On dynamical systems perturbed by a null-recurrent fast motion: the continuous coefficient case with independent driving noises (Q325917)

The aim of this paper is to study the system of diffusion processes \((X^\varepsilon (t), Y^\varepsilon (t))\in \mathbb R^{d+1}\) satisfying the stochastic differential equation \[ dX^\varepsilon (t)=\frac{1}{\varepsilon} \varphi(X^\varepsilon (t), Y^\varepsilon (t))dW(t),\qquad X(0)=x_0 \] and \[ dY^\varepsilon (t)= b(X^\varepsilon (t), Y^\varepsilon (t))dt + \sigma (X^\varepsilon (t), Y^\varepsilon (t))dW(t),\quad Y(0)=y_0, \] where \(b\) is a \(d\)-dimensional vector function and \(W(t)\) is an \(r\)-dimensional Brownian motion. The process depends on the parameter \(\varepsilon\), which is the ratio of the two different time scales. Here, \(X^\varepsilon\) changes faster in time as \(\varepsilon \to 0\). The authors are interested in the case when \(b(x,y)=b_1(x)+ b_2(x,y)\), where \(b_2\) and \(\sigma\) are very small as \(|x|\to \infty\). Then it follows that \(Y^\varepsilon\) converges to the solution of the ordinary differential equation \(\dot{y}= b_1(y)\) and slow motion is a perturbation of this ODE. The paper starts with the special case where the fast motion is a Brownian motion. Then, the authors prove some lemmas about continuity properties of a certain functional of the fast Brownian motion and show that the law of the process \(\zeta^\varepsilon (t)=\varepsilon^{-1/2}( Y^\varepsilon (t) - y(t))\) converges in distribution to the solution of a certain differential equation. Finally, they establish that there exist independent Brownian motions \(W_1\) and \(W_2\) such that the process \((\varepsilon X^\varepsilon (t), \zeta^\varepsilon (t))\) converges weakly to solutions of two stochastic differential equations with noises depending on \(W_1\) and \(W_2\), respectively.