Investiga-se, em tempo discreto, o problema multi-período de otimização de carteiras generalizado em média-variância cujos coeficientes de mercado são modulados por uma cadeia de Markov finita. O problema multi-período generalizado de média-variância com saltos Markovianos (PGMV ) é um problema de controle estocástico sem restrição cuja função objetivo consiste na maximização da soma ponderada ao longo do tempo da combinação linear de três elementos: o valor esperado da riqueza do investidor, o quadrado da esperança desta riqueza e a esperança do quadrado deste patrimônio. A principal contribuição deste trabalho é a derivação analítica de condições necessárias e suficientes para a determinação de uma estratégia ótima de investimento para o problema PGMV . A partir deste modelo são derivadas várias formulações de médiavariância, como o modelo tradicional cujo objetivo é maximizar o valor esperado da riqueza final do investidor, dado um nível de risco (variância) do portfólio no horizonte de investimento, bem como o modelo mais complexo que busca maximizar a soma ponderada das esperanças da riqueza ao longo do tempo, limitando a perda deste patrimônio em qualquer momento. Adicionalmente, derivam-se formas fechadas para a solução dos problemas citados quando as restrições incidem somente no instante final. Outra contribuição deste trabalho é a extensão do modelo PGMV para a solução do problema de seleção de carteiras em média-variância com o objetivo de superar um benchmark estocástico, com restrições sobre o valor esperado ou sobre a variância do tracking error do portfólio. Por fim, aplicam-se os resultados obtidos em exemplos numéricos cujo universo de investimento são todas as ações do IBOVESPA. / In this work we deal with a discrete-time multi-period mean-variance portfolio selection model with the market parameters subject to Markov regime switching. The multi-period generalized mean-variance portfolio selection model with regime switching (PGMV ) is an unrestricted stochastic control problem, in which the objective function involves the maximization of the weighted sum of a linear combination of three parts: the expected wealth, the square of the expected wealth and the expected value of the wealth squared. The main contribution of this work is the analytical derivation of necessary and sufficient conditions for the existence of an optimal control strategy to this PGMV model. We show that several mean-variance models are derived from the PGMV model, as the traditional formulation in which the objective is to maximize the expected terminal wealth for a given final risk (variance), or the complex one in which the objective function is to maximize the weighted sum of the wealth throughout its investment horizon, with control over maximum wealth lost. Additionally, we derive closed forms solutions for the above models when the restrictions are just in the final time. Another contribution of this work is to extend the PGMV model to solve the multi-period portfolio selection problem of beating a stochastic benchmark with control over the tracking error variance or its expected value. Finally, we run numerical examples in which the investment universe is formed by all the stocks belonging to the IBOVESPA.

Les processus markoviens déterministes par morceaux (PMDM) ont été introduits dans la littérature par M.H.A. Davis en tant que classe générale de modèles stochastiques non-diffusifs. Les PMDM sont des processus hybrides caractérisés par des trajectoires déterministes entrecoupées de sauts aléatoires. Dans cette thèse, nous développons des méthodes numériques adaptées aux PMDM en nous basant sur la quantification d'une chaîne de Markov sous-jacente au PMDM. Nous abordons successivement trois problèmes : l'approximation d'espérances de fonctionnelles d'un PMDM, l'approximation des moments et de la distribution d'un temps de sortie et le problème de l'arrêt optimal partiellement observé. Dans cette dernière partie, nous abordons également la question du filtrage d'un PMDM et établissons l'équation de programmation dynamique du problème d'arrêt optimal. Nous prouvons la convergence de toutes nos méthodes (avec le plus souvent des bornes de la vitesse de convergence) et les illustrons par des exemples numériques. / Piecewise-deterministic Markov processes (PDMP’s) have been introduced by M.H.A. Davis as a general class of non-diffusive stochastic models. PDMP’s are hybrid Markov processes involving deterministic motion punctuated by random jumps. In this thesis, we develop numerical methods that are designed to fit PDMP's structure and that are based on the quantization of an underlying Markov chain. We deal with three issues : the approximation of expectations of functional of a PDMP, the approximation of the moments and of the distribution of an exit time and the partially observed optimal stopping problem. In the latter one, we also tackle the filtering of a PDMP and we establish the dynamic programming equation of the optimal stopping problem. We prove the convergence of all our methods (most of the time, we also obtain a bound for the speed of convergence) and illustrate them with numerical examples.

The problem of creating dynamic fare model consists of two tasks - estimating demand for train tickets and multistage optimization of price of fare. We introduce a model of inhomogeneous Markov process for the process of selling the tickets in this thesis. Because of the complexity of the state space the optimization problem needs to be solved using simulation methods. The solution was implemented in R language for single-stage and two-stage problems. Before this application we summarize the theory of inhomogeneous Markov process with special attention to process with separable inhomogeneity. Then we propose methods for estimating the intensity using maximum likelihood theory. We also describe and compare two algorithms for simulated optimization. Powered by TCPDF (www.tcpdf.org)

Orientadores: Nancy Lopes Garcia, Alexsandro Giacomo Grimbert Gallo / Tese (doutorado) - Universidade Estadual de Campinas, Instituto de Matemática Estatística e Computação Científica / Made available in DSpace on 2018-08-24T12:54:22Z (GMT). No. of bitstreams: 1 Carvalho_WalterAugustoFonsecade_D.pdf: 1034671 bytes, checksum: 25dd72305f343655bedfde62a785a259 (MD5) Previous issue date: 2014 / Resumo: Esta tese é dedicada ao estudo dos processos de renovação binários obtidos como agregação de estados a partir de processos Markovianos com alfabeto finito. Na primeira parte, utilizamos uma abordagem matricial para obter condições sob as quais o processo agregado pertence a cada uma das seguintes classes: (1) Markoviano de ordem finita, (2) processo de ordem infinita com probabilidades de transição contínuas, (3) processo Gibbsiano. A segunda parte trata da distância d entre processos de renovação binários. Obtivemos condições sob as quais esta distância pode ser atingida entre tais processos / Abstract: This thesis is devoted to the study of binary renewal processes obtained as aggregation of states from Markov processes with finite alphabet. In the rst part, we use a matrix approach to obtain conditions under which the aggregated process belongs to each of the following classes: (1) Markov of finite order, (2) process of infinite order with continuous transition probabilities, (3) Gibbsian process. The second part deals with the distance d between binary renewal processes. We obtain conditions under which this distance can be achieved between these processes / Doutorado / Estatistica / Doutor em Estatística

Artificial agents are usually designed to achieve specific goals. An agent's competency can be defined as its ability to accomplish its goals under different conditions. This thesis restricts attention to a specific type of goal, namely reaching a desired state without exceeding a tolerance threshold of undesirable events in a first-order Markov process. For such goals, the state-dependent competency for an agent can be defined as the probability of reaching the desired state without exceeding the threshold and within a time limit given an initial state. The thesis further defines total competency as the set of state-dependent competency relationships over all possible initial states. The thesis uses a Monte Carlo approach to establish a baseline for estimating state-dependent competency. The Monte Carlo approach (a) uses trajectories sampled from an agent behaving in the environment, and then (b) uses nonlinear regression over the trajectory samples to estimate the competency curve. The thesis further presents an equation demonstrating recurrent relations for total competency and an algorithm based on that equation for computing total competency whose worst case computation time grows quadratically with the size of the state space. Simple maze-based Markov chains show that the Monte Carlo approach to estimating the competency agrees with the results computed by the proposed algorithm. Lastly, the thesis explores a special case where there are multiple sequential atomic goals that make up a complex goal. The thesis models a set of sequential goals as a Bayesian network and presents an equation based on the chain rule for deriving the competency for the complex goal from the competency for atomic goals. Experiments for the canonical taxi problem with sequential goals show the correctness of the Bayesian network-based decomposition approach.

Les processus markoviens déterministes par morceaux (PDMP) forment une vaste classe de processus stochastiques caractérisés par une évolution déterministe entre des sauts à mécanisme aléatoire. Ce sont des processus de type hybride, avec une composante discrète de mode et une composante d’état qui évolue dans un espace continu. Entre les sauts du processus, la composante continue évolue de façon déterministe, puis au moment du saut un noyau markovien sélectionne la nouvelle valeur des composantes discrète et continue. Dans cette thèse, nous construisons des PDMP évoluant dans des espaces de mesures (de dimension infinie), pour modéliser des population de cellules en tenant compte des caractéristiques individuelles de chaque cellule. Nous exposons notre construction des PDMP sur des espaces de mesure, et nous établissons leur caractère markovien. Sur ces processus à valeur mesure, nous étudions un problème d'arrêt optimal. Un problème d'arrêt optimal revient à choisir le meilleur temps d'arrêt pour optimiser l'espérance d'une certaine fonctionnelle de notre processus, ce qu'on appelle fonction valeur. On montre que cette fonction valeur est solution des équations de programmation dynamique et on construit une famille de temps d'arrêt $epsilon$-optimaux. Dans un second temps, nous nous intéressons à un PDMP en dimension finie, le TCP, pour lequel on construit un schéma d'Euler afin de l'approcher. Ce choix de modèle simple permet d'estimer différents types d'erreurs. Nous présentons des simulations numériques illustrant les résultats obtenus. / Piecewise deterministic Markov processes (PDMP) form a large class of stochastic processes characterized by a deterministic evolution between random jumps. They fall into the class of hybrid processes with a discrete mode and an Euclidean component (called the state variable). Between the jumps, the continuous component evolves deterministically, then a jump occurs and a Markov kernel selects the new value of the discrete and continuous components. In this thesis, we extend the construction of PDMPs to state variables taking values in some measure spaces with infinite dimension. The aim is to model cells populations keeping track of the information about each cell. We study our measured-valued PDMP and we show their Markov property. With thoses processes, we study a optimal stopping problem. The goal of an optimal stopping problem is to find the best admissible stopping time in order to optimize some function of our process. We show that the value fonction can be recursively constructed using dynamic programming equations. We construct some $epsilon$-optimal stopping times for our optimal stopping problem. Then, we study a simple finite-dimension real-valued PDMP, the TCP process. We use Euler scheme to approximate it, and we estimate some types of errors. We illustrate the results with numerical simulations.

No description available.

In this thesis a Markov chain model, which can be used for analysing students’ performance and their academic progress, is developed. Being able to evaluate students progress is useful for any educational system. It gives a better understanding of how students resonates and it can be used as support for important decisions and planning. Such a tool can be helpful for managers of the educational institution to establish a more optimal educational policy, which ensures better position in the educational market. To show that it is reasonable to use a Markov chain model for this purpose, a test for how well data fits such a model is created and used. The test shows that we cannot reject the hypothesis that the data can be fitted to a Markov chain model. / I detta examensarbete utvecklas en Markov-kedjemodell, som kan användas för att analysera studenters prestation och akademiska framsteg. Att kunna utvärdera studenters väg genom studierna är användbart för alla utbildningssystem. Det ger en bättre förståelse för hur studenter resonerar och det kan användas som stöd för viktiga beslut och planering. Ett sådant verktyg kan vara till hjälp för utbildningsinstitutionens chefer att upprätta en mer optimal utbildningspolitik, vilket säkerställer en bättre ställning på utbildningsmarknaden. För att visa att det är rimligt att använda en Markov-kedjemodell för detta ändamål skapas och används ett test för hur väl data passar en sådan modell. Testet visar att vi inte kan avvisa hypotesen att data kan passa en Markov-kedjemodell.

京都大学 / 新制・課程博士 / 博士(情報学) / 甲第24980号 / 情博第853号 / 新制||情||143(附属図書館) / 京都大学大学院情報学研究科システム科学専攻 / (主査)教授 田中 利幸, 教授 下平 英寿, 准教授 本多 淳也 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM

Markov processes have a long history of being used to model safety critical systems. However, with the development of autonomous vehicles and their increased complexity, Markov processes have been shown to not be sufficiently precise for reliability calculations. Therefore there has been the need to consider a more general stochastic process, namely the Semi-Markov process (SMP). SMPs allow for transitions with general distributions between different states and can be used to precisely model complex systems. This comes at the cost of increased complexity when calculating the reliability of systems. As such, methods to increase the interpretability of the system and allow for appropriate approximations have been considered and researched. In this thesis, a novel classification approach for transitions in SMP has been defined and complemented with different conjectures and properties. A transition is classified as good or bad by comparing the reliability of the original system with the reliability of any perturbed system, for which the studied transition is more likely to occur. Cases are presented to illustrate the use of this classification technique. Multiple suggestions and conjectures for future work are also presented and discussed. / Markovprocesser har länge använts för att modellera säkerhetskritiska system. Med utvecklingen av autonoma fordon och deras ökade komplexitet, har dock markovprocesser visat sig vara otillräckliga exakta för tillförlitlighetsberäkningar. Därför har det funnits ett behov för en mer allmän stokastisk process, nämligen semi-markovprocessen (SMP). SMP tillåter generella fördelningar mellan tillstånd och kan användas för att modellera komplexa system med hög noggrannhet. Detta innebär dock en ökad komplexitet vid beräkningen av systemens tillförlitlighet. Metoder för att öka systemets tolkningsbarhet och möjliggöra lämpliga approximationer har därför övervägts och undersökts. I den här masteruppsatsen har en ny klassificeringsmetod för övergångar i SMP definierats och kompletteras med olika antaganden och egenskaper. En övergång klassificeras som antingen bra eller dålig genom en jämförelse av tillförlitligheten i det ursprungliga systemets och ett ändrat system, där den studerade övergången har högre sannolikhet att inträffa. Fallstudier presenteras för att exemplifiera användningen av denna klassificeringsteknik. Flera förslag och antaganden för framtida arbete presenteras och diskuteras också.