Weak convergence of a nonlinear functional of a stationary Gaussian process. Application to the local time (Q5932676)

The problem of weak convergence for nonlinear functionals of stationary Gaussian processes indexed on the positive integers has been studied by many authors. \textit{X. Guyon} and \textit{J. Leon} [Ann. Inst. Henri Poincaré, Probab. Stat. 25, No. 3, 265-282 (1989; Zbl 0691.60017)] presented a different point of view. Beginning with a stationary Gaussian process on \([0,1]\), they build a sequence of discrete time processes by considering the values of the original process over the points of a set of increasing regular partitions of \([0,1]\). Depending on the behaviour of the correlation for nearby observations, they obtained a CLT or the convergence to a non-Gaussian distribution. They considered the weak convergence in \(R\). In this paper the convergence is extended to the weak convergence in \(C[0,1]\). Also the problem of the asymptotic independence (in a distributional sense) between the initial process and the limit process, based on the papers of \textit{H.-C. Ho} and \textit{T.-C. Sun} [Ann. Probab. 18, No. 3, 1159-1173 (1990; Zbl 0712.60021)], and \textit{C. Berzin, J. R. Leon} and \textit{J. Ortega} [Probab. Math. Stat. 18, 39-81 (1998)] is considered. The authors get new results on the distributional limit theorem for a random variable that involves the local time and the number of crossings. They study the weak convergence of functionals related to the increments of the local time for stable processes.