1: \begin{abstract}

2: %In this paper we study fully-coupled fast-slow systems of the form $\dot{x}_\epsilon 

3: %= a(x_\epsilon,y_\epsilon) +\epsilon^{-1}b(x_\epsilon,y_\epsilon)$, 

4: %$\dot{y}_\epsilon = \epsilon^{-2}g(x_\epsilon,y_\epsilon)$, where $\epsilon$ is a 

5: %small parameter and, for every fixed $x$, the fast dynamics are sufficiently chaotic 

6: %with ergodic invariant measure $\mu_x$ such that $b$ has average zero with respect to 

7: %$\mu_x$. Results of the form $x_\epsilon \to X$ in the limit $\epsilon\to 0$, where $X$ 

8: %solves a homogenized stochastic differential equation (SDE), have so far only been obtained 

9: %for the case $g= g(y)$, i.e., only for fast dynamics which evolve independently. In this 

10: %paper we give sufficient conditions for the convergence of the first moments of $x_\epsilon$ 

11: %in the \emph{fully-coupled case} $g = g(x,y)$, where previous approaches seemed to struggle.

12: %Our proof is based upon a stochastic regularization and functional-analytical techniques combined 

13: %via a double limit procedure involving a zero-noise limit as well as considering $\epsilon\ra 0$. 

14: %We also give exact formulas for the drift and diffusion coefficients for the limiting SDE of $X$. 

15: %Since our conditions are relatively strong for general systems, we also study weakly-coupled 

16: %systems of the form $\dot{x}_\epsilon = a(x_\epsilon, y_\epsilon) + \frac{1}{\epsilon} 

17: %b(x_\epsilon, y_\epsilon)$, $\dot{y}_\epsilon = \frac{1}{\epsilon^2} g(y_\epsilon) + \frac{1}{\epsilon} 

18: %h(x_\epsilon,y_\epsilon) + r(x_\epsilon,y_\epsilon)$, where coupling occurs only in lower time 

19: %scales and for which our theory is fully applicable.

20: %\end{abstract}

21: