Euler estimates for rough differential equations (Q2467726)

The authors consider controlled differential equation of the form \(dy=V_1(y)dx^1+\cdots+V_{d}(y)dx^{d}\) on the time interval \([0,1]\) with output signal \(y=y(t)\in \mathbb{R}^{l}\), an input or driving signal \(x=x(t)\in \mathbb{R}^{d}\) and \(d\) vector fields \(V_1,\ldots, V_{d}\) on \(\mathbb{R}^{l}\). It is presented a generalization of certain Euler estimates for controlled ordinary differential equations. These Euler estimates are extended to rough differential equation solutions. The authors obtained strong remainder estimates which remain valid in the context of rough differential equations. Then it is considered rough differential equations driven by random rough paths. The authors study stochastic Taylor expansion with strong remainder estimates. When specialized to Brownian motion, the authors find an improved version of estimate by Azencott. It is shown how the presented results lead to moment control of random rough differential equation solutions driven by random rough paths of sufficient integrability and it is given a criteria for \(L^{q}\)-convergence in the universal limit theorem.