Stable Convergence of Certain Functionals of Diffusions Driven by fBm

From MaRDI portal





Let \(B^H_t\) be a fractional Brownian motion with \(1/2< H< 3/4\) and \({\mathbf F}_t\) the filtration generated by \(B^H_t\). Let \(\sigma\), \(f\), \(G\) be functions satisfying some conditions. Define \(X_t\) by \[ X_t= x_0+ \int^t_0\sigma(X_s)\,dB^H_s \] and let \(X_n(s)\) be the broken-line approximation of \(X_t\) at step \(1/n\) in \([0,1]\). Let \(Z\) be an \(N(0,1)\) random variable, independent of \(B^H_t\). Denote by \(E_{{\mathbf F}_t}\) the conditional expectation given \({\mathbf F}_t\) and put \(G_t(\sigma(X_t))= E_{{\mathbf F}_t}G(\sigma(X_t) Z)\). The authors prove, among others, that the sequence of processes \[ \sqrt{n}\Biggl(\int^t_0 f(X_n(s)) G(\dot X_n(s) n^{H- 1})\,ds- \int^t_0 F(X(s)) G_s(\sigma(X_s))\,ds\Biggr) \] converges stably to a Gaussian process \(U(t)\) which can be written as \[ U(t)= \int^t_0 f(X_s) C^{{1/2}}_s (X_s)\,dW_s, \] where \(W\) is a Wiener process, indepenent of \(X_t\), and \[ C_s(X_s)= C_H E\{(G(\sigma(X_s) Z)- G_s(X_s))^2\mid {\mathbf F}_s\} \] with \(C_H\) some positive constant.











This page was built for publication: Stable Convergence of Certain Functionals of Diffusions Driven by fBm

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3158138)