µ3-OH)6(H2O)26(2,6-NDS)9(H2O)52]-(CuAl30) and [(Zn(NTA)H2O)2(Al(NTA)(µ2-OH)2)2(Al30(µ2-OH)54(µ3-OH)6(µ4-O)8(H2O)20(2,6-NDS)5(H2O)64]-(ZnAl32) where IDA = iminodiacetic acid, NTA- Nitrilotriacetic acid, and 2,6 NDS = 2,6 napthalene disulfonate. These compounds are the first ever reported Keggin-type aluminum species that have been functionalized with organics and heavy metal cations. Structural characterization of these compounds was done by means of single crystal X-ray diffraction along with FTIR, TGA, SEM/EDS and PXRD techniques for chemical characterization. This study provides more insight into the coagulation process and can be employed in developing optimized coagulants for enhanced water purification.

Atherosclerosis is accepted to be a chronic inflammatory disease of the arterial vessel wall. Several cellular subsets of the immune system are involved in its initiation and progression, such as monocytes, macrophages, T and B cells. Recent research has demonstrated that dendritic cells (DCs) contribute to atherosclerosis, too. DCs are defined by their ability to sense and phagocyte antigens, to migrate and to prime other immune cells, such as T cells. Although all DCs share these functional characteristics, they are heterogeneous with respect to phenotype and origin. Several markers have been used to describe DCs in different lymphoid and non-lymphoid organs; however, none of them has proven to be unambiguous. The expression of surface molecules is highly variable depending on the state of activation and the surrounding tissue. Furthermore, DCs in the aorta or the atherosclerotic plaque can be derived from designated precursor cells or from monocytes. In addition, DCs share both their marker expression and their functional characteristics with other myeloid cells like monocytes and macrophages. The repertoire of aortic DCs in healthy and atherosclerotic mice has just recently started to be explored, but yet there is no systemic study available, which describes the aortic DC compartment. Because it is conceivable that distinct aortic DC subsets exert dedicated functions, a detailed description of vascular DCs is required. The first part of this thesis characterizes DC subsets in healthy and atherosclerotic mice. It describes a previously unrecognized DC subset and also sheds light on the origin of vascular DCs. In recent years, microRNAs (miRNAs) have been demonstrated to regulate several cellular functions, such as apoptosis, differentiation, development or proliferation. Although several cell types have been characterized extensively with regard to the miRNAs involved in their regulation, only few studies are available that focus on the role of miRNAs in DCs. Because an improved understanding of the regulation of DC functions would allow for new therapeutic options, research on miRNAs in DCs is required. The second part of this thesis focuses on the role of the miRNA cluster miR- 17~92 in DCs by exploring its functions in healthy and atherosclerotic mice. This thesis clearly demonstrates for the first time an anti-inflammatory and atheroprotective role for the miR17-92 cluster. A model for its mechanism is suggested. / Atherosklerose ist eine chronisch-entzündliche Erkrankung der arteriellen Gefäßwand und zahlreiche Zellen des Immunsystems, wie zum Beispiel Monozyten, Makrophagen, T und B Zellen sind an der Entstehung und Entwicklung beteiligt. Aktuelle Forschungsergebnisse haben gezeigt, dass auch dendritische Zellen (DCs) zur Atherosklerose beitragen. DCs sind durch ihre Fähigkeit gekennzeichnet, Antigene zu erkennen, aufzunehmen, zu migrieren und andere Immunzellen, wie zum Beispiel T Zellen, zu aktivieren. Auch wenn alle DCs diese funktionellen Merkmale teilen, so sind sie in Bezug auf ihren Phänotyp oder Ursprung eine eher heterogene Gruppe. Zahlreiche Oberflächenmoleküle wurden in der Vergangenheit genutzt, um DCs in lymphatischen und nicht-lymphatischen Geweben zu beschreiben. Allerdings hat sich keines dieser Moleküle als spezifisch und unverwechselbar erwiesen. Die Expression von Oberflächenmolekülen ist sehr variabel und hängt nicht nur vom Aktivierungszustand der DCs, sondern auch vom umliegenden Gewebe ab. Dazu kommt, dass DCs in der Aorta, beziehungsweise im atherosklerotischen Plaque, von designierten Vorläuferzellen, aber auch von Monozyten abstammen können und DCs das Profil ihrer Oberflächenmoleküle, sowie ihre funktionellen Eigenschaften, mit anderen myeloiden Zellen wie Monozyten und Makrophagen teilen. Neuere Arbeiten haben damit begonnen das Repertoire an DCs in der Aorta von gesunden und atherosklerotischen Mäusen zu untersuchen. Da es naheliegt, dass verschiedene DC Untergruppen ganz bestimmte Funktionen ausüben, wird eine detaillierte Beschreibung vaskulärer DCs in der Forschung benötigt. Weil es hierzu allerdings bislang kaum Studien gibt, untersucht der erste Teil dieser Arbeit zum ersten Mal systematisch die in gesunden und atherosklerotischen Mäusen vorkommenden Gruppen an DCs. Sie beschreibt außerdem eine zuvor nicht beachtete DC-Untergruppe und gibt Aufschluss über den Ursprung vaskulärer DCs. In den letzten Jahren wurde gezeigt, dass microRNAs (mirRNAs) zahlreiche zelluläre Vorgänge wie Apoptose, Differenzierung, Entwicklung und Proliferation regulieren. Obwohl viele Zelltypen in Bezug auf die in ihrer Regulation eingebundenen mirRNAs charakterisiert wurden, gibt es nur wenige Studien, die sich mit der Rolle von mirRNAs in DCs beschäftigen. Der zweite Teil dieser Arbeit konzentriert sich auf die Rolle der miRNA Gruppe miR-17~92 in DCs und untersucht deren Rolle in gesunden und atherosklerotischen Mäusen. Diese Arbeit zeigt erstmals eine deutliche anti-inflammatorische und protektive Rolle dieser miRNA und schlägt ein Modell für die entdeckten Mechanismen vor.

Doctor en Ciencias de la Ingeniería, Mención Ingeniería Química y Biotecnología / Streptomyces leeuwenhoekii C34 es una nueva cepa que fue aislada desde la laguna Chaxa ubicada en el Desierto de Atacama, Chile. Esta cepa produce metabolitos especializados con actividad contra Staph. aureus resistente a meticilina (MRSA): chaxamicinas y chaxalactinas. La secuencia genómica de S. leeuwenhoekii C34 se obtuvo mediante las tecnologías de Illumina Miseq y PACbio RS II SMRT. El genoma se utilizó para identificar clústers de genes biosintéticos (BGCs) que codifican para metabolitos especializados a través de minería de genomas, y para desarrollar un modelo a escala genómica (GSM) para estudiar las rutas de biosíntesis de producción de metabolitos especializados. Se encontraron 34 BGCs en el genoma de S. leeuwenhoekii C34, más un BGC ubicado en el plásmido pSLE2. Se encontró tres BGCs para lazo-péptidos. Específicamente, se identificó el producto del BGC del lazo-péptido 3 en el sobrenadante de S. leeuwenhoekii C34 cultivado en medio TSB/YEME y se expresó exitosamente en el huésped heterólogo S. coelicolor M1152. Se confirmó que este lazo-péptido era el mismo que la chaxapeptina, recientemente descrita para S. leeuwenhoekii C58. Por otra parte, se identificó un BGC de 64 kb (locus 1083651 a 1147687) que codifica para un híbrido trans-AT PKS/NRPS. Es probable que el producto de este BGC sea un compuesto halogenado debido a la presencia de un gen, sle09470, que codifica para una enzima cloradora. Para estudiar este clúster de genes, se desarrollaron diferentes cepas derivadas de S. leeuwenhoekii. También, el BGC se clonó en huéspedes heterólogos: S. coelicolor M1152, M1154 and S. albus. A través de análisis de HPLC MS/MS y comparación de perfiles de metabolitos, se identificó un grupo de compuestos con patrón clorado, sin embargo se descartaron como posibles productos del BGC ya que además de encontrarse en las cepas de S. leeuwenhoekii también se encontraron en muestras de S. coelicolor M1152. Por otra parte, se detecto un metabolito con una señal de m/z 611.53 [M + H]+ solamente en las muestras de S. leeuwenhoekii M1614 ( chaxamycin BGC) y M1619 ( chaxamycin BGC; sle09560). Se requieren msá estudios para confirmar si los metabolitos expresados diferencialmente corresponden a un producto del híbrido transAT-PKS/NRPS BGC. Para construir el GSM de S. leeuwenhoekii C34 se desarrolló una interfaz basada en python, que permite: buscar genes de Streptomyces asociados a reacciones en la base de datos KEGG, realizar BLAST local contra S. leeuwenhoekii C34, comparar los dominios de proteínas, descargar información de los metabolitos, construir el GSM y realizar simulaciones usando COBRApy. Las rutas biosintéticas de chaxamicinas, chaxalactinas, desferrioxaminas, ectoina y el producto del híbrido transAT-PKS/NRPS BGC (híbrido PK-NP) se incluyeron en el modelo. El modelo, iVR1007, consiste de 1722 reacciones, 1463 metabolitos y 1007 genes, y se validó usando información experimental de crecimiento en diferentes fuentes de carbono, nitrógeno y fósforo, mostrando un 83.7 % de precisión. El modelo se usó para encontrar deleción y sobre-expresión de genes no intuitivas que predicen un aumento en la producción de precursores de chaxamicinas, chaxalactinas e híbrido PK-NP. Las modificaciones predichas podrán ser usadas para realizar ingeniería metabólica de S. leeuwenhoekii C34 para incrementar la producción de metabolitos especializados.

Licensed valuers in the State of Victoria, Australia currently appraise rural land using manual techniques. Manual techniques typically involve site visits to the property, liaison with property owners through interview, and require a valuer experienced in agricultural properties to determine a value. The use of manual techniques typically takes longer to determine a property value than for valuations performed using automated techniques, providing appropriate data are available. Manual methods of valuation can be subjective and lead to bias in valuation estimates, especially where valuers have varying levels of experience within a specific regional area. Automation may lend itself to more accurate valuation estimates by providing greater consistency between valuations. Automated techniques presently in use for valuation include artificial neural networks, expert systems, case based reasoning and multiple regression analysis. The latter technique appears mo st widely used for valuation. The research aimed to develop a conceptual rural property valuation model, and to develop and evaluate quantitative models for rural property valuation based on the variables identified in the conceptual model. The conceptual model was developed by examining peer research, Valuation Best Practice Standards, a standard in use throughout Victoria for rating valuations, and rural property valuation texts. Using data that are only available digitally and publicly, the research assessed this conceptualisation using properties from four LGAs in the Wellington and Wimmera Catchment Management Authority (CMAs) areas in Victoria. Cluster analysis was undertaken to assess if the use of sub-markets, that are determined statistically, can lead to models that are more accurate than sub-markets that have been determined using geographically defined areas. The research is divided into two phases; the 'available data phase' and the 'restricted data phase'. The 'available data phase' used publicly available digital data to build quantitative models to estimate the value of rural properties. The 'restricted data phase' used data that became available near the completion of the research. The research examined the effect of using statistically derived sub-markets as opposed to geographically derived ones for property valuation. Cluster analysis was used during both phases of model development and showed that one of the clusters developed in the available data phase was superior in its model prediction compared to the models produced using geographically derived regions. A number of limitations with the digital property data available for Victoria were found. Although GIS analysis can enable more property characteristics to be derived and measured from existing data, it is reliant on having access to suitable digital data. The research also identified limitations with the metadata elements in use in Victoria (ANZMETA DTD version 1). It is hypothesised that to further refine the models and achieve greater levels of price estimation, additional properties would need to be sourced and added to the current property database. It is suggested that additional research needs to address issues associated with sub-market identification. If results of additional modelling indicated significantly different levels of price estimation, then these models could be used with manual techniques to evaluate manually derived valuation estimates.

Partially observable Markov decision processes are interesting because of their ability to model most conceivable real-world learning problems, for example, robot navigation, driving a car, speech recognition, stock trading, and playing games. The downside of this generality is that exact algorithms are computationally intractable. Such computational complexity motivates approximate approaches. One such class of algorithms are the so-called policy-gradient methods from reinforcement learning. They seek to adjust the parameters of an agent in the direction that maximises the long-term average of a reward signal. Policy-gradient methods are attractive as a \emph{scalable} approach for controlling partially observable Markov decision processes (POMDPs). ¶ In the most general case POMDP policies require some form of internal state, or memory, in order to act optimally. Policy-gradient methods have shown promise for problems admitting memory-less policies but have been less successful when memory is required. This thesis develops several improved algorithms for learning policies with memory in an infinite-horizon setting. Directly, when the dynamics of the world are known, and via Monte-Carlo methods otherwise. The algorithms simultaneously learn how to act and what to remember. ¶ Monte-Carlo policy-gradient approaches tend to produce gradient estimates with high variance. Two novel methods for reducing variance are introduced. The first uses high-order filters to replace the eligibility trace of the gradient estimator. The second uses a low-variance value-function method to learn a subset of the parameters and a policy-gradient method to learn the remainder. ¶ The algorithms are applied to large domains including a simulated robot navigation scenario, a multi-agent scenario with 21,000 states, and the complex real-world task of large vocabulary continuous speech recognition. To the best of the author's knowledge, no other policy-gradient algorithms have performed well at such tasks. ¶ The high variance of Monte-Carlo methods requires lengthy simulation and hence a super-computer to train agents within a reasonable time. The ANU ``Bunyip'' Linux cluster was built with such tasks in mind. It was used for several of the experimental results presented here. One chapter of this thesis describes an application written for the Bunyip cluster that won the international Gordon-Bell prize for price/performance in 2001.

Any computation is facilitated by some physical process, and the observable quantities of any physical process can be viewed as a computation. These close ties suggest that the study of what universal computers are capable of may lead to additional insight about the physical universe, and vice versa. In his thesis, we explore three lines of research that are linked to this central theme. The first partition shows how notions of non-computability and undecidability eventually led to evidence of emergence, the concept that even if a ‘theory of everything’ governing all microscopic interactions were discovered, the understanding of macroscopic order is likely to require additional insights. The second partition proposes a physically motivated model of computation that relates quantum complexity, quantum optimal control, and Riemannian geometry. Thus insights in any one of these disciplines could also lead to insights in the others. The remainder of this partition explores a simple application of these relations. The final partition proposes a model of quantum computation that generalizes measurement based computation to continuous variables. We outline its optical implementation, whereby any computation can be performed by single mode measurements on a resource state that can be prepared by passing a collection of squeezed states through a beamsplitter network.

Cette thèse rassemble une série de travaux sur les émissions naturelles observées dans la partie externe de la plasmasphère terrestre par la mission multi-satellites CLUSTER, et plus précisément par l'instrument WHISPER. Parmi les émissions observées, citons les ondes électrostatiques électroniques observées entre deux gyrofréquences électroniques successives. Ces ondes sont couramment appelées (n+½)fce et sont aussi bien observées dans la magnétosphère terrestre que dans celles d'autres planètes magnétisées. Nous proposons une classification de toutes les émissions naturelles observées dans la plasmasphère, basée sur les caractéristiques spectrales déterminées lors d'études d'événements. Nous nous intéresserons ensuite aux les trois types principaux d'ondes rencontrées : émissions aux n(1.1)fce, aux (n+½)fce et aux nfce. Ces dernières, mises en évidence grâce à la bonne résolution de l'instrument WHISPER, n'avaient jamais été observées auparavant. Une étude systématique sur trois années de données (la période 2002 - 2004) a permis de localiser les différentes émissions observées en secteur MLT et en latitude magnétique. Par des études d'événements multi-satellites, et par l'étude statistique, nous avons pu mettre en évidence l'importance de la distance radiale à la plasmapause sur les caractéristiques (intensité, nombre d'harmoniques présentes) des émissions aux (n+½)fce observées. Pour quantifier cette distance à la plasmapause, nous avons défini un indicateur utilisant la valeur de la fréquence plasma, mesurée par l'instrument WHISPER, lors de la traversée de l'équateur magnétique. Nous avons aussi étudié l'influence de l'activité géomagnétique, à travers les variations des indices AE, Kp et Dst, sur les trois types d'ondes étudiés. En examinant les données de l'instrument PEACE de la missions CLUSTER, deux formes différentes de la fonction de distribution, dépendant de l'intensité des ondes aux (n+½)fce, ont été mises en évidence. Enfin, nous nous sommes familiarisés avec l'utilisation d'un code particulaire et avons retrouvé les modes propres du plasma prédits par la théorie linéaire.

Les travaux menés dans cette thèse concernent les problèmes d'ordonnancement sur des plates-formes de calcul dynamiques et hétérogènes et s'appuient sur le modèle de communication "multi-port" pour les communications. Nous avons considéré le problème de l'ordonnancement des tâches indépendantes sur des plates-formes maîtres-esclaves, dans les contextes statique et dynamique. Nous nous sommes également intéressé au problème de la redistribution de fichiers répliqués dans le cadre de l'équilibrage de charge. Enfin, nous avons étudié l'importance des mécanismes de partage de bande passante pour obtenir une meilleure efficacité du système.

<p>Abstract</p><p>Entrepreneurship has become an important issue in the modern society where the business landscape consists of many small and medium sized firms. These firms have all been started by an entrepreneur who has identified and business idea and developed that idea into a new venture. We therefore see the idea identification and idea development process as a central part of the creation of a new firm.</p><p>A concept that has become popular in the discussion about entrepreneurship is the cluster concept and dynamic business settings. These concepts are based on the fact that in some geographical regions more firms are started than in other geographical locations. These firms are often started in the same field or industry which has lead to theories that the environment supports and stimulates entrepreneurship. The studies about clusters have often been conducted with an organizational perspective and studied formal contacts within clusters. Therefore we in this study want to study how cluster environments support entrepreneurs in their idea identification and idea development process with the perspective of the individual entrepreneur.</p><p>The study is based on a theoretical framework built around the cluster theories develop by Michael Porter who is considered to be the father to this concept. In addition to this we also present theories about the importance of untraded interdependencies, the triple helix, spin-offs and entrepreneurs as organizational products.</p><p>The empirical material for this study has been collected through interviews with five entrepreneurs in Sweden’s famous ICT cluster Kista Science City just outside Stockholm. These interviews are later presented as five individual cases which describe the idea identification and development process. The results from these interview shows that the cluster environment can support the entrepreneurs in several ways. The support we have found in this study could be linked to the existence of untraded interdependencies that provided the entrepreneurs with access to information, knowledge and experts through networks. These factors were useful both in the idea identification and the idea development process.</p>

<p>Within this thesis an algorithm for object recognition called Cluster Matching has been developed, implemented and evaluated. The image information is sampled at arbitrary sample points, instead of interest points, and local image features are extracted. These sample points are used as a compact representation of the image data and can quickly be searched for prior known objects. The algorithm is evaluated on a test set of images and the result is surprisingly reliable and time efficient.</p>