Examples of moderate deviation principle for diffusion processes (Q2471398): Difference between revisions

The authors study the moderate deviation principle (MDP) for the family \((S_t^k)_{t\to+\infty}, k\in (0.5,1):\) \[ S_t^k:=\frac{1}{t^k}\int_0^tH(X_s)ds, \] where \(X=(X_t)_{t\geq 0}\in R^d,d\geq 1\) is an ergodic diffusion process with unique invariant measure \(\mu(dz)\) possessing a density \(p(z).\) The function \(H\) is assumed to be integrable relative to \(\mu(dz)\) and has zero barycenter \[ \int_{R^d}H(z)p(z)dz=0. \] The SDE under consideration is \(dX=b(X)dt+\sigma(X)dW(t),\) generated by a standard vector-valued Wiener process \(W.\) In most papers on the topic, the Laplace transform is used to study the MDP. In this paper, authors replace this technique by `stochastic exponential' one, enabling to formulate the MDP conditions in terms of `drift-diffusion' parameters and \(H.\)