No label defined (Q4359378)

We consider a process \(x\) belonging (in quadratic mean) to a Hölder space with exponent \(s\). The process is observed at times \((n\Delta)_{n=0,\dots, N}\) with \(T=N\Delta\). We use a kernel \(E(u,v)\) such that: \(E(u,v)\) is a measurable function from \(\mathbb{R}^2\) to \(\mathbb{R}\) satisfying \(|E(u,v)|\leq \Phi(u-v)\) with \(\Phi\in L^1(\mathbb{R})\). In order to take into account the smoothness of the process, we assume that \(E(u,v)\) satisfies \[ \forall u\in\mathbb{R}:\quad\sum_{p\in\mathbb{Z}} E(u,p) (p-u)^k= \delta_0(k) \quad\text{for}\quad k=0,\dots, L=[s]. \] First a nonparametric estimate is build using the discrete observed data and the kernel \(E\). The strong consistency and the asymptotic normality are proved for \(T\to\infty\), \(\Delta\to 0\) and \(\sqrt{T} \Delta^s\to 0\). The case where the covariance function depends on a parameter is then studied. For the entire collection see [Zbl 0841.00017].