Cette thèse est divisée en deux parties. Dans la première partie, on considère le système de réaction-diffusion-advection (Pε), qui est un modèle d'haptotaxie, mécanisme lié à la dissémination de tumeurs cancéreuses. Le résultat principal concerne la convergence de la solution du systeme (Pε) vers la solution d'un problème à frontière libre (P0) qui est bien défini. Dans la seconde partie, on considère une classe générale d'équations elliptiques du type Hénon:−∆u = |x|^{α} f(u) dans Ω ⊂ R^N avec α > -2. On examine deux cas classiques : f(u) = e^u, |u|^{p−1} u et deux autres cas : f(u) = u^{p}_{+} puis f(u) nonlinéarité générale. En étudiant les solutions stables en dehors d'un ensemble compact (en particulier, solutions stables et solutions avec indice de Morse fini) avec différentes méthodes, on obtient des résultats de classification. / This thesis is divided into two main parts. In the first part, we consider an example of reaction-diffusion-taxis system (Pε), which is a haptotaxis model - a mechanism about the spread of cancer cells. The main result concerns the convergence of the solution of System (Pε) to the solution of a free boundary problem (P0), where system (P0) is well-posed. In the second part, we consider a general class of Hénon type elliptic equations : −∆u = |x|^{α} f(u) in Ω ⊂ R^Nwith α > −2. We investigate two classical cases f(u) = e^u, |u|^{p−1} u and two others cases f(u) = u^{p}_{+} , f(u) is a general function. By studying the solutions which are stable outside a compact set (in particular, stable solutions and finite Morse index solutions) with different methods, we establish some classification results.

CoordenaÃÃo de AperfeÃoamento de Pessoal de NÃvel Superior / Conselho Nacional de Desenvolvimento CientÃfico e TecnolÃgico / In this work we establish local regularity estimates for at solutions to non-convex fully nonlinear elliptic equations and we study cavitation type equations modeled within coef- icients bounded and measurable. / Neste trabalho estabelecemos estimativas de regularidade local para soluÃÃes "flat" de equaÃÃes elÃpticas totalmente nÃo-lineares nÃo-convexas e estudamos equations do tipo cavidade com coeficientes meramente mensurÃveis.

Cette thèse est dédiée à l'analyse mathématique de quelques problèmes variationnels motivés par le modèle de Ginzburg-Landau en théorie de la supraconductivité. Dans la première partie on étudie l'existence de solutions pour les équations de Ginzburg-Landau sans champ magnétique et avec données au bord de type semi-rigides. Ces données consistent à prescrire le module de la fonction sur le bord du domaine ainsi que son degré topologique. C'est un cas particulier de problèmes à bord libre, ou la donnée complète de la fonction sur le bord est une inconnue du problème. L'existence de solutions à ce problème n'est pas assurée. En effet la méthode directe du calcul des variations ne peut pas s'appliquer car le degré sur le bord n'est pas continu pour la convergence faible dans l'espace de Sobolev adapté. On dit que c'est un problème sans compacité. En étudiant le phénomène de "bubbling" qui apparaît dans l'étude de tels problèmes on donne des résultats d'existence et de non existence de solutions. Dans le Chapitre 1 on étudie des conditions qui permettent d'affirmer que la différence entre deux niveaux d'énergie est strictement optimale. Pour cela on adapte une technique due à Brezis-Coron. Ceci nous permet de redémontrer un résultat (précédemment obtenu par Berlaynd Rybalko et Dos Santos) d'existence de solutions stables pour les équations de Ginzburg-Landau dans des domaines multiplement connexes. Dans le Chapitre 2 on considère les applications harmoniques a valeurs dans $R^2$ avec des conditions au bord de type degrés prescrits sur un anneau. On fait un lien entre ce problème et la théorie des surfaces minimales dans $R^3$ grâce à la différentielle quadratique de Hopf. Ceci nous conduit à l'étude des surfaces minimales bordées par deux cercles dans des plans parallèles. On prouve l'existence de telles surfaces qui ne sont pas des catenoides grâce a un résultat de bifurcation. On utilise alors les résultats obtenus pour déduire des théorèmes d'existence et de non existence de minimiseurs de l'énergie de Ginzburg-Landau à degrés prescrits dans un anneau. Dans ce troisième Chapitre on obtient des résultats pour une valeur du paramètre " grand. Le Chapitre 4 a pour objet l'étude des problèmes a degrés prescrits en dimension n3. On y montre la non existence des minimiseurs de la n-énergie de Ginzburg-Landau a degrés prescrits dans un domaine simplement connexe. On étudie ensuite des points critiques de type min-max pour une énergie perturbée. La deuxième partie est consacrée a l'analyse asymptotique des solutions des équations deGinzburg-Landau lorsque " tend vers zero. Sandier et Serfaty ont étudié le comportement asymptotique des mesures de vorticité associées aux équations. Ils ont notamment trouvé des conditions critiques sur les mesures limites dans le cas des équations avec et sans champ magnétique. Nous nous intéressons alors à ces conditions critiques dans le cas sans champ magnétique. Le problème de la régularité locale des mesures limites se ramène ainsi a l'étude de la régularité des fonctions stationnaires harmoniques dont le Laplacien est une mesure. Nous montrons que localement de telles mesures sont supportées par une union de lignes appartenant à l'ensemble des zéros d'une fonction harmonique / This thesis is devoted to the mathematical analysis of some variational problems. These problem sare motivated by the Ginzburg-Landau model related to the super conductivity. In the first part we study existence of solutions of the Ginzburg-Landau equations without magnetic eld but with semi-sti boundary conditions. These conditions are obtained by prescribing the modulus of the function on the boundary of the domain along with its topological degree. This is a particular case of free boundary problems, where the function on the boundary is an unknown of the problem. Existence of solutions of that problem does not necessary hold. Indeed we can not apply the direct method of the calculus of variations since the degree on the boundaryis not continuous with respect to the weak convergence in an appropriated Sobolev space. This is problem with loss of compactness. By studying the bublling" phenomenon which come upin such problems we obtain some existence and non existence results .In Chapter 1 we study conditions under which the dierence between two energy levels is strictly optimal. In order to do that we adapt a technique due to Brezis-Coron. This allow us to recover known existence results (previously obtained by Berlyand and Rybalko and DosSantos) for stable solutions of the Ginzburg-Landau equations in multiply connected domains. In Chapter 2 we are interested in harmonic maps with values in $R^2$ with prescribed degree boundary condition in an annulus. We make a link between this problem and the minimal surface theory in $R^3$ thanks to the so-called Hopf quadratic differential. This leads us to study immersed minimal surfaces bounded by two circles in parallel planes. We prove the existence of such surfaces die rent from catenoids by using a bifurcation argument. We then apply the results obtained to deduce existence and non existence results for minimizers of the Ginzburg-Landau energy with prescribed degrees. This is done in Chapter 3 where the results are obtained for large ".Chapter 4 is devoted to prescribed degree problems in dimension n3 . We prove the non existence of minimizers of the Ginzburg-Landau energy in simply connected domains. We then study min-max critical points of a perturbed energy. The second part is devoted to the asymptotic analysis of solutions of the Ginzburg-Landau equations when "goes to zero. Sandier and Serfaty studied the asymptotic behavior of the vorticity measures associated to these equations. They derived critical conditions on the limiting measures both with and without magnetic Field. We are interested by these conditions when there is no magnetic Field. The problem of the local regularity of the limiting measures is then equivalent to the study of regularity of stationary harmonic functions whose Laplacianis a measure. We show that locally such measures are concentrated on a union of lines which belong to the zero set of an harmonic function

Dans le cadre de la thèse des modèles physico-mathématiques pour des microsystèmes électromécaniques avec une permittivité générale sont développés et analysés par des méthodes mathématiques modernes du domaine des équations aux dérivées partielles. En particulier ces systèmes sont à frontière libre et pour conséquence difficiles à traiter. Des méthodes numériques ont été développées pour valider les résultats analytiques obtenus. / In the framework of this thesis physical/mathematical models for microelectromechanical systems with general permittivity have been developed and analysed with modern mathematical methods from the domain of partial differential equations. In particular these systems are moving boundary problems and thus difficult to handle. Numerical methods have been developed in order to validate the obtained analytical results.

[pt] Primeiro examinamos soluções de viscosidade Lp para equações elípticas totalmente não lineares com ingredientes de fronteira mensuráveis. Ao considerar p0 < p < d, focamos nas estimativas da regularidade dos gradientes derivadas de potenciais não lineares. Encontramos condições para Lipschitz-continuidade local das soluções e continuidade do gradiente. Examinamos avanços recentes na teoria da regularidade decorrentes de estimativas potenciais (não lineares). Nossas descobertas decorrem de – e são inspiradas por – fatos fundamentais na teoria de soluções de Lp-viscosidade, e resultados do trabalho de Panagiota Daskalopoulos, Tuomo Kuusi e Giuseppe Mingione (DKM2014). Na segunda parte provamos a regularidade parcial de mapas harmônicos com peso fracamente estacionários com dados de fronteira livre em um cone. Como ponto de partida, damos uma olhada na teoria da regularidade parcial interior para mapas harmônicos fracionários de minimização de energia intrínseca do espaço euclidiano em variedades Riemannianas compactas e suaves para potências fracionárias estritamente entre zero e um. Mapas harmônicos fracionários intrínsecos podem ser estendidos para mapas harmônicos com peso, então provamos regularidade parcial para mapas harmônicos minimizantes locais com dados de fronteira (parcialmente) livres em meios-espaços, mapas harmônicos fracionários então herdam essa regularidade. / [en] We first examine Lp-viscosity solutions to fully nonlinear elliptic equations with bounded measurable ingredients. By considering p0 < p < d, we focus on gradient-regularity estimates stemming from nonlinear potentials. We find conditions for local Lipschitz-continuity of the solutions and continuity of the gradient. We survey recent breakthroughs in regularity theory arising from (nonlinear) potential estimates. Our findings follow from – and are inspired by – fundamental facts in the theory of Lp-viscosity solutions, and results in the work of Panagiota Daskalopoulos, Tuomo Kuusi and Giuseppe Mingione (DKM2014). In the second part we prove partial regularity of weakly stationary weighted harmonic maps with free boundary data on a cone. As a starting point we take a look at the interior partial regularity theory for intrinsic energy minimising fractional harmonic maps from Euclidean space into smooth compact Riemannian manifolds for fractional powers strictly between zero and one. Intrinsic fractional harmonic maps can be extended to weighted harmonic maps, so we prove partial regularity for locally minimising harmonic maps with (partially) free boundary data on half-spaces, fractional harmonic maps then inherit this regularity.

We consider a variational formulation of a Bernoulli-type free boundary problem for the Laplacian operator with discontinuous boundary data. We show the existence of a weak solution to the problem. Moreover, we show that the solution has symmetry properties inherited by symmetric data. These results are achieved through the use of comparison arguments, the celebrated method of moving planes, and several elaborated techniques from existing literature. / Vi studerar ett Bernoulli frirandsproblem för Laplaceoperatorn med diskontinuerliga randdata. Detta görs via en variationsformulering av problemet. Vi visar att en svag lösning existerar för problemet. Utöver det visar vi bland annat att den svaga lösningen har symmetriegenskaper. Dessa resultat uppnås genom jämförelseargument, den välkända "moving-plane” metoden, samt flera utarbetade tekniker från befintlig litteratur.

過去的三十年間由於評價美式選擇權所產生的自由邊界問題已經有相當的研究成果。本論文將證明自由邊界問題的解為遞增函數。更進一步提出自由邊界凹性的嚴謹証明。利用我們的結論可以得知美式選擇權的最佳履約邊界對時間而言為嚴格遞減的凹函數。這個結果對可用來求導最佳履約邊界的漸近解。 對於美式交換選擇權，我們將其自由邊界問題轉換成單變數的積分方程，同時提供一個永續型美式交換選擇權的評價公式。對於有限時間的美式交換選擇權的最佳履約邊界，我們將提供一個接近到期日的漸近解並發展一個數值方法求其數值解。數值計算的結果顯示漸近解在接近到期日時與數值解非常接近。 對於評價美式選擇權，我們提出使用混合整數非線性規劃(MINLP)的模型，這個模型的最佳解同時提供賣方的完全避險策略、買方的最佳交易策略與美式選擇權的公平價格。因為求算MINLP模型的解需耗用大量的計算時間，我們證明此模型和其非線性規劃的寬鬆問題有相同的最佳解，所以只需求算寬鬆問題即可。觀察數值結果亦顯示非線性規劃的寬鬆問題可以大幅的降低計算的時間。此外，當市場的價格低於公平價格時，我們提出一個最小化賣方期望損失的數學規劃模型，此模型的解提供賣方最小化其期望損失的避險策略。 / In the past three decades, a great deal of effort has been made on solving the free boundary problem (FBP) arising from American option valuation problems. In this dissertation, we show that the solutions, the price and the free boundary, of this FBP are increasing functions. Furthermore, we provide a rigorous verification that the free boundary of this problem is concave. Our results imply that the optimal exercise boundary of an American call is a strictly decreasing concave function of time. These results will provide a useful information to obtain an asymptotic formula for the optimal exercise boundary. For pricing of American exchange options (AEO), we convert the associated FBP into a single variable integral equation (IE) and provide a formula for valuating the perpetual AEO. For the finite horizon AEO, we propose an asymptotic solution as time is near to expiration and develop a numerical method for its optimal exercise boundary. Compared with the computational results, the values of our asymptotic solution are close to the computational results as time is near to expiration. For valuating American options, we develop a mixed integer nonlinear programming (MINLP) model. The solution of the MINLP model provides a hedging portfolio for writers, the optimal trading strategy for buyers, and the fair price for American options at the same time. We show that it can be solved by its nonlinear programming (NLP) relaxation. The numerical results reveal that the use of NLP relaxation reduces the computation time rapidly. Moreover, when the market price is less than the fair price, we propose a minimum expected loss model. The solution of this model provides a hedging strategy that minimizes the expected loss for the writer.

A numerical treatment of non-linear higher-order geometric evolution equations with the level set and the finite element method is presented. The isotropic, weak anisotropic and strong anisotropic situation is discussed. Most of the equations considered in this work arise from the field of thin film growth. A short introduction to the subject is given. Four different models are discussed: mean curvature flow, surface diffusion, a kinetic model, which combines the effects of mean curvature flow and surface diffusion and includes a further kinetic component, and an adatom model, which incorporates in addition free adatoms. As an introduction to the numerical schemes, first the isotropic and weak anisotropic situation is considered. Then strong anisotropies (non-convex anisotropies) are used to simulate the phenomena of faceting and coarsening. The experimentally observed effect of corner and edge roundings is reached in the simulation through the regularization of the strong anisotropy with a higher-order curvature term. The curvature regularization leads to an increase by two in the order of the equations, which results in highly non-linear equations of up to 6th order. For the numerical solution, the equations are transformed into systems of second order equations, which are solved with a Schur complement approach. The adatom model constitutes a diffusion equation on a moving surface. An operator splitting approach is used for the numerical solution. In difference to other works, which restrict to the isotropic situation, also the anisotropic situation is discussed and solved numerically. Furthermore, a treatment of geometric evolution equations on implicitly given curved surfaces with the level set method is given. In particular, the numerical solution of surface diffusion on curved surfaces is presented. The equations are discretized in space by standard linear finite elements. For the time discretization a semi-implicit discretization scheme is employed. The derivation of the numerical schemes is presented in detail, and numerous computational results are given for the 2D and 3D situation. To keep computational costs low, the finite element grid is adaptively refined near the moving curves and surfaces resp. A redistancing algorithm based on a local Hopf-Lax formula is used. The algorithm has been extended by the authors to the 3D case. A detailed description of the algorithm in 3D is presented in this work. / In der Arbeit geht es um die numerische Behandlung nicht-linearer geometrischer Evolutionsgleichungen höherer Ordnung mit Levelset- und Finite-Elemente-Verfahren. Der isotrope, schwach anisotrope und stark anisotrope Fall wird diskutiert. Die meisten in dieser Arbeit betrachteten Gleichungen entstammen dem Gebiet des Dünnschicht-Wachstums. Eine kurze Einführung in dieses Gebiet wird gegeben. Es werden vier verschiedene Modelle diskutiert: mittlerer Krümmungsfluss, Oberflächendiffusion, ein kinetisches Modell, welches die Effekte des mittleren Krümmungsflusses und der Oberflächendiffusion kombiniert und zusätzlich eine kinetische Komponente beinhaltet, und ein Adatom-Modell, welches außerdem freie Adatome berücksichtigt. Als Einführung in die numerischen Schemata, wird zuerst der isotrope und schwach anisotrope Fall betrachtet. Anschließend werden starke Anisotropien (nicht-konvexe Anisotropien) benutzt, um Facettierungs- und Vergröberungsphänomene zu simulieren. Der in Experimenten beobachtete Effekt der Ecken- und Kanten-Abrundung wird in der Simulation durch die Regularisierung der starken Anisotropie durch einen Krümmungsterm höherer Ordnung erreicht. Die Krümmungsregularisierung führt zu einer Erhöhung der Ordnung der Gleichung um zwei, was hochgradig nicht-lineare Gleichungen von bis zu sechster Ordnung ergibt. Für die numerische Lösung werden die Gleichungen auf Systeme zweiter Ordnungsgleichungen transformiert, welche mit einem Schurkomplement-Ansatz gelöst werden. Das Adatom-Modell bildet eine Diffusionsgleichung auf einer bewegten Fläche. Zur numerischen Lösung wird ein Operatorsplitting-Ansatz verwendet. Im Unterschied zu anderen Arbeiten, die sich auf den isotropen Fall beschränken, wird auch der anisotrope Fall diskutiert und numerisch gelöst. Außerdem werden geometrische Evolutionsgleichungen auf implizit gegebenen gekrümmten Flächen mit Levelset-Verfahren behandelt. Insbesondere wird die numerische Lösung von Oberflächendiffusion auf gekrümmten Flächen dargestellt. Die Gleichungen werden im Ort mit linearen Standard-Finiten-Elementen diskretisiert. Als Zeitdiskretisierung wird ein semi-implizites Diskretisierungsschema verwendet. Die Herleitung der numerischen Schemata wird detailliert dargestellt, und zahlreiche numerische Ergebnisse für den 2D und 3D Fall sind gegeben. Um den Rechenaufwand gering zu halten, wird das Finite-Elemente-Gitter adaptiv an den bewegten Kurven bzw. den bewegten Flächen verfeinert. Es wird ein Redistancing-Algorithmus basierend auf einer lokalen Hopf-Lax Formel benutzt. Der Algorithmus wurde von den Autoren auf den 3D Fall erweitert. In dieser Arbeit wird der Algorithmus für den 3D Fall detailliert beschrieben.

The Black-Scholes model and its assumptions has endured its fair share of criticism. One problematic issue is the model’s assumption that market volatility is constant. The past decade has seen numerous publications addressing this issue by adapting the Black-Scholes model to incorporate stochastic volatility. In this dissertation, American put options are priced under the Heston stochastic volatility model using the Crank- Nicolson finite difference method in combination with the Projected Over-Relaxation method (PSOR). Due to the early exercise facility, the pricing of American put options is a challenging task, even under constant volatility. Therefore the pricing problem under constant volatility is also included in this dissertation. It involves transforming the Black-Scholes partial differential equation into the heat equation and re-writing the pricing problem as a linear complementary problem. This linear complimentary problem is solved using the Crank-Nicolson finite difference method in combination with the Projected Over-Relaxation method (PSOR). The basic principles to develop the methods necessary to price American put options are covered and the necessary numerical methods are derived. Detailed algorithms for both the constant and the stochastic volatility models, of which no real evidence could be found in literature, are also included in this dissertation. / MSc (Applied Mathematics), North-West University, Potchefstroom Campus, 2013

The Black-Scholes model and its assumptions has endured its fair share of criticism. One problematic issue is the model’s assumption that market volatility is constant. The past decade has seen numerous publications addressing this issue by adapting the Black-Scholes model to incorporate stochastic volatility. In this dissertation, American put options are priced under the Heston stochastic volatility model using the Crank- Nicolson finite difference method in combination with the Projected Over-Relaxation method (PSOR). Due to the early exercise facility, the pricing of American put options is a challenging task, even under constant volatility. Therefore the pricing problem under constant volatility is also included in this dissertation. It involves transforming the Black-Scholes partial differential equation into the heat equation and re-writing the pricing problem as a linear complementary problem. This linear complimentary problem is solved using the Crank-Nicolson finite difference method in combination with the Projected Over-Relaxation method (PSOR). The basic principles to develop the methods necessary to price American put options are covered and the necessary numerical methods are derived. Detailed algorithms for both the constant and the stochastic volatility models, of which no real evidence could be found in literature, are also included in this dissertation. / MSc (Applied Mathematics), North-West University, Potchefstroom Campus, 2013