A standard problem in mathematical finance is the calculation of the price of some financial derivative such as various types of options. Since there exists analytical solutions in only a few cases it will often boil down to estimating the price with Monte Carlo simulation in conjunction with some numerical discretization scheme. The upside of using what we can call standard Monte Carlo is that it is relative straightforward to apply and can be used for a wide variety of problems. The downside is that it has a relatively slow convergence which means that the computational cost or complexity can be very large. However, this slow convergence can be improved upon by using Multilevel Monte Carlo instead of standard Monte Carlo. With this approach it is possible to reduce the computational complexity and cost of simulation considerably. The aim of this thesis is to introduce the reader to the Multilevel Monte Carlo method with applications to European and Asian call options in both the Black-Scholes-Merton (BSM) model and in the Heston model. To this end we first cover the necessary background material such as basic probability theory, estimators and some of their properties, the stochastic integral, stochastic processes and Ito’s theorem. We introduce stochastic differential equations and two numerical discretizations schemes, the Euler–Maruyama scheme and the Milstein scheme. We define strong and weak convergence and illustrate these concepts with examples. We also describe the standard Monte Carlo method and then the theory and implementation of Multilevel Monte Carlo. In the applications part we perform numerical experiments where we compare standard Monte Carlo to Multilevel Monte Carlo in conjunction with the Euler–Maruyama scheme and Milsteins scheme. In the case of a European call in the BSM model, using the Euler–Maruyama scheme, we achieved a cost O(ε-2(log ε)2) to reach the desired error in accordance with theory in comparison to the O(ε-3) cost for standard Monte Carlo. When using Milsteins scheme instead of the Euler–Maruyama scheme it was possible to reduce the cost in terms of the number of simulations needed to achieve the desired error even further. By using Milsteins scheme, a method with greater order of strong convergence than Euler–Maruyama, we achieved the O(ε-2) cost predicted by the complexity theorem compared to the standard Monte Carlo cost of order O(ε-3). In the final numerical experiment we applied the Multilevel Monte Carlo method together with the Euler–Maruyama scheme to an Asian call in the Heston model. In this case, where the coefficients of the Heston model do not satisfy a global Lipschitz condition, the study of strong or weak convergence is much harder. The numerical experiments suggested that the strong convergence was slightly slower compared to what was found in the case of a European call in the BSM model. Nevertheless, we still achieved substantial savings in computational cost compared to using standard Monte Carlo.

In the Thesis we study the problem of linear filtration of Gaussian signals in finite-dimensional space. We use the Kalman-type equations for the filter to show that the filter depends continuously on the signal. Secondly, we show the same continuity property for the covariance of the error and verify existence and uniqueness of a solution to an integral equation that is satisfied by the filter even under more general assumptions. We present several examples of application of the continuity property that are based on the theory of stochastic differential equations driven by fractional Brownian motion. 1

In the Thesis the problem of estimating an unknown parameter in a stochastic dif- ferential equation is studied. Linear equations with Volterra process as the source of noise are considered. Firstly, the properties of Volterra processes and the properties of stochastic integral with respect to a Volterra process are presented. Secondly, the prop- erties of the solution to the equation under consideration are discussed. This includes the existence of the strictly stationary solution, the properties of such solution and ergodic results. These results are then generalized to equations with a mixed noise. Ergodic results are used to derive strongly consistent estimators of the unknown parameter. 1

In this thesis, we address several problems arising in the study of nondegenerate and degenerate parabolic SPDEs, stochastic hyper- bolic conservation laws and SDEs with continues coefficients. In the first part, we are interested in degenerate parabolic SPDEs, adapt the notion of kinetic formulation and kinetic solution and establish existence, uniqueness as well as continuous dependence on initial data. As a preliminary result we obtain regularity of solutions in the nondegenerate case under the hypothesis that all the coefficients are sufficiently smooth and have bounded derivatives. In the second part, we consider hyperbolic conservation laws with stochas- tic forcing and study their approximations in the sense of Bhatnagar-Gross- Krook. In particular, we describe the conservation laws as a hydrodynamic limit of the stochastic BGK model as the microscopic scale vanishes. In the last part, we provide a new and fairly elementary proof of Skorkohod's classical theorem on existence of weak solutions to SDEs with continuous coefficients satisfying a suitable Lyapunov condition. 1

In this paper, we examine the feasibility of extending the Akaike information criterion (AIC) for deterministic systems as a potential model selection criteria for stochastic models. We discuss the implementation method for three different classes of stochastic models: continuous time Markov chains (CTMC), stochastic differential equations (SDE), and random differential equations (RDE). The effectiveness and limitations of implementing the AIC for comparison of stochastic models is demonstrated using simulated data from the three types of models and then applied to experimental longitudinal growth data for algae.

Many phenomena in our world can be described as differential equations in high dimensions. However, they are notoriously challenging to solve numerically due to the exponential growth in computational cost with increasing dimensions. This thesis explores an algorithm, known as deep BSDE, for solving high-dimensional partial differential equations and applies it to finance, namely European option pricing. In addition, an implementation of the method is provided that seemingly shortens the runtime by a factor of two, compared with the results in previous studies. From the results, we can conclude that the deep BSDE method does handle high-dimensional problems well. Lastly, the thesis gives the relevant prerequisites required to be able to digest the theory from an undergraduate level.

In this project, we are aiming to solve option pricing and hedging problems numerically via Backward Stochastic Differential Equations (BSDEs). We use Markovian BSDEs to formulate nonlinear pricing and hedging problems of both European and American option types. This method of formulation is crucial for pricing financial instruments since it enables consideration of market imperfections and computations in high dimensions. We conduct numerical experiments of the pricing and hedging problems, where there is a higher interest rate for borrowing than lending, using the least squares Monte Carlo and deep neural network methods. Moreover, based on the experiment results, we point out which method to chooseover the other depending on the the problem at hand.

Stochastic differential equations(SDEs) play an important role in many branches of engineering and science including economics, finance, chemistry, biology, mechanics etc. SDEs (with m-dimensional Wiener process) arising in many applications do not have explicit solutions, which implies the development of effective numerical methods for such systems. For SDEs, one can classify the numerical methods into three classes: fully implicit methods, semi-implicit methods and explicit methods. In order to solve SDEs, the computation of Newton iteration is necessary for the implicit and semi-implicit methods whereas for the explicit methods we do not need such computation. In this thesis the common theme is to construct explicit numerical methods with strong order 1.0 and 1.5 for solving Itˆo SDEs. The five-stage Milstein(FSM)methods, split-step forward Milstein(SSFM)methods and M-stage split-step strong Taylor(M-SSST) methods are constructed for solving SDEs. The FSM, SSFM and M-SSST methods are fully explicit methods. It is proved that the FSM and SSFM methods are convergent with strong order 1.0, and M-SSST methods are convergent with strong order 1.5.Stiffness is a very important issue for the numerical treatment of SDEs, similar to the case of deterministic ordinary differential equations. Stochastic stiffness can lead someone to use smaller step-size for the numerical simulation of the SDEs. However, such issues can be handled using numerical methods with better stability properties. The analysis of stability (with multidimensional Wiener process) shows that the mean-square stable regions of the FSM methods are unbounded. The analysis of stability shows that the mean-square stable regions of the FSM and SSFM methods are larger than the Milstein and three-stage Milstein methods. The M-SSST methods possess large mean square stability region as compared to the order 1.5 strong Itˆo-Taylor method. SDE systems simulated with the FSM, SSFM and M-SSST methods show the computational efficiency of the methods. In this work, we also consider the problem of computing numerical solutions for stochastic delay differential equations(SDDEs) of Itˆo form with a constant lag in the argument. The fully explicit methods, the predictor-corrector Euler(PCE)methods, are constructed for solving SDDEs. It is proved that the PCE methods are convergent with strong order γ = ½ in the mean-square sense. The conditions under which the PCE methods are MS-stable and GMS-stable are less restrictive as compared to the conditions for the Euler method.

Depuis la publication de l'ouvrage de Choquet (1955), la théorie d'espérance non linéaire a attiré avec grand intérêt des chercheurs pour ses applications potentielles dans les problèmes d'incertitude, les mesures de risque et le super-hedging en finance. Shige Peng a construit une sorte d'espérance entièrement non linéaire dynamiquement cohérente par l'approche des EDP. Un cas important d'espérance non linéaire cohérente en temps est la G-espérance, dans laquelle le processus canonique correspondant (B_{t})_{t≥0} est appelé G-mouvement brownien et joue un rôle analogue au processus de Wiener classique. L'objectif de cette thèse est d'étudier, dans le cadre de la G-espérance, certaines équations différentielles stochastiques rétrogrades (G-EDSR) à croissance quadratique avec applications aux problèmes de maximisation d'utilité robuste avec incertitude sur les modèles, certaines équations différentielles stochastiques (G-EDS) réfléchies et équations différentielles stochastiques rétrogrades réfléchies avec générateurs lipschitziens. On considère d'abord des G-EDSRs à croissance quadratique. Dans le Chapitre 2 nous fournissons un resultat d'existence et unicité pour des G-EDSRs à croissance quadratique. D'une part, nous établissons des estimations a priori en appliquant le théorème de type Girsanov, d'où l'on en déduit l'unicité. D'autre part, pour prouver l'existence de solutions, nous avons d'abord construit des solutions pour des G-EDSRs discretes en résolvant des EDPs non-linéaires correspondantes, puis des solutions pour les G-EDSRs quadratiques générales dans les espaces de Banach. Dans le Chapitre 3 nous appliquons les G-EDSRs quadratiques aux problèmes de maximisation d'utilité robuste. Nous donnons une caratérisation de la fonction valeur et une stratégie optimale pour les fonctions d'utilité exponentielle, puissance et logarithmique. Dans le Chapitre 4, nous traitons des G-EDSs réfléchies multidimensionnelles. Nous examinons d'abord la méthode de pénalisation pour résoudre des problèmes de Skorokhod déterministes dans des domaines non convexes et établissons des estimations pour des fonctions α-Hölder continues. A l'aide de ces résultats obtenus pour des problèmes déterministes, nous définissons le G-mouvement Brownien réfléchi et prouvons son existence et son unicité dans un espace de Banach. Ensuite, nous prouvons l'existence et l'unicité de solution pour les G-EDSRs multidimensionnelles réfléchies via un argument de point fixe. Dans le Chapitre 5, nous étudions l'existence et l'unicité pour les équations différentielles stochastiques rétrogrades réfléchies dirigées par un G-mouvement brownien lorsque la barrière S est un processus de G-Itô. / Since the publication of Choquet's (1955) book, the theory of nonlinear expectation has attracted great interest from researchers for its potential applications in uncertainty problems, risk measures and super-hedging in finance. Shige Peng has constructed a kind of fully nonlinear expectation dynamically coherent by the PDE approach. An important case of time-consistent nonlinear expectation is G-expectation, in which the corresponding canonical process (B_{t})_{t≥0} is called G-Brownian motion and plays a similar role to the classical Wiener process. The objective of this thesis is to study, in the framework of the G-expectation, some backward stochastic differential equations (G-BSDE) under a quadratic growth condition on their coefficients with applications to robust utility maximization problems with uncertainty on models, Reflected stochastic differential equations (reflected G-SDE) and reflected backward stochastic differential equations with Lipschitz coefficients (reflected G-BSDE). We first consider G-BSDE with quadratic growth. In Chapter 2 we provide a result of existence and uniqueness for quadratic G-BSDEs. On the one hand, we establish a priori estimates by applying the Girsanov-type theorem, from which we deduce the uniqueness. On the other hand, to prove the existence of solutions, we first constructed solutions for discrete G-BSDEs by solving corresponding nonlinear PDEs, then solutions for the general quadratic G-BSDEs in the spaces of Banach. In Chapter 3 we apply quadratic G-BSDE to robust utility maximization problems. We give a characterization of the value function and an optimal strategy for exponential, power and logarithmic utility functions. In Chapter 4, we discuss multidimensional reflected G-SDE. We first examine the penalization method to solve deterministic Skorokhod problems in non-convex domains and establish estimates for continuous α-Hölder functions. Using these results for deterministic problems, we define the reflected G-Brownian motion and prove its existence and its uniqueness in a Banach space. Then we prove the existence and uniqueness of the solution for the multidimensional reflected G-SDE via a fixed point argument. In Chapter 5, we study the existence and uniqueness of the reflected backward stochastic differential equations driven by a G-Brownian motion when the obstacle S is a G-Itô process.

L’objectif de cette thèse est l’étude d’un schéma numérique pour l’approximation des solutions d’équations différentielles doublement stochastiques rétrogrades (EDDSR). Durant les deux dernières décennies, plusieurs méthodes ont été proposées afin de permettre la résolution numérique des équations différentielles stochastiques rétrogrades standards. Dans cette thèse, on propose une extension de l’une de ces méthodes au cas doublement stochastique. Notre méthode numérique nous permet d’attaquer une large gamme d’équations aux dérivées partielles stochastiques (EDPS) nonlinéaires. Ceci est possible par le biais de leur représentation probabiliste en termes d’EDDSRs. Dans la dernière partie, nous étudions une nouvelle méthode des particules dans le cadre des études de protection en neutroniques. / The purpose of this thesis is to study a numerical method for backward doubly stochastic differential equations (BDSDEs in short). In the last two decades, several methods were proposed to approximate solutions of standard backward stochastic differential equations. In this thesis, we propose an extension of one of these methods to the doubly stochastic framework. Our numerical method allows us to tackle a large class of nonlinear stochastic partial differential equations (SPDEs in short), thanks to their probabilistic interpretation. In the last part, we study a new particle method in the context of shielding studies.