| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 808.55 KB | Adobe PDF |
Authors
Abstract(s)
We consider Forward Backward Stochastic Differential Equations (FBSDEs for short) with different assumptions on its coefficients. In a first part we present results of existence, uniqueness and dependence upon initial conditions and on the coefficients. There are two main methodologies employed in this study. The first one presented is the Four Step Scheme, which makes very clear the connection of FBSDEs with quasilinear parabolic systems of Partial Differential Equations (PDEs for short). The weakness of this methodology is the smoothness and regularity assumptions recquired on the coefficients of the system, which motivate the employment of Banach`s Fixed Point Theorem in the study of existence and uniqueness results. This classic analytical tool requires less regularity on the coefficients, but gives only local existence of solution in a small time duration. In a second stage, with the help of the previous work with a running-down induction on time, we can assure the existence and uniqueness of solution for the FBSDE problem in global time. The second goal of this work is the study of the assymptotic behaviour of the FBSDEs solutions when the diffusion coefficient of the forward equation is multiplicatively perturbed with a small parameter that goes to zero. This question adresses the problem of the convergence of the classical/viscosity solutions of the quasilinear parabolic system of PDEs associated to the system. When this quasilinear parabolic system of PDEs takes the form of the backward Burgers Equation, the problem is the convergence of the solution when the viscosity parameter goes to zero. To study conveniently this problem with a probabilistic approach , we present a concise survey of the classical Large Deviations Principles and the basics of the so-called "Freidlin-Wentzell Theory". This theory is mainly concerned with the study of the Itô Diffusions with the diffusion term perturbed by a small parameter that converges to zero and the richness of properties of the FBSDEs shows us that (even in a coupled FBSDE system) this approach is a good one, since we can extract for the solutions of the perturbed systems a Large Deviations Principle and state convergence of the perturbed solutions to a solution of a deterministic system of ordinary differential equations.
Description
Tese de mestrado em Matemática, apresentada à Universidade de Lisboa, através da Faculdade de Ciências, 2011
Keywords
Existence and uniqueness Forward backward stochastic differential Gradient estimates Quasilinear equations of parabolic type Four step scheme Large deviations Freidlin-Wentzell theory Burgers equations type Viscosity solutions Teses de mestrado - 2011