1: \begin{abstract}

2: The Brownian frame process $\mathcal{T}^{B}$ is defined as 

3: \begin{equation*}

4: \mathcal{T}_{t}^{B}:=\left( B_{t-1+u}\right) _{0\leq u\leq 1},\text{ \ \ }%

5: t\in \left[ 0,1\right] ,

6: \end{equation*}%

7: where $B$ is a real-valued Brownian motion with parameter set $\left[ -1,1%

8: \right] $. This thesis investigates properties of the path-valued Brownian

9: frame process relevant to establishing an integration theory based on the

10: theory of rough paths (\cite{Lyons98}). The interest in studying this object

11: comes from its connection with Gaussian Volterra processes (e.g. \cite%

12: {Decreusefond05}) and stochastic delay differential equations (e.g. \cite%

13: {Mohammed84}). Chapter 2 establishes the existence of $\mathcal{T}^{B}$. We

14: then examine the convergence of dyadic polygonal approximations to $\mathcal{%

15: T}^{B}$ if the path-space $V$ where $\mathcal{T}^{B}$ takes its values is

16: equipped with first the $p$-variation norm ($p>2$) and second the $\sup $%

17: -norm. In the case of the $p$-variation norm, the Brownian frame process is

18: shown to have finite $\acute{p}$-variation for $\acute{p}>\frac{2p}{p-2}$.

19: In the case of the $\sup $-norm, it is shown to have finite $\acute{p}$%

20: -variation for $\acute{p}>2$. Chapter 3 provides a tail estimate for the

21: probability that two evaluations of the Brownian frame process are far apart

22: in the $p$-variation norm. Chapter 4 shows that $\mathcal{T}^{B}$ does not

23: have a L\'{e}vy area if $V\otimes V$ is equipped with the injective tensor

24: product norm (where $V=C\left( \left[ 0,1\right] \right) $).

25: \end{abstract}

26: