The 1/\(H\)-variation of the divergence integral with respect to the fractional Brownian motion for \(H>1/2\) and fractional Bessel processes (Q2485819): Difference between revisions

Stochastic processes used to model the input of a real system are often supposed to be Markovian, but for some important applications in finance, telecommunication networks, biology etc. it is required that the underlying process satisfies the long-range dependence criterion; this means that the behavior of the process after a time \(t\) depends not only on the situation at time \(t\) but also on the previous history of the process. The essential feature that makes the fractional Brownian motion (fBm) an appropriate model for many applications is the fact that fBm is \(H\)-self-similar (\(H \in (0,1)\)) Gaussian process whose increments are stationary and exhibit long-range dependence. The authors consider the stochastic integral with respect to fBm in the case \(\frac{1}{2} <H<1,\) as defined in the framework of Malliavin calculus. In particular, they consider the divergence integral \(\int_0^t u_s \delta B_s,\) as defined by \textit{E. Alòs} and \textit{D. Nualart} [Stochastics Stochastics Rep. 75, 129--152 (2002; Zbl 1028.60048)]. The main result is that the \(\frac{1}{H}\)-variation of the divergence integral \(\int_0^{\cdot} u_s \delta B_s\) on a time interval \([0,T]\) is equal to \(C_H \int_0^T | u_s| ^{1/H} \,ds,\) where \(C_H\) is a constant depending on \(H.\) As an application, they discuss the \(\frac{1}{H}\)-variation properties of a particular representation of Bessel processes with respect to fBm.