\(L^p\)-variations for multifractal fractional random walks (Q930681)

Let \(M\) be a continuous multifractal random measure (MRM) as defined by \textit{E. Bacry} and \textit{J. F. Muzy} [Commun. Math. Phys. 236, No. 3, 449--475 (2003; Zbl 1032.60046)] and let \(\{B(t), t \geq 0\}\) be a Brownian motion which is independent of \(M\). The subordinated process \(Y_t = B(M([0, t])),\;(t \geq 0)\) is called a multifractals random walk (MRW). Bacry and Muzy [loc. cit.] proved that the process \(\{Y_t, t \geq 0\}\) is equivalent in law to the limit process \[ Y_t' = \lim_{l \to 0^+} \int_0^t e^{w_l(s)/2}\, dB(s). \] See [loc. cit.] or the paper under review for the definition of \(w_l(s)\). In this paper the author extends the above result of [loc. cit.] by replacing Brownian motion \(B\) by a fractional Brownian motion \(B^H\) with Hurst index \(H > 1/2\). The author proposes two approaches to define the so-called multifractal fractional random walk (MFRW) as limit in distribution of a sequence of stochastic integrals with respect to fractional Brownian motion, where the integrands are of the form \(l^{-\psi(1/2)} e^{w_l(s)/2}\) and \(e^{w_l(s)}\), respectively. See the paper under review for the precise conditions and the undefined notation. It should be mentioned that the author shows that both approaches lead to MFRWs which are different from the fractional Brownian motion in multifractal time \(U_t= B^H(M[0, t])\). Based on the scaling properties of MRM and the conditionally Gaussian structure, the author studies in Section 3 the asymptotic properties of MFRW for \(H \in (1/2, 3/4)\). Multifractal fractional random walks are useful models for scaling phenomena. The scaling property of a MFRW \(X(t)\) can be described by the scaling exponent \(\zeta(p)\) defined by \[ \zeta(p) := \frac{\log {\mathbb E}\big(|X(t+\tau) - X(t)|^p\big)} {\log \tau} \] at different scales \(0 < \tau < T\). It is interesting from the theoretical and applied point of view to estimate the scaling function \(\zeta(p)\). As applications of the asymptotic results in Section 3, the author studies the properties of a moment based estimator of the scaling function \(\zeta(p)\) and provides a construction of asymptotic confidence intervals.