Au cours de la dernière décennie, l'industrie du service informatique s'est métamorphosée afin de répondre à des besoins client croissants en termes de disponibilité, de performance ou de capacité de stockage des systèmes informatisés. Afin de faire face à ces demandes, les hébergeurs d'infrastructures ont naturellement adopté le partage de systèmes où les charges de travail de différents clients sont exécutées simultanément. Cette technique, mutualisant les ressources à disposition d'un système entre les utilisateurs, permet aux hébergeurs de réduire le coût de maintenance de leurs infrastructures, mais pose des problèmes d'interférence de performance et d'équité d'accès aux ressources. Nous désignons par le terme systèmes partagés best-effort les systèmes dont la gestion de ressources est centrée autour d'une maximisation de l'usage des ressources à disposition, tout en garantissant une répartition équitable entre les différents utilisateurs. Dans ce travail, nous soulignons la possibilité pour un utilisateur abusif d'attaquer les ressources d'une plateforme partagée afin de réduire de manière significative la qualité de service fournie aux autres utilisateurs concurrents. Le manque de métriques génériques aux différentes ressources, ainsi que le compromis naturel entre équité et optimisation des performances forment les causes principales des problèmes rencontrés dans ces systèmes. Nous introduisons le temps d'utilisation comme métrique générique de consommation des ressources, métrique s'adaptant aux différentes ressources gérées par les systèmes partagés best-effort. Ceci nous amène à la spécification de couches de contrôles génériques, transparentes et automatisées d'application de politiques d'équité garantissant une utilisation maximisée des ressources régulées. Notre prototype, implémenté au sein du noyau Linux, nous permet d'évaluer l'apport de notre approche pour la régulation des surcharges d'utilisation mémoire. Nous observons une amélioration significative de la performance d'applications typiques des systèmes partagés best-effort en temps de contention mémoire. De plus, notre technique borne l'impact d'applications abusives sur d'autres applications légitimes concurrentes, puisque l'incertitude sur les durées d'exécution est naturellement amoindrie.

由於交換合約為店頭市場交易，故其違約風險的考量為一重要因素。本文依據Wang and Liao（2003）對於權益交換的研究，以及Hübner（2001）對於信用風險的設定，將之結合，在完全市場的假設下，不考慮交易成本以及賦稅影響下，推導出考慮信用風險後的一般化權益交換評價模型，對於各類型的權益交換評價，只需將本文模型假設簡化即可運用。而依據本文推導結果在跨通貨的權益交換模型中，較無跨通貨的權益交換模型多了一個匯率風險調整項，另外在考慮信用風險之後，則會再多出一信用風險調整項。

This paper empirically explores how the introduction of Credit Default Swap (CDS) trading affects firm systematic risk. By treating the introduction as an event study and imploring propensity score matching and difference-in-differences analysis, this research finds that firm exposure to market risk increases after the introduction of CDS instruments, controlling for higher debt levels. These findings change, however, in times of financial crisis when the impact of CDS trading actually reduces systematic risk. These results show that CDS introduction enables a firm to more dramatically change its exposure to systematic risk in comparison to its counterpart to reflect market conditions.

一籃子信用違約交換評價上並不存在公式解，一般是用蒙地卡羅模擬來推估商品價格，然而，因為蒙地卡羅執行速度較慢，往往會需要能夠大規模運行的計算資源以及高成本的硬體，為了減少成本和提高蒙地卡羅的效率就必須從其演算法改進，於是本文利用Chiang et al.(2007)所提出的一籃子信用違約交換演算法來提升一籃子信用違約交換的評價效率，但是該方法採用多元常態分佈假設下的Factor gaussian copula模型進行評價，並不符合市場實際金融市場資料具有不對稱的偏態現象，尤其對未來的環境危機發生的頻率不斷增加，極端事件可能出現的機會也越來越高，基於此問題，本文將Factor t copula、Factor clayton copula、Factor NIG copula以及Modify factor NIG copula與重要性抽樣演算法結合來提昇商品評價的準確度，並且分析各模型與該演算法結合的效果。

Univ., Diss.--Genève, 2007.

Thesis (doctoral)--Universität St. Gallen, 2009.

Η τρέχουσα χρηματοπιστωτική κρίση, έδωσε τη δυνατότητα σε μεγάλο αριθμό ερευνητών να προσπαθησουν να ερμηνεύσουν συγκεκριμένες διαδικασίες που εμφανίζονται σε αυτή την κατάσταση. Ενδιαφέρον αποτελεί η σχέση μεταξύ κρατικών ομολόγων και των CDS. Η ακόλουθη εργασία θα προσπαθήσει μέσα από θεωρητική και εμπειρική ανάλυση να μελετήσει το ασφάλιστρο κινδύνου μεταξύ των δύο μέσων με τη χρήση της θεωρητικών οικονομετρικών μεθόδων. / The current financial crisis, has enabled a large number of researchers trying to interpret specific processes that occur in it. Such interest is the relationship between government bonds and CDS. The following study will attempt to theoretical and empirical study of co-movements of the spread between the two instruments with the use of theoretical econometric methods.

The thesis deals with bubbles in financial markets, causes these bubbles and the ant model. The aim is to reveal the causes of bubbles in financial markets. Researched bubbles are dot-com, mortgages and crypto-currencies. In the first teoretical part are described the financial markets, investment instruments, theory of the business cycle, investment analysis and historical speculative bubbles. In the second practical part are identified the causes examined speculative bubbles. In the third part are summarized the results of the practical part.

De nos jours, de plus en plus d'organisations déploient des infrastructures matérielles telles que des clusters ou des grilles. Elles sont utilisées pour héberger des services internet communs tels que l'email, les réseaux sociaux ou le commerce électronique ou pour exécuter des applications scientifiques telles que les simulations nucléaires ou les prédictions météorologiques. La capacité de traitement et de stockage demandée pour répondre à la charge de travail de ces applications ne peut être fournie que par le biais de ces infrastructures matérielles. Ces infrastructures matérielles embarquent des systèmes d'exploitation, qui peuvent potentiellement coopérer dans le but de gérer au mieux les ressources disponibles. Ces systèmes gèrent alors l'allocation des ressources aux applications en fonction des besoins de ces dernières. Ces systèmes visent à garantir la qualité de service et en même temps à gérer de façon optimale les ressources dans le but de limiter les coûts, notamment l'énergie. La communauté scientifique s'est intéressée à la problématique de la gestion des ressources. De nombreuses approches ont été proposées et des solutions ont été mises en œuvre. En réalisant un état de l'art de ces approches, nous constatons que la plupart d'entre elles s'intéressent à la gestion des nœuds dans l'objectif de répartir les calculs d'une façon adéquate pour exploiter de manière optimale la charge processeur. La gestion globale de la mémoire dans de telles infrastructures n'a pas été suffisamment étudiée. En effet, la mémoire est souvent considérée comme une ressource avec une capacité théoriquement illimitée grâce aux mécanismes de swap, mais ces derniers ont des conséquences importantes sur les performances des applications et le coût de fonctionnement de l'infrastructure. Dans cette thèse, nous étudions la conception et l'implantation d'un service de gestion globale de la mémoire dans une infrastructure matérielle. Ce service de gestion de mémoire doit éviter le gaspillage de mémoire et ne doit pas pénaliser les performances des applications hébergées. Nous proposons un service de swap à distance permettant à une machine, plutôt que swapper sur son disque local, de swapper sur la mémoire distante d'une autre machine ayant de la mémoire disponible. Les pages distantes peuvent être déplacées dynamiquement afin d'équilibrer la charge entre les machines. Ceci permet de mutualiser la mémoire et d'économiser les ressources. Un prototype a été implémenté et évalué. / Nowadays, more and more organizations are deploying large scale infrastructures such as clusters or grids. They are used to host common Internet services such as email, social networks, e-commerce applications or to run scientific applications such as nuclear simulations and weather predictions. Processing power and storage capacities satisfying the workload of these applications can only be provided by such infrastructure. The operating systems deployed on these nodes manage the allocation of application resources and can potentially cooperate in order to manage the available resources according of the application needs. The scientific community is usually interested in the resource management problematic. Many approaches have been proposed and solutions have been implemented. However, we find out that most of them focus on the node management in order to adequately distribute calculations to optimally exploit the CPU load. The global memory management in such infrastructures has not been enough studied. Indeed, memory is often considered as a resource with a theoretically unlimited capacity thanks to the swap capabilities, but swapping has a significant impact on the system performance and the operation cost. In this thesis, we study the design and the implementation of a global memory service management in a large scale infrastructure. This memory management service must avoid wasting memory resources and should not penalize the performance of hosted applications. It is based on remote swap mechanisms. A prototype has been implemented and evaluated.

Dissertação (mestrado)—Universidade de Brasília, Departamento de Economia, 2011. / Submitted by Shayane Marques Zica (marquacizh@uol.com.br) on 2011-09-12T20:34:16Z No. of bitstreams: 1 2011_RodrigoCaldasGonçalves.pdf: 805939 bytes, checksum: 362aab0bea917f3c353b5ea5c9f18b29 (MD5) / Approved for entry into archive by LUCIANA SETUBAL MARQUES DA SILVA(lucianasetubal@bce.unb.br) on 2011-09-20T15:04:47Z (GMT) No. of bitstreams: 1 2011_RodrigoCaldasGonçalves.pdf: 805939 bytes, checksum: 362aab0bea917f3c353b5ea5c9f18b29 (MD5) / Made available in DSpace on 2011-09-20T15:04:47Z (GMT). No. of bitstreams: 1 2011_RodrigoCaldasGonçalves.pdf: 805939 bytes, checksum: 362aab0bea917f3c353b5ea5c9f18b29 (MD5) / Este trabalho aborda o modelo de precificação do CDS de emissões soberanas, proposto por Remolona in ‘A Ratings Based Approach to Measuring Sovereign Risk’ (International Journal of Finance and Economics, vol. 13, issue 1, 26-39) como forma de evidenciar parcelas do risco soberano não amparadas pela perda esperada, calculada de acordo com o rating soberano dos emissores, originando o que a literatura acadêmica chama de Credit Spread Puzzle. Foi avaliado o modelo para CDS com maturidades de 2, 3, 5, 7 e 10 anos, para grupos de 7 e 8 países emergentes, incluindo-se sempre o Brasil, considerando os períodos entre janeiro de 2002 e junho de 2006, e janeiro de 2002 e junho de 2010, utilizando para tal regressões em painel. Complementarmente, foram realizadas regressões lineares individuais pelo método OLS de 12 países, sendo 10 emergentes e 2 da zona do Euro que atualmente enfrentam problemas em relação à gestão de suas dívidas externas. Foi também avaliado o comportamento do indicador de Volatilidade VIX, elaborado pela Chicago Board of Options Exchange, e as implicações que possui na formação do CDS. Com base em dados de expectativas de perdas de todos os países, calculado com base no rating individual divulgado pela agência de classificação de Risco Moody’s, e nos CDS dos diversos países analisados, foi calculada individualmente a parcela de prêmio decorrente de perdas inesperadas, ou prêmio de risco, e feita análise comparativa com o prêmio de risco brasileiro. Os resultados mostraram que o modelo proposto por Remolona sofreu forte influência da crise subprime ocorrida entre 2008 e 2009, o que mudou os parâmetros dos coeficientes da regressão em painel, sem, no entanto, invalidar o modelo. Foi constatado que existem restrições para aplicação do modelo aos CDS individualmente, não sendo confiável sua utilização sem adaptações. Foi constatada a existência de autocorrelação de resíduos, demonstrando que existem fatores que não foram incluídos na modelagem. Em relação à análise de prêmios, constatou-se que o Brasil, se comparado aos demais países avaliados, vem apresentando significativas melhoras na taxa de CDS, além de ter apresentado perdas menores em razão da crise do subprime, principalmente nas maturidades de 2, 3 e 5 anos, indicando uma incompatibilidade entre as perdas esperadas, e consequentemente a classificação de risco atribuída, e a precificação feita pelo mercado, sendo que muitas vezes a última apresentou valor inferior à precificação esperada. _______________________________________________________________________________ ABSTRACT / This paper discusses the model for pricing sovereign CDS emissions proposed by Remolona in 'A Ratings Based Approach to Measuring Sovereign Risk' (International Journal of Finance and Economics, vol. 13, issue 1, 26-39) as a way of showing portions of sovereign risk is not supported by the expected loss calculated in accordance with the sovereign rating of the issuers, resulting in what the academic literature calls the Credit Spread Puzzle. We evaluated the model for CDS with maturities of 2, 3, 5, 7 and 10 years, for groups of 7 and 8 developing countries, always including Brazil, considering the periods between January 2002 and June 2006 and January 2002 and June 2010, using such panel regressions. In addition, individual linear regressions were performed by OLS from 12 countries, 10 emerging and 2 of the Eurozone which currently face problems in relation to the management of foreign debts. It was also rated the behavior of the VIX volatility indicator, developed by the Chicago Board of Options Exchange, and the implications it has on the formation of the CDS. Based on data from expected losses of all countries, calculated on the basis of individual rating issued by rating agency Moody's, and the CDS of the analyzed countries, was calculated individually to share premium arising from unexpected losses, or premium risk, and made comparison with the Brazilian risk premium. The results showed that the model proposed by Remolona was strongly influenced by the subprime crisis that occurred between 2008 and 2009, which changed the parameters of the regression coefficients in the panel, without, however, invalidate the model. It was noted that restrictions apply to individual CDS, its use is not reliable without adaptations. It has been found the existence of autocorrelation of residues, demonstrating that there are factors that were not included in the modeling. On the analysis of premiums, it was found that Brazil, as compared to other countries evaluated, has shown significant improvements in the rate of CDS, and also presented lower losses due to subprime crisis, primarily with maturities of 2, 3 and 5 years, indicating a mismatch between the expected losses, and consequently the risk ratings assigned, and pricing by the market, and often the latter showed a value below the expected pricing.