I characterize optimal strategies and equilibria in two bandit-type games of strategic information acquisition. The latter game admits a missing data problem, which I develop an estimator to address. I consider a third environment with an additional missing data problem, characterize estimation bias both statistically and empirically, and develop a pair of easily implementable unbiased estimators. In the first chapter, I analyze a two-player, two-armed exponential bandit model in continuous time across several monitoring and disclosure environments, in which breakthroughs are fully informative of the state of the world. In particular, I consider an environment where player actions are observed but the results of experimentation are unobserved, and an environment where both actions and experimentation outcomes are unobserved. In the former environment, I construct a Perfect Bayesian Equilibrium that induces the efficient (cooperative) experimentation path. This equilibrium relies on strategies that deter free-riding by moving to an undesirable, asymmetric experimentation path when the safe arm is played. In the latter environment, I provide conditions on starting beliefs for which it is possible to construct a Perfect Bayesian Equilibrium that induces the efficient experimentation path. This equilibrium requires that players privately, separately experiment and share the results of experimentation at a single point in time. I show how this hybrid Markov approach can be adapted to a general discrete-time setting and provide conditions for which strategies measurable with respect to both the Markov partition and a finite automaton are consistent. In the second chapter, I consider the problem of a rideshare company which makes sequences of take it or leave it offers to drivers for individual trips. When drivers are privately and commonly informed of trip quality, rejected offers are informative about trip quality to the rideshare company. As such, the rideshare company faces a strategic information acquisition problem: offers influence both the current-period payoff and the expectation of future payoffs via posterior beliefs. When unobserved quality is Bernoulli, the value function increases in beliefs over quality, and I characterize the value function and optimal offer sequence using dynamic programming. Moreover, I show that a heuristic 𝑛-offer look-ahead sequence converges uniformly to the optimal revenue, and characterize bounds on the revenue gap. For a general bounded distribution of unobserved trip quality, I show that the value function is decreasing after rejections, and define analogous convergent bounds. Among heterogeneous drivers, driver and platform value can be collapsed into a single-dimensional match value parameter that governs the optimal order of offers. I also illustrate a source of potential bias when estimating driver preferences using a data set with repeated offers, and construct a consistent, easily-implementable estimator to address the issue. In the third chapter, I consider a choice environment where an econometrician only observes certain covariates for goods that are chosen. This missing data may reflect literal omissions from a dataset or inherently counterfactual outcomes, such as in the Roy (1951) model. When partially observed covariates vary stochastically across consumers there is selection bias in the distribution of observed covariates: consumers are more likely to pick goods with preferential characteristic draws. I show that imputation methods that use the observed distribution of covariates to replace missing data bias discrete choice parameter estimates toward zero, regardless of whether missing data or all data are imputed. Moreover, this bias rapidly increases with the variance of the covariate distribution. Instead, I propose two full-information maximum likelihood estimation procedures that jointly estimate preferences and the underlying distribution of covariates. While these estimators necessarily involve Monte Carlo simulation when covariates are distributed continuously, I show that it is possible to avoid simulation when covariates take a discrete set of values.

Made available in DSpace on 2015-03-04T18:50:49Z (GMT). No. of bitstreams: 1 tese.pdf: 5590910 bytes, checksum: a89ccd96ade2b696f0e5b9163dc31bf5 (MD5) Previous issue date: 2006-11-13 / Lagrangian stochastic models are a largely used tool in the study of passive substances dispersion inside the Atmospheric Boundary Layer. Its application is related to the trajectory computation of thousands of particles, that numerically simulate the dispersion of suspense substances in the atmosphere. In this study, the basic concepts related to the Lagrangian stochastic modelling are presented and discussed together with its main characteristics and its computational implementation, to the study of particles dispersion in the atmosphere. In a computational experiment, the obtained results are compared with observational data from the TRACT experiment, that took place in Europe in 1992. The input data needed for the dispersion model are extracted from simulations with the numerical weather forecast model RAMS. Dispersion over Rio de Janeiro region is also tested in a second experiment. / Modelos Lagrangianos estocásticos constituem ferramenta muito utilizada no estudo da dispersão de substâncias passivas na Camada Limite Atmosférica. Sua aplicação consiste em calcular a trajetória de milhares de partículas, que simulam numericamente a dispersão de uma substância em suspensão na atmosfera. Nesta tese, são apresentados e discutidos os conceitos básicos relacionados à Modelagem Lagrangiana Estocástica de Partículas, bem como suas principais características e sua implementação computacional, para o estudo da dispersão de partículas na atmosfera. Numa experimentação computacional, comparam-se os resultados obtidos com dados observacionais provenientes do experimento TRACT, realizado na Europa em 1992. Os dados de entrada necessários ao modelo de dispersão são extraídos de simulações do modelo de previsão numérica do tempo RAMS. A dispersão sobre o Estado do Rio de Janeiro é também testada em um segundo experimento.

[pt] Sintetizadores de séries temporais de atenuação por chuva constituem um importante recurso para o projeto, teste e otimização de Técnicas de Mitigação de Desvanecimento. Como dados experimentais de propagação não estão disponíveis para todas as configurações possíveis de enlaces Terra-espaço, a síntese de séries temporais de atenuação por chuva torna-se uma solução interessante, permitindo a reprodução das características dinâmicas de longa duração da atenuação pela chuva. Nesta tese de doutorado, modelos de canal capazes de sintetizar períodos curtos e longos de atenuação por chuva são analisados, testados e validados para os climas tropical e equatorial brasileiros. Aspectos críticos que determinam o correto comportamento destes geradores de séries temporais são analisados e sua faixa de validade é apresentada. Em regiões de clima tropical e equatorial, o desenvolvimento insuficiente da estrutura de telecomunicações terrestre e, por vezes, a necessidade de vencer distâncias continentais, amplificam a necessidade de implementação de sistemas de comunicação por satélite no médio prazo. Tais sistemas apresentam a vantagem de servir grandes áreas com curto tempo de implantação e a possibilidade de atingir mercado considerável. Para as novas aplicações, fortemente focadas em conteúdo multimídia para consumidores corporativos e residenciais, a banda C não é atraente porque o desenvolvimento de novos sistemas com pequenos terminais, como desejado, é incompatível com as grandes antenas necessárias para a proteção dos sistemas existentes nesta banda. Para o fornecimento de serviços multimídia a altas taxas é previsto o uso da banda Ka. Problemas de propagação são severos nestas altas faixas de frequência de forma que figuras de mérito padrão para desempenho e disponibilidade são difíceis de ser obtidas em regiões climáticas tropicais e equatoriais. Por esta razão, metodologias tradicionais de cálculo de enlaces, que levam ao uso de altas margens fixas não são as mais indicadas. Lança-se mão de Técnicas de Mitigação de Desvanecimento (Fade Mitigation Techniques, FMT). No projeto e otimização de FMTs, o conhecimento do comportamento dinâmico do canal de radiopropagação é necessário. Objetivando preencher este requisito, são usadas séries temporais de atenuação por chuva nas simulações de sistema. Porém, séries temporais experimentais não estão disponíveis em nível global em todas as frequências e inclinações de enlace desejadas, para que se projete e teste FMTs. Portanto, a alternativa ao uso de séries experimentais é a geração (sintetização) de séries temporais de atenuação por chuva fazendo uso de características climatológicas bem como de parâmetros geométricos e de propagação relativos ao enlace. Neste trabalho, modelos de canal baseados na abordagem original de Maseng e Bakken são testados com dados de beacon em 11,5 GHz (banda Ku), medidos em quatro localidades Brasileiras, representativas de climas tropical, subtropical e equatorial. Ainda, um modelo misto usando cadeias de Markov e um modelo, já existente, gerador de eventos de atenuação por chuva sob demanda é proposto e analisado. Validação é realizada por meio de variáveis de teste propostas pela ITU-R e também pela análise da fidelidade que características específicas dos eventos sintetizados possuem em relação às características de eventos reais medidos. / [en] Rain attenuation time series synthesizers constitute an important resource to the design, test and optimization of Fade Mitigation Techniques (FMTs). In the absence of experimental propagation data for every possible Earth-Space configuration, the synthesis of rain attenuation time series becomes an interesting solution to allow for the reproduction of the long-term dynamic characteristics of rain attenuation. In this thesis, channel models able to synthesize long- and shortterm rain attenuation periods are discussed, tested and validated for the Brazilian tropical and equatorial climates. Critical issues determining the correct behavior of such time series generators are analyzed and their range of validity is presented.

Lagrangian stochastic models are a largely used tool in the study of passive substances dispersion inside the Atmospheric Boundary Layer. Its application is related to the trajectory computation of thousands of particles, that numerically simulate the dispersion of suspense substances in the atmosphere. In this study, the basic concepts related to the Lagrangian stochastic modelling are presented and discussed together with its main characteristics and its computational implementation, to the study of particles dispersion in the atmosphere. In a computational experiment, the obtained results are compared with observational data from the TRACT experiment, that took place in Europe in 1992. The input data needed for the dispersion model are extracted from simulations with the numerical weather forecast model RAMS. Dispersion over Rio de Janeiro region is also tested in a second experiment. / Modelos Lagrangianos estocásticos constituem ferramenta muito utilizada no estudo da dispersão de substâncias passivas na Camada Limite Atmosférica. Sua aplicação consiste em calcular a trajetória de milhares de partículas, que simulam numericamente a dispersão de uma substância em suspensão na atmosfera. Nesta tese, são apresentados e discutidos os conceitos básicos relacionados à Modelagem Lagrangiana Estocástica de Partículas, bem como suas principais características e sua implementação computacional, para o estudo da dispersão de partículas na atmosfera. Numa experimentação computacional, comparam-se os resultados obtidos com dados observacionais provenientes do experimento TRACT, realizado na Europa em 1992. Os dados de entrada necessários ao modelo de dispersão são extraídos de simulações do modelo de previsão numérica do tempo RAMS. A dispersão sobre o Estado do Rio de Janeiro é também testada em um segundo experimento.

The treatment of human immunodeficiency virus (HIV-1) infection faces the challenge of drug resistance. The high mutation rate of HIV-1 allows it to develop resistance against all available drugs. New mechanisms of intervention that do not succumb to failure through resistance are thus being explored. Mutagens that increase the viral mutation rate are a promising class of drugs. They can drive HIV-1 past a critical mutation rate, called the error threshold, and induce a catastrophic loss of genetic information. The treatment duration for a mutagen to drive HIV-1 beyond this error threshold is not yet estimated. We devise a detailed stochastic simulation of HIV-1 infection to estimate this duration. The simulations predict that the required duration is inversely proportional to the difference between the mutation rate induced by a mutagen and the error threshold. This scaling is robust to changes in simulation parameters. Using this scaling, we estimate the required duration of treatment with mutagens to be many years. Unfortunately, all available drugs, including mutagens, fail to clear the infection because HIV-1 establishes a reservoir of latently infected cells harbouring silent HIV-1 integrated genomes. A new \shock and kill" strategy that aims to activate latent cells and render them susceptible to immune killing or viral cytopathicity and thus to eradicate the HIV-1 latent reservoir has been suggested. Several latency reversal agents (LRAs) have been developed. Individual LRAs fail to show any decline in the HIV-1 latent reservoir in clinical trials. Combinations of LRAs have been tested in a few in-vitro and ex-vivo experiments. It has been found that in combination LRAs act synergistically. Finding the drug concentrations that yield the maximum synergy may be helpful in achieving a sterilizing cure. Here, we develop an intracellular model to estimate these drug concentrations. We choose drugs from two different classes of LRAs and show that our model captures quantitatively recent in-vitro experiments of their activity individually and in combination. With this model, we estimate the concentrations of the drugs required to obtain the maximum synergy. Strong CD8+ T cell responses against viruses have been associated with low levels of viremia. Elite controllers of HIV-1, who are known to have low or undetectable viremia, mount a cross-reactive CD8+ T cell response against the pathogen which controls viral mutation-driven escape from immune activity. These cross-reactive responses are against specific epitopes of HIV-1. Our goal was to examine whether such epitopes could be identified systematically so that a cross-reactive immune response could be induced by using these epitopes as immunogens. Immune recognition of an epitope involves two parts: presentation of the epitope, or peptide, by the major histocompatibility complex (MHC) molecules in the host and high a finity binding of the peptide-MHC complex with a T cell receptor (TCR). Immune escape could occur at either of these steps. Here, we examined the first step. We devise the following procedure to identify peptides that sustain HLA binding despite mutations. First, from the full length HIV-1 (HCV) proteome, we identify viral peptides that bind tightly with MHC molecules using the software NetMHCpan2.8. Next, we pick the peptides and their complementary MHC molecules that yield tight binding and mutate the peptides bit by bit to examine whether binding was compromised. We identify several viral peptide-MHC pairs that display tight binding despite all possible single mutations of the peptides both with HIV-1 and HCV. These peptides present candidates which can be tested for their TCR binding and cross-reactive immunogenic potential.

Submitted by Nelson Danilo Sykora (sykora1@gmail.com) on 2013-07-08T21:29:23Z No. of bitstreams: 1 Tese _ Nelson D Sykora.pdf: 1461672 bytes, checksum: 6346c144802522b574a243adb52ade3c (MD5) / Approved for entry into archive by Marcia Bacha (marcia.bacha@fgv.br) on 2013-08-13T19:44:44Z (GMT) No. of bitstreams: 1 Tese _ Nelson D Sykora.pdf: 1461672 bytes, checksum: 6346c144802522b574a243adb52ade3c (MD5) / Made available in DSpace on 2013-08-13T19:45:09Z (GMT). No. of bitstreams: 1 Tese _ Nelson D Sykora.pdf: 1461672 bytes, checksum: 6346c144802522b574a243adb52ade3c (MD5) Previous issue date: 2013-01-28 / The relationship between spot and future market and commodity price Backwardation have had an emphasis on the literature of economics and finance. The aim of this paper is to present the main causes responsible for the behavior of Backwardation and to identify the properties that characterize the equilibrium in agricultural commodit y prices. Be the existence of risk premium or the convenience yield, the unders tanding of the effects on the replication of future price and on the term structure of commodity prices remains an open issue. On the other hand, t he p remise of perfect p ortfolio replication and the absence of market frictions imply that the understanding of c ommodity price Backwardation comes from the understanding of the stochastic process of the underlying asset itself . The risk - neutral pricing allied with signs of reversion in prices supports pricing models such as Schwartz and Smith (2000), whose calibration and results will be presented for soybeans. / A relação entre preços do mercado spot e do mercado futuro e a evidência de Mercado Invertido (backwardation) na estrutura a termo de commodities têm tido ênfase na literatura de economia e de finanças. O objetivo deste trabalho é apresentar as principais causas responsáveis pelo comportamento de Mercado Invertido e identificar as propriedades que caracterizam o equilíbrio de preços em commodities agrícolas. Seja pela existência de prêmio de risco ou do benefício de conveniência, o entendimento dos efeitos sobre a replicação do preço futuro e sobre a estrutura a termo de preços ainda permanece em aberto. A premissa de perfeita replicação de portfólios e a ausência de fricções de mercado implicam, por outro lado, que o entendimento do comportamento de Mercado Invertido advém da compreensão do processo estocástico do próprio ativo subjacente. O apreçamento neutro ao risco, amparado pelos sinais de reversão de preços, permite a modelagem de preços conforme o proposto em Schwartz e Smith (2000), cuja calibração e os resultados serão apresentados para a soja.

Le comportement humain est modélisé de manière sommaire dans les logiciels de simulation énergétique des bâtiments. Or son impact est considérable et il est à l'origine d'écarts importants entre résultats de simulation et mesures in situ. Les occupants influencent les consommations d'énergie des bâtiments par leur présence et leurs activités, les ouvertures/fermetures de fenêtres, la gestion des dispositifs d'occultation, l'utilisation de l'éclairage artificiel et des appareils électriques, la gestion des consignes de chauffage et les puisages d'eau chaude sanitaire. La thèse propose une modélisation de l'occupation incluant l'ensemble de ces aspects suivant une approche stochastique statistique, pour les bâtiments résidentiels et de bureaux. La construction des modèles fait appel à un grand nombre de données issues de campagnes de mesures, d'enquêtes sociologiques et de la littérature scientifique. Le modèle d'occupation proposé est couplé à l'outil de simulation thermique dynamique Pléiades+COMFIE. En propageant les incertitudes des facteurs du modèle d'occupation et du modèle thermique (enveloppe, climat, systèmes), un intervalle de confiance des résultats de simulation peut être estimé, ouvrant ainsi la voie à un processus de garantie de performance énergétique. / Human behaviour is modelled in a simplistic manner in building energy simulation programs. However, it has a considerable impact and is identified as a major explanatory factor of the discrepancy between simulation results and in situ measurements. Occupants influence buildings energy consumption through their presence and activities, the opening/closing of windows, the actions on blinds, the use of artificial lighting and electrical appliances, the choices of temperature setpoints, and the water consumptions. The thesis proposes a model of occupants' behaviour including all these aspects, according to a stochastic approach, for residential and office buildings. Models' development is based on numerous data from measurements campaigns, sociological surveys and from the scientific literature. The proposed model for occupants' behaviour is coupled to the simulation tool Pléiades+COMFIE. By propagating the uncertainties of factors from the occupants' behaviour model and the thermal model (envelope, climate, systems), the simulation results confidence interval can be estimated, opening the way to an energy performance guarantee process.

Self-organizing patterns arise in a variety of ways in nature, the complex patterning observed on animal coats is such an example. It is already known that the mechanisms responsible for pattern formation starts at the developmental stage of an embryo. However, the actual process determining cell fate has been, and still is, unknown. The mathematical interest for pattern formation emerged from the theories formulated by the mathematician and computer scientist Alan Turing in 1952. He attempted to explain the mechanisms behind morphogenesis and how the process of spatial cell differentiation from homogeneous cells lead to organisms with different complexities and shapes. Turing formulated a mathematical theory and proposed a reaction-diffusion system where morphogens, a postulated chemically active substance, moderated the whole mechanism. He concluded that this process was stable as long as diffusion was neglected; otherwise this would lead to a diffusion-driven instability, which is the fundamental part of pattern formation. The mathematical theory describing this process consists of solving partial differential equations and Turing considered deterministic reaction-diffusion systems.   This thesis will start with introducing the reader to the problem and then gradually build up the mathematical theory needed to get an understanding of the stochastic reaction-diffusion systems that is the focus of the thesis. This study will to a large extent simulate stochastic systems using numerical computations and in order to be computationally feasible a compartment-based model will be used. Noise is an inherent part of such systems, so the study will also discuss the effects of noise and morphogen kinetics on different geometries with boundaries of different complexities from one-dimensional cases up to three-dimensions.

Le travail présenté dans ce manuscrit traite des mécanismes d'émergence des maladies infectieuses. Une première partie du texte est consacrée à la réflexion menée sur ces processus, et à leurs conséquences pour la gestion des maladies. Dans une deuxième partie nous exposons les travaux de modélisation réalisés ; l’objectif est de tester si des changements dans l'expression d'une infection peuvent être expliqués par la dynamique de circulation de l'agent infectieux dans la population hôte, à travers la protection de groupe. Nous montrons que l'apparition de la maladie de Carré chez les lions du Serengeti peut être due à une perte d’immunité de groupe suite à l’extinction du virus. Nous étudions ensuite les conséquences sur le patron épidémique d'une protection partiellement croisée entre souches. Enfin, nous intégrons à cette problématique la structure spatiale de la population hôte, pour expliquer des observations faites sur les systèmes lagomorphes/lagovirus en France. / The study presented here aims to suggest and to test several mechanisms to explain how infectious diseases impact could increase, in relation to microparasite circulation and loss of herd protection. We first introduce the main characteristics of host/microparasite interaction and display some knowledge about emerging infectious diseases, and their consequences for diseases control. Stochastic modelling is then used to study epidemic patterns under several hypotheses and for different host/parasite systems: i) we explain canine distemper emergence in Serengeti lions through a break in virus circulation and the resulting loss of herd immunity; ii) we study the consequences of partial crossprotection induced in hosts by different strains in a small homogeneous host population; and iii) we extend this study to spatially structured host populations to explain data observed in lagomorphs/lagoviruses in France.

Le but de cette thèse est de développer des outils statistiques permettant d'initialiser et de calibrer les modèles de microsimulation dynamique stochastique, en partant de l'exemple du modèle SimVillages (développé dans le cadre du projet Européen PRIMA). Ce modèle couple des dynamiques démographiques et économiques appliquées à une population de municipalités rurales. Chaque individu de la population, représenté explicitement dans un ménage au sein d'une commune, travaille éventuellement dans une autre, et possède sa propre trajectoire de vie. Ainsi, le modèle inclut-il des dynamiques de choix de vie, d'étude, de carrière, d'union, de naissance, de divorce, de migration et de décès. Nous avons développé, implémenté et testé les modèles et méthodes suivants: * un modèle permettant de générer une population synthétique à partir de données agrégées, où chaque individu est membre d'un ménage, vit dans une commune et possède un statut au regard de l'emploi. Cette population synthétique est l'état initial du modèle. * un modèle permettant de simuler une table d'origine-destination des déplacements domicile-travail à partir de données agrégées. * un modèle permettant d'estimer le nombre d'emplois dans les services de proximité dans une commune donnée en fonction de son nombre d'habitants et de son voisinage en termes de service. * une méthode de calibration des paramètres inconnus du modèle SimVillages de manière à satisfaire un ensemble de critères d'erreurs définis sur des sources de données hétérogènes. Cette méthode est fondée sur un nouvel algorithme d'échantillonnage séquentiel de type Approximate Bayesian Computation.