Book contents
- Frontmatter
- Contents
- Preface
- Introduction
- The story in a nutshell
- I Basics
- II Abstract theory of rough paths
- III Stochastic processes lifted to rough paths
- IV Applications to stochastic analysis
- 17 Stochastic differential equations and stochastic flows
- 18 Stochastic Taylor expansions
- 19 Support theorem and large deviations
- 20 Malliavin calculus for RDEs
- Appendices
- Frequently used notation
- References
- Index
19 - Support theorem and large deviations
from IV - Applications to stochastic analysis
Published online by Cambridge University Press: 05 June 2012
- Frontmatter
- Contents
- Preface
- Introduction
- The story in a nutshell
- I Basics
- II Abstract theory of rough paths
- III Stochastic processes lifted to rough paths
- IV Applications to stochastic analysis
- 17 Stochastic differential equations and stochastic flows
- 18 Stochastic Taylor expansions
- 19 Support theorem and large deviations
- 20 Malliavin calculus for RDEs
- Appendices
- Frequently used notation
- References
- Index
Summary
We now discuss some classical results of diffusion theory: the Stroock–Varadhan support theorem and Freidlin–Wentzell large deviation estimates. Everything relies on the fact that the Stratonovich SDE
can be solved as an RDE solution which depends continuously on enhanced Brownian motion in rough path topology, subject to the suitable Lipregularity assumptions on the vector fields. (A summary of the relevant continuity results was given in Section 17.1.)
Support theorem for SDEs driven by Brownian motion
Theorem 19.1 (Stroock–Varadhan support theorem)Assume that V = (V1, …, Vd) is a collection of Lip2-vector fields on ℝe, and V0is a Lip1-vector field on ℝe. Let B be a d-dimensional Brownian motion and consider the unique (up to indistinguishability) Stratonovich SDE solution Y on [0, T] to
Let us write yh = π(V, V0) (0, y0; (h, t)) for the ODE solution to
started at y0 ∈ ℝewhere h is a Cameron–Martin path, i.e. h. Then, for any α ∈ [0, 1/2) and any δ > 0,
(where the Euclidean norm is used for conditioning |B − h|∞;[0, T] < ∈).
Proof. Without loss of generality, α ∈ (1/3, 1/2). Let us write, for a fixed Cameron–Martin path h,
From Theorem 17.3, there is a unique solution π(V, V0) (0, y0; (B, t)) to the RDE with drift
which then solves the Stratonovich equation (19.1).
- Type
- Chapter
- Information
- Multidimensional Stochastic Processes as Rough PathsTheory and Applications, pp. 533 - 544Publisher: Cambridge University PressPrint publication year: 2010