The entropy distance between the Wiener and stationary Gaussian measures (Q1306254)

Let \(H(\cdot,\cdot)\) denote the entropy (or Hellinger) ``distance'' on the set of probability measures on \({\mathcal C}=C^0([1-\varepsilon, 1+\varepsilon]; \mathbb{R})\), \(0<\varepsilon <1\). Let \(\gamma\) denote the standard Wiener measure on \({\mathcal C}\), and \(\mathbb{P},\mathbb{Q}\) denote generic absolutely continuous stationary Gaussian measures on \({\mathcal C}\). It is shown that (i) \(H(\gamma,\mathbb{P})\geq c>0\) if \(\mathbb{P}\) is not centered or with variance \(\neq 1\); (ii) \(H(\gamma,\mathbb{P})\sim\varepsilon/2\) if \(\varepsilon\to 0\) and \(\mathbb{P}\) is centered with variance \(=1\); (iii) \(H(\mathbb{Q},\mathbb{P}) ={\mathcal O}(\varepsilon^2)\) if \(\mathbb{P},\mathbb{Q}\) are centered with same variance and \(C^2\) covariance.