Ergodicity of stochastic differential equations driven by fractional Brownian motion (Q1775448): Difference between revisions

The paper considers finite-dimensional stochastic differential equations of the form \(dx=f(x)\,dt+\sigma\,dB_H(t)\), where \(x\in\mathbb R^n\), \(B_H\) is an \(n\)-dimensional fractional Brownian motion with Hurst parameter \(H\in(0,1)\), and \(\sigma\) is a constant invertible matrix. Assuming \(f\) to be sufficiently regular and dissipative, existence of a unique stationary solution is established, which has, in addition, the property that for every initial condition the solution converges to the stationary solution in total variation norm with polynomial speed, where small (\(H<\frac12\)) and large (\(\frac12<H<1\)) Hurst parameters are dealt with separately. For \(H\neq\frac12\) a general framework has to be developed in order to give a meaning to the assertions made in the previous sentence and, in particular, to investigate joint properties of two or more solutions. In this context the notion of a stochastic dynamical system is introduced, which is a concept between the notion of a random dynamical system [see \textit{L. Arnold}, ``Random dynamical systems'' (1998; Zbl 0906.34001)] and Markov semigroup theory used in classical stochastic analysis. The main technical ingredient is a coupling construction, as it has been used previously by \textit{M. Hairer} [Probab. Theory Relat. Fields 124, 345--380 (2002; Zbl 1032.60056)] and by \textit{J. C. Mattingly} [Commun. Math. Phys. 230, 421--462 (2002; Zbl 1054.76020)].