MDP for integral functionals of fast and slow processes with averaging

We establish large deviation principle (LDP) for the family of vector-valued random processes $(X^\epsilon,Y^\epsilon),\epsilon\to 0$ defined as $$X^\epsilon_t=\frac{1}{\epsilon^\kappa}\int_0^t H(\xi^\epsilon_s,Y^\epsilon_s)ds, dY^\epsilon_t=F(\xi^\epsilon_t,Y^\epsilon_t)dt+ D\epsilon^{1/2-\kappa}G(\xi^\epsilon_t,Y^\epsilon_t)dW_t,$$ where $W_t$ is Wiener process and $\xi^\epsilon_t$ is fast ergodic diffusion. We show that, under $\kappa<{1/2}$ or less and Veretennikov-Khasminskii type condition for fast diffusion, the LDP holds with rate function of Freidlin-Wentzell's type.