Large-deviation principle for conditional distributions of diffusion processes (Q1567745)

The authors deal with the stochastic differential equation \[ dx^\varepsilon_t= \sigma(x^\varepsilon_t, \xi_{t/ \varepsilon})dw^x_t+ b(x^\varepsilon_t, \xi_{t/ \varepsilon})dt, \quad x_0=\eta, \] where \(\xi_t\) is a stationary and ergodic process, the r.v. \(\eta\) has a Lebesgue density and \((w^x_t, \xi_t,\eta)\) are independent. The main result is a large deviation principle for the family of conditional distributions \(\mu_t^\varepsilon (\varepsilon, A)=P (x^\varepsilon_t\in A\mid\sigma \{\xi_s,0\leq s\leq t/ \varepsilon\})\) defined on the space of trajectories.