In this thesis we study the optimal stochastic control problem of the drift of a Lévy process. We show that, for a broad class of Lévy processes, the partial integro-differential Hamilton-Jacobi-Bellman equation for the value function admits classical solutions and that control policies exist in feedback form. We then explore the class of Lévy processes that satisfy the requirements of the theorem, and find connections between the uniform integrability requirement and the notions of the score function and Fisher information from information theory. Finally we present three different numerical implementations of the control problem: a traditional dynamic programming approach, and two iterative approaches, one based on a finite difference scheme and the other on the Fourier transform. / text

L'objectif de ma thèse est d'étudier mathématiquement un indicateur de rupture de volatilité très utilisé par les praticiens en salle de marché. L'indicateur bandes de Bollinger appartient à la famille des méthodes dites d'analyse technique et donc repose exclusivement sur l'historique récente du cours considéré et un principe déduit des observations passées des marchés, indépendamment de tout modèle mathématique. Mon travail consiste à étudier les performances de cet indicateur dans un univers qui serait gouverné par des équations différentielles stochastiques (Black -Scholes) dont le coefficient de diffusion change sa valeur à un temps aléatoire inconnu et inobservable, pour un praticien désirant maximiser une fonction objectif (par exemple, une certaine utilité espérée de la valeur du portefeuille à une certaine maturité). Dans le cadre du modèle, l'indicateur de Bollinger peut s'interpréter comme un estimateur de l'instant de la prochaine rupture. On montre dans le cas des petites volatilités, que le comportement de la densité de l'indicateur dépend de la volatilité, ce qui permet pour un ratio de volatilité assez grand, de détecter via l'estimation de la distribution de l'indicateur dans quel régime de volatilité on se situe. Aussi, dans le cas des grandes volatilités, on montre par une approche via la transformée de Laplace, que le comportement asymptotique des queues de distribution de l'indicateur dépend de la volatilité. Ce qui permet de détecter le changement des grandes volatilités. Ensuite, on s'intéresse à une étude comparative entre l'indicateur de Bollinger et l'estimateur classique de la variation quadratique pour la détection de changement de la volatilité. Enfin, on étudie la gestion optimale de portefeuille qui est décrite par un problème stochastique non standard en ce sens que les contrôles admissibles sont contraints à être des fonctionnelles des prix observés. On résout ce problème de contrôle en s'inspirant de travaux de Pham and Jiao pour décomposer le problème initial d'allocation de portefeuille en un problème de gestion après la rupture et un problème avant la rupture, et chacun de ces problèmes est résolu par la méthode de la programmation dynamique . Ainsi, un théorème de verification est prouvé pour ce problème de contrôle stochastique. / The aim of my thesis is to study mathematically an indicator widely used by the practitioners in the trading market, and designed to detect changes in the volatility term . The Bollinger Bands indicator belongs to the family of methods known as technical analysis which consist in looking t the past price movement in order to predict its future price movements independently of any mathematical model. We study the performance of this indicator in a universe that is governed by a stochastic differential equations (Black-Scholes) such that the volatility changes at an unknown and unobservable random time, for a practitioner seeking to maximize an objective function (for instance, the expected utility of the wealth at a certain maturity). Within the framework of the model, Bollinger indicator can be interpreted as an estimator of the time at which the volatility changes its value. We show that in the case of small volatilities, the density behavior of the indicator depends on the value of the volatility, which allows that for large ratio of volatility, to detect via the distribution estimation in which regime of volatility we are. Also , for the case of large volatilities, we show by an approach via the Laplace transform that the asymptotic tails behavior of the indictor depends on the volatility value. This allows to detect a change for large volatilities. Next, we compare two indicators designed to detect a volatility change: the Bollinger bands and the quadratic variation indicators. Finally, we study the optimal portfolio allocation which is described by a non-standard stochastic problem in view of that the admissible controls need to be adapted to the filtration generated by the prices. We resolve this control problem by an approach used by Pham and Jiao to separate the initial allocation problem into an allocation problem after the rupture and an problem before the rupture, and each one of these problems is resolved by the dynamic programming method. Also, a verification theorem is proved for this stochastic control problem.

In this thesis, we study homogenization of optimal control problems in various oscillatory domains. Specifically, we consider four types of domains given in Figure 1 below. Figure 1: Oscillating Domains The thesis is organized into six chapters. Chapter 1 provides an introduction to our work and the rest of the thesis. The main contributions of the thesis are contained in Chapters 2-5. Chapter 6 presents the conclusions of the thesis and possible further directions. A brief description of our work (Chapters 2-5) follows: Chapter 2: Asymptotic behaviour of a fourth order boundary optimal control problem with Dirichlet boundary data posed on an oscillating domain as in Figure 1(A) is analyzed. We use the unfolding operator to study the asymptotic behavior of this problem. Chapter 3: Homogenization of a time dependent interior optimal control problem on a branched structure domain as in Figure 1(B) is studied. Here we pose control on the oscillating interior part of the domain. The analysis is carried out by appropriately defined unfolding operators suitable for this domain. The optimal control is characterized using various unfolding operators defined at each branch of every level. Chapter 4: A new unfolding operator is developed for a general oscillating domain as in Figure 1(C). Homogenization of a non-linear elliptic problem is studied using this new un-folding operator. Using this idea, homogenization of an optimal control problem on a circular oscillating domain as in Figure 1(D) is analyzed. Chapter 5: Homogenization of a non-linear optimal control problem posed on a smooth oscillating domain as in Figure 1(C) is studied using the unfolding operator.

This dissertation contains two parts: Commutative and Noncommutative Multidimensional ($d$-D) Linear Systems Theory. The first part focuses on the development of the interpolation theory to solve the $H^{\infty}$ control problem for $d$-D linear systems. We first review the classical discrete-time 1D linear system in the operator theoretical viewpoint followed by the formulations of the so-called Givone-Roesser and Fornasini-Marchesini models. Application of the $d$-variable $Z$-transform to the system of equations yields the transfer function which is a rational function of several complex variables, say $\mathbf{z} = (z_{1}, \dots, z_{d})$. We then consider the output feedback stabilization problem for a plant $P(\mathbf{z})$. By assuming that $P(\mathbf{z})$ admits a double coprime factorization, then a set of stabilizing controllers $K(\mathbf{z})$ can be parametrized by the Youla parameter $Q(\mathbf{z})$. By doing so, one can convert such a problem to the model matching problem with performance index $F(\mathbf{z})$, affine in $Q(\mathbf{z})$. Then, with $F(\mathbf{z})$ as the design parameter rather than $Q(\mathbf{z})$, one has an interpolation problem for $F(\mathbf{z})$. Incorporation of a tolerance level on $F(\mathbf{z})$ then leads to an interpolation problem of multivariable Nevanlinna-Pick type. We also give an operator-theoretic formulation of the model matching problem which lends itself to a solution via the commutant lifting theorem on the polydisk. The second part details a system whose time-axis is described by a free semigroup $\mathcal{F}_{d}$. Such a system can be represented by the so-called noncommutative Givone-Roesser, or noncommutative Fornasini-Marchesini models which are analogous to those in the first part. Application of a noncommutative $d$-variable $Z$-transform to the system of equations yields the transfer function expressed by a formal power series in several noncommuting indeterminants, say $T(z) = \sum_{v \in \mathcal{F}_{d}}T_{v}z^{v}$ where $z^{v} = z_{i_{n}} \dotsm z_{i_{1}}$ if $v = g_{i_{n}} \dotsm g_{i_{1}} \in \mathcal{F}_{d}$ and $z_{i}z_{j} \neq z_{j}z_{i}$ unless $i = j$. The concepts of reachability, controllability, observability, similarity, and stability are introduced by means of the state-space interpretation. Minimal realization problems for noncommutative Givone-Roesser or Fornasini-Marchesini systems are solved directly by a shift-realization procedure constructed from appropriate noncommutative Hankel matrices. This procedure adapts the ideas of Schützenberger and Fliess originally developed for "recognizable series" to our systems. / Ph. D.

The work selects a specific issue from the numerical analysis of optimal control problems. We investigate a linear-quadratic optimal control problem based on a partial differential equation on 3-dimensional non-convex domains. Based on efficient solution methods for the partial differential equation an algorithm known from control theory is applied. Now the main objectives are to prove that there is no degradation in efficiency and to verify the result by numerical experiments. We describe a solution method which has second order convergence, although the intermediate control approximations are piecewise constant functions. This superconvergence property is gained from a special projection operator which generates a piecewise constant approximation that has a supercloseness property, from a sufficiently graded mesh which compensates the singularities introduced by the non-convex domain, and from a discretization condition which eliminates some pathological cases. Both isotropic and anisotropic discretizations are investigated and similar superconvergence properties are proven. A model problem is presented and important results from the regularity theory of solutions to partial differential equation in non-convex domains have been collected in the first chapters. Then a collection of statements from the finite element analysis and corresponding numerical solution strategies is given. Here we show newly developed tools regarding error estimates and projections into finite element spaces. These tools are necessary to achieve the main results. Known fundamental statements from control theory are applied to the given model problems and certain conditions on the discretization are defined. Then we describe the implementation used to solve the model problems and present all computed results.

Submitted by DANIELLA PORTO null (danielinha.dani@gmail.com) on 2016-03-24T18:05:56Z No. of bitstreams: 1 TESE Daniella Porto.pdf: 1058349 bytes, checksum: ed5227eb69daeb674962db0bf4513f1f (MD5) / Approved for entry into archive by Ana Paula Grisoto (grisotoana@reitoria.unesp.br) on 2016-03-24T19:42:30Z (GMT) No. of bitstreams: 1 porto_d_dr_sjrp.pdf: 1058349 bytes, checksum: ed5227eb69daeb674962db0bf4513f1f (MD5) / Made available in DSpace on 2016-03-24T19:42:30Z (GMT). No. of bitstreams: 1 porto_d_dr_sjrp.pdf: 1058349 bytes, checksum: ed5227eb69daeb674962db0bf4513f1f (MD5) Previous issue date: 2016-03-15 / Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) / Consideramos dois tipos de problemas de controle ótimo: a) Problemas de controle impulsivo e b) problemas de controle ótimo sobre domínios estratificados. Organizamos o trabalho em duas partes distintas. A primeira parte é dedicada ao estudo de um problema de controle impulsivo onde a técnica de reparametrização usual do problema impulsivo é usada para obter um problema regular. Então nós damos resultados de aproximações consistentes via discretização de Euler em que uma sequência de problemas aproximados é obtida com a propriedade que se existe uma subsequência de processos que são ótimos para o correspondente problema discreto que converge para algum processo limite, então o último é ótimo para o problema reparametrizado original. A partir da solução ótima reparametrizada somos capazes de fornecer a solução do problema impulsivo original. A segunda parte considera o problema de tempo mínimo definido sobre domínios estratificados. Definimos o problema e estabelecemos desigualdades de Hamilton Jacobi. Então, damos algumas motivações via Lei de Snell e o problema do Elvis e finalmente fornecemos condições de otimalidade necessárias e suficientes. / We consider two types of optimal control problems: a) Impulsive control problems and b) optimal control problems in stratified domains. So we organize this work in two distinct parts. The first part is dedicated to the study of an impulsive optimal control problem where the usual reparametrization technique of the impulsive problem is used to obtain a regular problem. Then we provide consistent approximation results via Euler discretization in which a sequence of related approximated problems is obtained with the property that if there is a subsequence of processes which are optimal for the corresponding discrete problems which converge to some limit process, then the latter is optimal to the original reparametrized problem. From the reparametrized optimal solution we are able to provide the solution to the original impulsive problem. The second part is regarding the minimal time problem defined on stratified domains. We sate the problem and establish Hamilton-Jacobi inequalities. Then we give some motivation via Snell's law and the Elvis problem and finally we provide necessary and sufficient conditions of optimality. / FAPESP: 2011/14121-9

Made available in DSpace on 2015-04-11T14:02:46Z (GMT). No. of bitstreams: 1 Adriana Gomes Penaranda.pdf: 2772639 bytes, checksum: e4d23c72018fc1400d20f9996f6aacc1 (MD5) Previous issue date: 2013-03-01 / FAPEAM - Fundação de Amparo à Pesquisa do Estado do Amazonas / Wireless Sensor Networks (WSNs) are composed of a large number of sensor nodes. These networks require density control to ensure a better functioning because the high concentration of sensor nodes generates collision data, interference, and retransmittions. In addition, sensor nodes have limited energy, processing, and communication, therefore is interesting to optimize the energy consumption of the network in order to extend its lifetime. Density control schemes have been used to prolong the network lifetime. The Density Control Problem in Wireless Sensor Networks (DCP-WSNs) minimizes the energy consumed by the sensor nodes active, choosing a subset of sensor nodes that meets the application requirements and maximize the use of network resources. This paper presents two approaches to treat DCP-WSN: Periodic and Multiperiod. The Periodic Approach always chooses the best solution for a given period, having a local view of the network lifetime and repeats this proceduce periodically. The Multiperiod Approach defines an expected life time of the network and divide it into periods. For each period the solution is chosen taking into consideration the other periods, thus with an global view of the network lifetime and periods. Both approaches are modeled with Integer Linear Programming and solved by an optimization software. For the Periodic Approach model is proposed a Lagrangean Relaxation with a Lagrangean Heuristic which relax difficults constraints in order to make the problem easier to be solved. We also present a Genetic Algorithm Hybrid (GA) which uses the Periodic Approach to generate the solution of each period and execute a refinement stage based on concepts of the Multiperiod Approach. The proposed heuristics are compared with algorithms of the literature and results show that the Lagrangean Relaxation and Heuristic reach better energy consumption and solution time. Furthermore the Lagrangean relaxation generates lower bounds for the DCP-WSN that may be used to evaluate other algorithms Density Control. / As Redes de Sensores Sem Fios (RSSFs) são redes compostas por um grande número de nós de sensores. Estas redes necessitam de controle de densidade para garantir um melhor funcionamento, pois a alta concentração de nós sensores gera colisão de dados, interferências e consequentemente retransmissão de dados. Os nós sensores possuem limitações de energia, processamento e comunicação e por isto é interessante otimizar o consumo de energia da rede com o objetivo de estender seu tempo de vida. Esquemas de controle de densidade têm sido utilizados como recursos para prolongar o tempo de vida da rede. O Problema de Controle de Densidade em Redes de Sensores Sem Fios (PCD-RSSFs) consiste em minimizar a energia consumida pelos nós sensores ativos, escolhendo um subconjunto de nós que atenda os requisitos da aplicação e maximize a utilização dos recursos da rede. Este trabalho apresenta duas abordagens para tratar o PCD-RSSFs: Periódica e Multiperíodo. A Abordagem Periódica escolhe a melhor solução para um dado período, tendo uma visão local do tempo de vida da rede e repete este procedimento periodicamente. A Abordagem Multiperíodo consiste em definir um tempo esperado de vida da rede e dividí-lo em períodos. Para cada período a solução é escolhida levando em consideração os outros períodos, caracterizando uma visão global do tempo de vida da rede e dos períodos. Ambas as abordagens foram modeladas com Programação Linear Inteira e resolvidas por um software de otimização. Para a modelagem da Abordagem Periódica é proposta uma Relaxação Lagrangeana em conjunto com uma Heurística Lagrangeana onde a ideia é relaxar restrições difíceis com o intuito de deixar o problema mais simples de ser resolvido. Também é apresentado um Algoritmo Genético (AG) híbrido que utiliza Abordagem Periódica para gerar a solução de cada período e em seguida uma fase de refinamento baseada nos conceitos da Abordagem Multiperíodo. As heurísticas implementadas são comparadas com algoritmos da literatura e os resultados mostram que a combinação Relaxação Lagrangeana e Heurística Lagrangeana obtêm melhor desempenho tanto em consumo de energia quanto em tempo de solução. Além disso a Relaxação Lagrangeana gera limites inferiores para o PCD-RSSFs que podem ser utilizados para avaliação de outros algoritmos de controle de Densidade

The work selects a specific issue from the numerical analysis of optimal control problems. We investigate a linear-quadratic optimal control problem based on a partial differential equation on 3-dimensional non-convex domains. Based on efficient solution methods for the partial differential equation an algorithm known from control theory is applied. Now the main objectives are to prove that there is no degradation in efficiency and to verify the result by numerical experiments. We describe a solution method which has second order convergence, although the intermediate control approximations are piecewise constant functions. This superconvergence property is gained from a special projection operator which generates a piecewise constant approximation that has a supercloseness property, from a sufficiently graded mesh which compensates the singularities introduced by the non-convex domain, and from a discretization condition which eliminates some pathological cases. Both isotropic and anisotropic discretizations are investigated and similar superconvergence properties are proven. A model problem is presented and important results from the regularity theory of solutions to partial differential equation in non-convex domains have been collected in the first chapters. Then a collection of statements from the finite element analysis and corresponding numerical solution strategies is given. Here we show newly developed tools regarding error estimates and projections into finite element spaces. These tools are necessary to achieve the main results. Known fundamental statements from control theory are applied to the given model problems and certain conditions on the discretization are defined. Then we describe the implementation used to solve the model problems and present all computed results.

<p>Numerical trajectory optimization is helpful not only for mission planning but also design</p> <p>space exploration and quantifying vehicle performance. Direct methods for solving the opti-</p> <p>mal control problems, which first discretize the problem before applying necessary conditions</p> <p>of optimality, dominate the field of trajectory optimization because they are easier for the</p> <p>user to set up and are less reliant on a forming a good initial guess. On the other hand,</p> <p>many consider indirect methods, which apply the necessary conditions of optimality prior to</p> <p>discretization, too difficult to use for practical applications. Indirect methods though provide</p> <p>very high quality solutions, easily accessible sensitivity information, and faster convergence</p> <p>given a sufficiently good guess. Those strengths make indirect methods especially well-suited</p> <p>for generating large data sets for system analysis and worth revisiting.</p> <p>Recent advancements in the application of indirect methods have already mitigated many</p> <p>of the often cited issues. Automatic derivation of the necessary conditions with computer</p> <p>algebra systems have eliminated the manual step which was time-intensive and error-prone.</p> <p>Furthermore, regularization techniques have reduced problems which traditionally needed</p> <p>many phases and complex staging, like those with inequality path constraints, to a signifi-</p> <p>cantly easier to handle single arc. Finally, continuation methods can circumvent the small</p> <p>radius of convergence of indirect methods by gradually changing the problem and use previ-</p> <p>ously found solutions for guesses.</p> <p>The new optimal control problem solver Giuseppe incorporates and builds upon these</p> <p>advancements to make indirect methods more accessible and easily used. It seeks to enable</p> <p>greater research and creative approaches to problem solving by being more flexible and</p> <p>extensible than previous solvers. The solver accomplishes this by implementing a modular</p> <p>design with well-defined internal interfaces. Moreover, it allows the user easy access to and</p> <p>manipulation of component objects and functions to be use in the way best suited to solve</p> <p>a problem.</p> <p>A new technique simplifies and automates what was the predominate roadblock to using</p> <p>continuation, the generation of an initial guess for the seed solution. Reliable generation of</p> <p>a guess sufficient for convergence still usually required advanced knowledge optimal contrtheory or sometimes incorporation of an entirely separate optimization method. With the</p> <p>new method, a user only needs to supply initial states, a control profile, and a time-span</p> <p>over which to integrate. The guess generator then produces a guess for the “primal” problem</p> <p>through propagation of the initial value problem. It then estimates the “dual” (adjoint)</p> <p>variables by the Gauss-Newton method for solving the nonlinear least-squares problem. The</p> <p>decoupled approach prevents poorly guessed dual variables from altering the relatively easily</p> <p>guess primal variables. As a result, this method is simpler to use, faster to iterate, and much</p> <p>more reliable than previous guess generation techniques.</p> <p>Leveraging the continuation process also allows for greater insight into the solution space</p> <p>as there is only a small marginal cost to producing an additional nearby solutions. As a</p> <p>result, a user can quickly generate large families of solutions by sweeping parameters and</p> <p>modifying constraints. These families provide much greater insight in the general problem</p> <p>and underlying system than is obtainable with singular point solutions. One can extend</p> <p>these analyses to high-dimensional spaces through construction of compound continuation</p> <p>strategies expressible by directed trees.</p> <p>Lastly, a study into common convergence explicates their causes and recommends mitiga-</p> <p>tion strategies. In this area, the continuation process also serves an important role. Adaptive</p> <p>step-size routines usually suffice to handle common sensitivity issues and scaling constraints</p> <p>is simpler and out-performs scaling parameters directly. Issues arise when a cost functional</p> <p>becomes insensitive to the control, which a small control cost mitigates. The best perfor-</p> <p>mance of the solver requires proper sizing of the smoothing parameters used in regularization</p> <p>methods. An asymptotic increase in the magnitude of adjoint variables indicate approaching</p> <p>a feasibility boundary of the solution space.</p> <p>These techniques for indirect methods greatly facilitate their use and enable the gen-</p> <p>eration of large libraries of high-quality optimal trajectories for complex problems. In the</p> <p>future, these libraries can give a detailed account of vehicle performance throughout its flight</p> <p>envelope, feed higher-level system analyses, or inform real-time control applications.</p>

In dieser Arbeit untersuchen wir das Problem der Maximierung bestimmter konkaver Funktionale auf dem Raum der optionalen, zufälligen Maße. Deartige Funktionale treten in der mikroökonomischen Literatur auf, wo ihre Maximierung auf die Bestimmung des optimalen Konsumplans eines ökomischen Agenten hinausläuft. Als Alternative zu den wohlbekannten Methoden der dynamischen Programmierung wird ein neuer Zugang vorgestellt, der es erlaubt, die Struktur der maximierenden Maße in einem über den üblicherweise angenommenen Markovschen Rahmen hinausgehenden, allgemeinen Semimartingalrahmen zu klären. Unser Zugang basiert auf einer unendlichdimensionalen Version des Kuhn-Tucker-Theorems. Die implizierten Bedingungen erster Ordnung erlauben es uns, das Maximierungsproblem auf ein neuartiges Darstellungsproblem für optionale Prozesse zu reduzieren, das damit als ein nicht-Markovsches Substitut für die Hamilton-Jacobi-Bellman Gleichung der dynamischen Programmierung dient. Um dieses Darstellungsproblem im deterministischen Fall zu lösen, führen wir eine zeitinhomogene Verallgemeinerung des Konvexitätsbegriffs ein. Die Lösung im allgemeinen stochastischen Fall ergibt sich über eine enge Beziehung zur Theorie des Gittins-Index der optimalen dynamischen Planung. Unter geeigneten Annahmen gelingt ihre Darstellung in geschlossener Form. Es zeigt sich dabei, daß die maximierenden Maße absolutstetig, diskret und auch singulär sein können, je nach Struktur der dem Problem zugrundeliegenden Stochastik. Im mikroökonomischen Kontext ist es natürlich, daß Problem in einen Gleichgewichtsrahmen einzubetten. Der letzte Teil der Arbeit liefert hierzu ein allgemeines Existenzresultat für ein solches Gleichgewicht. / In this thesis, we study the problem of maximizing certain concave functionals on the space of optional random measures. Such functionals arise in microeconomic theory where their maximization corresponds to finding the optimal consumption plan of some economic agent. As an alternative to the well-known methods of Dynamic Programming, we develop a new approach which allows us to clarify the structure of maximizing measures in a general stochastic setting extending beyond the usually required Markovian framework. Our approach is based on an infinite-dimensional version of the Kuhn-Tucker Theorem. The implied first-order conditions allow us to reduce the maximization problem to a new type of representation problem for optional processes which serves as a non-Markovian substitute for the Hamilton-Jacobi-Bellman equation of Dynamic Programming. In order to solve this representation problem in the deterministic case, we introduce a time-inhomogeneous generalization of convexity. The stochastic case is solved by using an intimate relation to the theory of Gittins-indices in optimal dynamic scheduling. Closed-form solutions are derived under appropriate conditions. Depending on the underlying stochastics, maximizing random measures can be absolutely continuous, discrete, and also singular. In the microeconomic context, it is natural to embed the above maximization problem in an equilibrium framework. In the last part of this thesis, we give a general existence result for such an equilibrium.