1: \begin{abstract}

2: Consider a family of random ordinary differential equations on a manifold driven by vector fields of the form $\sum_kY_k\alpha_k(z_t^\epsilon(\omega))$ where $Y_k$ are vector fields, $\epsilon$ is a positive number, $z_t^\epsilon$ is a  ${1\over \epsilon} \L_0$ diffusion process taking values in possibly a different manifold, $\alpha_k$ are annihilators of $\ker (\L_0^*)$. Under H\"ormander type conditions on $\L_0$ we prove that, as $\epsilon $ approaches zero,

3: the stochastic processes $y_{t\over \epsilon}^\epsilon$ converge weakly and in the Wasserstein topologies. We describe this limit and give an  upper bound for the rate of the convergence. 

4: 

5: 

6: %We study limits of solutions $(y_t^\epsilon)$ to a family of random ordinary differential equations on a state space $M$ with parameter $\epsilon$

7: %where the fast motions $(z_t^\epsilon)$ are Markov processes on a compact space $G$ with generator of the form ${1\over \epsilon} \L_0$.  (1) The stochastic processes  $\{(y^\epsilon_{t\over \epsilon}), \epsilon\in (0,1)\}$ are tight if $\L_0$ is a regularity improving Fredholm operator; $(y^\epsilon_{t\over \epsilon})$ converges weakly as $\epsilon\to 0$ if $\L_0$ satisfies H\"ormander's condition and has Fredholm index $0$; an estimate for the rate of convergence in the Wasserstein 1-distance is obtained if $\L_0$ satisfies strong H\"ormander's condition.  (2)

8: %If $\L={1\over 2} \sum_{k=1}^m (Y_k)^2+Y_0$ and $Y_k$ are controlled by a pair of functions $(V, \tilde V)$, we give estimates for the derivatives of  its probabilistic semigroup $P_t$, of all orders.

9:  \end{abstract}

10: