An extension of the Lévy characterization to fractional Brownian motion (Q2431515)

Let \(X\) be a continuous square integrable zero mean random process on some probability space. The main theorem states that for a continuous square integrable centered random process \(X\), the following properties are equivalent: {\parindent=7mm \begin{itemize}\item[(i)]the process \(X\) is a fractional Brownian motion with self-similarity index \(H\in(0,1)\); and \item[(ii)]the process \(X\) has the properties (a), (b), (c) for some \(H\in[0,1]\). \end{itemize}} These properties are as follows: {\parindent=7mm \begin{itemize}\item[(a)]the sample paths of the process \(X\) are \(\beta\)-Hölder continuous for any \(\beta\in(0,H)\); \item[(b)]for \(t>0\), \[ n^{2H-1}\sum^{n}\limits_{k=1}\left(X_{t_k}-X_{t_{k-1}}\right)\overset{L^{1}(p)}{\longrightarrow }t^{2H} \quad \text{as} \quad n\rightarrow\infty; \] \item[(c)]the process \[ M_t=\int^{t}\limits_{0}s^{\frac{1}{2}-H}(t-s)^{\frac{1}{2}-H}dX_s \] is a martingale with respect to the filtration \(F^X\). \end{itemize}} This theorem extends the classical characterization theorem due to P. Lévy for the Brownian motion \((H=\frac{1}{2})\) to the fractional Brownian motion. Section 2 contains one auxiliary result for orthogonal martingales. In Sections 3 and 4, the theorem is proved for \(H>\frac{1}{2}\) and \(H<\frac{1}{2}\). In order to prove that \(X\) is a special Gaussian process, the representation of the process \(X\) with respect to a certain martingale is applied. A different proof of the theorem appeared in \textit{Y. Mishura} [Stochastic calculus for fractional Brownian motion and related processes. Lecture Notes in Mathematics 1929. Berlin: Springer (2008; Zbl 1138.60006)].