Artificial vision is the problem of creating systems capable of processing visual information. A fundamental sub-problem of artificial vision is image segmentation, the problem of detecting a structure from a digital image. Examples of segmentation problems include the detection of a road from an aerial photograph or the determination of the boundaries of the brain's ventricles from medical imagery. The extraction of structures allows for subsequent higher-level cognitive tasks. One of them is shape comparison. For example, if the brain ventricles of a patient are segmented, can their shapes be used for diagnosis? That is to say, do the shapes of the extracted ventricles resemble more those of healthy patients or those of patients suffering from schizophrenia? This thesis deals with the problem of image segmentation and shape comparison in the mathematical framework of partial differential equations. The contribution of this thesis is threefold: 1. A technique for the segmentation of regions is proposed. A cost functional is defined for regions based on a non-parametric functional of the distribution of image intensities inside the region. This cost is constructed to favor regions that are homogeneous. Regions that are optimal with respect to that cost can be determined with limited user interaction. 2. The use of direction information is introduced for the segmentation of open curves and closed surfaces. A cost functional is defined for structures (curves or surfaces) by integrating a local, direction-dependent pattern detector along the structure. Optimal structures, corresponding to the best match with the pattern detector, can be determined using efficient algorithms. 3. A technique for shape comparison based on the Laplace equation is proposed. Given two surfaces, one-to-one correspondences are determined that allow for the characterization of local and global similarity measures. The local differences among shapes (resulting for example from a segmentation step) can be visualized for qualitative evaluation by a human expert. It can also be used for classifying shapes into, for example, normal and pathological classes.

We have directed much attention towards developing quantum trajectory methods which can accurately predict the transmission probabilities for a variety of quantum mechanical barrier scattering processes. One promising method involves solving the complex quantum Hamilton-Jacobi equation with the Derivative Propagation Method (DPM). We present this method, termed complex valued DPM (CVDPM(n)). CVDPM(n) has been successfully employed in the Lagrangian frame to accurately compute transmission probabilities on 'thick' one dimensional Eckart and Gaussian potential surfaces. CVDPM(n) is able to reproduce accurate results with a much lower order of approximation than is required by real valued quantum trajectory methods, from initial wave packet energies ranging from the tunneling case (E[subscript o]=0) to high energy cases (twice the barrier height). We successfully extended CVDPM(n) to two-dimensional problems (one translational degree of freedom representing an Eckart or Gaussian barrier coupled to a vibrational degree of freedom) in the Lagrangian framework with great success. CVDPM helps to explain why barrier scattering from "thick" barriers is a much more well posed problem than barrier scattering from "thin" barriers. Though results in these two cases are in very good agreement with grid methods, the search for an appropriate set of initial conditions (termed an 'isochrone) from which to launch the trajectories leads to a time-consuming search problem that is reminiscent of the rootsearching problem from semi-classical dynamics. In order to circumvent the isochrone problem, we present CVDPM(n) equations of motion which are derived and implemented in the arbitrary Lagrangian-Eulerian frame for a metastable potential as well as the Eckart and Gaussian surfaces. In this way, the isochrone problem can be circumvented but at the cost of introducing other computational difficulties. In order to understand why CVDPM may give better transmission probabilities than real valued counterparts, much attention we have been studying and applying numerical analytic continuation techniques to visualize complex-extended wave packets as well as the complex-extended quantum potential. Numerical analytic continuation techniques have also been used to analytically continue a discrete real-valued potential into the complex plane for CVDPM with very promising results.

The focus of this thesis is optimal path planning. The path planning problem is posed as an optimal control problem, for which the viscosity solution to the static Hamilton-Jacobi-Bellman (HJB) equation is used to determine the optimal path. The Ordered Upwind Method (OUM) has been previously used to numerically approximate the viscosity solution of the static HJB equation for direction-dependent weights. The contributions of this thesis include an analytical bound on the convergence rate of the OUM for the boundary value problem to the viscosity solution of the HJB equation. The convergence result provided in this thesis is to our knowledge the tightest existing bound on the convergence order of OUM solutions to the viscosity solution of the static HJB equation. Only convergence without any guarantee of rate has been previously shown. Navigation functions are often used to provide controls to robots. These functions can suffer from local minima that are not also global minima, which correspond to the inability to find a path at those minima. Provided the weight function is positive, the viscosity solution to the static HJB equation cannot have local minima. Though this has been discussed in literature, a proof has not yet appeared. The solution of the HJB equation is shown in this work to have no local minima that is not also global. A path can be found using this method. Though finding the shortest path is often considered in optimal path planning, safe and energy efficient paths are required for rover path planning. Reducing instability risk based on tip-over axes and maximizing solar exposure are important to consider in achieving these goals. In addition to obstacle avoidance, soil risk and path length on terrain are considered. In particular, the tip-over instability risk is a direction-dependent criteria, for which accurate approximate solutions to the static HJB equation cannot be found using the simpler Fast Marching Method. An extension of the OUM to include a bi-directional search for the source-point path planning problem is also presented. The solution is found on a smaller region of the environment, containing the optimal path. Savings in computational time are observed. A comparison is made in the path planning problem in both timing and performance between a genetic algorithm rover path planner and OUM. A comparison in timing and number of updates required is made between OUM and several other algorithms that approximate the same static HJB equation. Finally, the OUM algorithm solving the boundary value problem is shown to converge numerically with the rate of the proven theoretical bound.

Thesis (Ph. D.)--University of Washington, 1999. / Vita. Includes bibliographical references (p. 87-91).

Thesis (Ph. D.)--University of Texas at Austin, 2009. / Title from PDF title page (University of Texas Digital Repository, viewed on Aug. 6, 2009). Vita. Includes bibliographical references.

In this dissertation, we present two research projects, namely finite-state mean-field games and the Hughes model for the motion of crowds. In the first part, we describe finite-state mean-field games and some applications to socio-economic sciences. Examples include paradigm shifts in the scientific community and the consumer choice behavior in a free market. The corresponding finite-state mean-field game models are hyperbolic systems of partial differential equations, for which we propose and validate a new numerical method. Next, we consider the dual formulation to two-state mean-field games, and we discuss numerical methods for these problems. We then depict different computational experiments, exhibiting a variety of behaviors, including shock formation, lack of invertibility, and monotonicity loss. We conclude the first part of this dissertation with an investigation of the shock structure for two-state problems. In the second part, we consider a model for the movement of crowds proposed by R. Hughes in [56] and describe a numerical approach to solve it. This model comprises a Fokker-Planck equation coupled with an Eikonal equation with Dirichlet or Neumann data. We first establish a priori estimates for the solutions. Next, we consider radial solutions, and we identify a shock formation mechanism. Subsequently, we illustrate the existence of congestion, the breakdown of the model, and the trend to the equilibrium. We also propose a new numerical method for the solution of Fokker-Planck equations and then to systems of PDEs composed by a Fokker-Planck equation and a potential type equation. Finally, we illustrate the use of the numerical method both to the Hughes model and mean-field games. We also depict cases such as the evacuation of a room and the movement of persons around Kaaba (Saudi Arabia).

The aim of this thesis is to study a class of left-invariant optimal control problems on the matrix Lie group SE(2). We classify, under detached feedback equivalence, all controllable (left-invariant) control affine systems on SE(2). This result produces six types of control affine systems on SE(2). Hence, we study six associated left-invariant optimal control problems on SE(2). A left-invariant optimal control problem consists of minimizing a cost functional over the trajectory-control pairs of a left-invariant control system subject to appropriate boundary conditions. Each control problem is lifted from SE(2) to T*SE(2) ≅ SE(2) x se (2)*and then reduced to a problem on se (2)*. The maximum principle is used to obtain the optimal control and Hamiltonian corresponding to the normal extremals. Then we derive the (reduced) extremal equations on se (2)*. These equations are explicitly integrated by trigonometric and Jacobi elliptic functions. Finally, we fully classify, under Lyapunov stability, the equilibrium states of the normal extremal equations for each of the six types under consideration.

Cette thèse concerne les jeux différentiels à somme nulle et à deux joueurs avec information incomplète. La structure de l'information est liée à un signal que reçoivent les joueurs. Cette information est dite symétrique quand la connaissance du signal est la même pour les deux joueurs (le signal est public), et asymétrique quand les signaux reçus par les joueurs peuvent être différents (le signal est privé).Ces signaux sont révélés au cours du jeu. Dans plusieurs situations de tels jeux, il est montré dans cette thèse, l'existence d'une valeur du jeu et sa caractérisation comme unique solution d'une équation aux dérivées partielles.Un type de structure d'information concerne le cas symétrique où le signal est réduit à la connaissance par les joueurs de l'état du système au moment où celui-ci atteint une cible donnée (les données initiales inconnues sont alors révélées). Pour ce type du jeu, nous avons introduit des stratégies non anticipatives qui dépendent du signal et nous avons obtenu l'existence d'une valeur.Comme les fonctions valeurs sont en général irrégulières (seulement continues), un des points clefs de notre approche est de prouver des résultats d'unicité et des principes de comparaison pour des solutions de viscosité lipschitziennes de nouveaux types d'équation d'Hamilton-Jacobi-Isaacs associées aux jeux étudiés. / In this thesis we investigate two-person zero-sum differential games with incomplete information. The information structure is related to a signal communicated to the players during the game.In such games, the information is symmetric if both players receive the same signal (namely it is a public signal). Otherwise, if the players could receive different signals (i.e. they receive private signals), the information is asymmetric. We prove in this thesis the existence of value and the characterization of the value function by a partial differential equation for various types of such games.A particular type of such information structure is the symmetric case in which the players receive as their signal the current state of the dynamical system at the moment when the state of the dynamic hits a fixed target set (the unknown initial data are then revealed to both players). For this type of games, we introduce the notion of signal-depending non-anticipative strategies with delay and we prove the existence of value with such strategies.As the value functions are in general irregular (at most continuous), a crucial step of our approach is to prove the uniqueness results and the comparison principles for viscosity solutions of new types of Hamilton-Jacobi-Isaacs equation associated to the games studied in this thesis.

Orientador: Edmundo Capelas de Oliveira / Dissertação (mestrado profissional) - Universidade Estadual de Campinas, Instituto de Matemática, Estatística e Computação Científica / Made available in DSpace on 2018-08-17T14:35:07Z (GMT). No. of bitstreams: 1 Rosa_EsterCristinaFontesdeAquino_M.pdf: 847998 bytes, checksum: d177526572b19cc1fdd5eeccdf511380 (MD5) Previous issue date: 2011 / Resumo: Este trabalho tem por objetivo modelar e resolver, matematicamente, um problema físico conhecido como pêndulo simples. Discutimos, como caso particular, as chamadas oscilações de pequena amplitude, isto é, uma aproximação que nos leva a mostrar que o período de oscilação é proporcional à raiz quadrada do quociente entre o comprimento do pêndulo e a aceleração da gravidade. Como vários outros problemas oriundos da Física, o pêndulo simples é representado através de equações diferenciais parciais. Assim, na busca de sua solução, aplicamos a metodologia de separação de variáveis que nos leva a um conjunto de equações ordinárias passíveis de simples integração. Escolhendo um sistema de coordenadas adequado, é conveniente usar o método de Hamilton-Jacobi, discutindo, antes, o problema do oscilador harmónico, apresentando, em seguida, o problema do pêndulo simples e impondo condições a fim de mostrar que as equações diferenciais associadas a esses dois sistemas são iguais, ou seja, suas soluções são equivalentes. Para tanto, estudamos o método de separação de variáveis associado às equações diferenciais parciais, lineares e de segunda ordem, com coeficientes constantes e três variáveis independentes, bem como a respectiva classificação quanto ao tipo. Posteriormente, estudamos as equações hipergeométricas, cujas soluções, as funções hipergeométricas. podem ser encontradas pelo método de Frobenius. Apresentamos o método de Hamilton-Jacobi, já mencionado, para o enfren-tamento do problema apresentado. Fizemos no capítulo final um apêndice sobre a função gama por sua presente importância no trato de funções hipergeométricas, em especial a integral elíptica completa de primeiro tipo que compõe a solução exata do período do pêndulo simples / Abstract: This work aims to present and solve, mathematically, the physics problem that is called simple pendulum. We reasoned, as an specific case, the so called low amplitude oscillation, that is, a convenient approximation that make us show that the period of oscillation is proportional to the quotient square root between the pendulum length and the gravity acceleration. Like several other problems arising from the physics, we are going to broach it through partial differential equations. Thus, in the search of its solution, we made use of the variable separation methodology that leads us to a body of ordinary equations susceptible of simple integration. Choosing an appropriate coordinate system, it is convenient to use the method Hamilton-Jacobi, arguing, first, the problem of the harmonic oscillator, with, then the problem of sf simple pendulum and imposing conditions to show that the differential equations associated with these two systems are equal, that is, their solutions are equivalent. With the purpose of reaching the objectives, we studied the variable separation method associated with partial differential equations, linear and of second order, with constant coefficient and three independent variables, as well as the respective classification about the type. Afterwards, we studied the hypergeometrical equations whose solutions, the hypergeometrical functions, are found by the Frobenius method. Introducing the Hamilton-Jacobi method, already mentioned, for addressing the problem presented. We made an appendix in the final chapter on the gamma function by its present importance in dealing with hypergeometric functions, in particular the elliptic integral of first kind consists of the exact period of sf simple pendulum / Mestrado / Fisica-Matematica / Mestre em Matemática

Cette thèse se compose de deux parties principales. Dans la première partie, le Chapitre 3 est consacré à l'étude du comportement limite d'un système contrôlé singulièrement perturbé avec deux variables d'état qui sont faiblement couplées. Afin de prouver notre résultat d'approximation, nous utilisons la méthode de moyennisation et un résultat récent sur le contrôle nonexpansif. La principale nouveauté de notre approche est de permettre la dynamique limite de dépendre de l'état initial du système rapide. Notons que dans la littérature, le comportement limite d'un tel système a été généralement traité dans des conditions qui garantissent que la limite est indépendante de l'état initial du système rapide. Dans le Chapitre 4, nous généralisons les résultats du Chapitre 3 supposant une condition de nonexpansivité plus générale. De plus, nous considérons un exemple ou la nouvelle condition de nonexpansivité est satisfaite, mais pas la condition de nonexpansivité du Chapitre 3. Dans la deuxième partie de la thèse, le Chapitre 5 porte sur les représentations stables des Hamiltoniens convexes associant à un Hamiltonien donné des fonctions correspondant au problème de Bolza en controle optimal. Dans le Chapitre 6 nous étudions également la stabilité des solutions des équations d'Hamilton-Jacobi-Bellman sous contraintes d'état en exploitant la stabilité des fonctions valeur d'une famille de problèmes de contrôle optimal de Bolza sous contraintes d'état. Nous montrons que sous des hypothèses appropriées, la fonction valeur est la solution unique d'équation d'Hamilton-Jacobi-Bellman et que les solutions sont stables par rapport à l'Hamiltonien et les contraintes d'état. / This thesis consists of two main parts. In the first part, Chapter 3 is devoted to the investigation of the limit behavior of a singularly perturbed control system with two state variables which are weakly coupled. In order to prove our approximation result we use the so called averaging method and a recent result on nonexpansive control. The main novelty of our averaging approach lies in the fact that the limit dynamic may depend on the initial condition of the fast system. In the literature, the investigation of the limit behavior of such systems has been usually addressed under conditions that ensure that the limit dynamic is independent from the initial condition of the fast system. In Chapter 4, we generalise the results of Chapter 3 by considering a more general nonexpansivity condition. Moreover, we consider an example where the new nonexpansity condition is satisfied but the nonexpansivity condition of Chapter 3 does not hold true. The second part deals with Hamilton-Jacobi equations under state constraints. Chapter 5 focuses on the stable representation of convex Hamiltonians by functions describing a Bolza optimal control problem. In Chapter 6 we investigate stability of solutions of Hamilton-Jacobi-Bellman equations under state constraints by studying stability of value functions of a suitable family of Bolza optimal control problems under state constraints. We show that under suitable assumptions, the value function is a unique viscosity solution to Hamilton-Jacobi-Bellman equation and that solutions are stable with respect to Hamiltonians and state constraints.