Convergence of Appell polynomials of long range dependent moving averages in martingale differences (Q1969274)

Consider sums \(S_{N,k}(t)= \sum_{s=1}^{[Nt]} A_k(X_s)\), \(t\geq 0\), \(k,N\in \mathbb{N}_+= \{1,2,\dots\}\), of Appell polynomials of the long-range dependent moving average process \(X_s= \sum_{i\leq s} b_{s-i} \zeta_i\), \(s\in \mathbb{N}_+\), where \((\zeta_i)_{i\in \mathbb{Z}}\) is a strictly stationary weakly dependent martingale difference sequence, and \(b_j\sim j^{d-1}\), \(j\in \mathbb{N}_+\), with \(0< d< 1/2\). It is proved that if \(k(1- 2d)< 1\), then, suitably normalized, \((S_{N,k} (t))_{t\geq 0}\) converges in distribution to the so-called \(k\)th order Hermite process as \(N\to \infty\). This generalizes the corresponding results of the first named author [Lith. Math. J. 22, 327-340 (1982); translation from Litov. Mat. Sb. 22, No. 3, 185-201 (1983; Zbl 0515.60057)] and \textit{F. Avram} and \textit{M. S. Taqqu} [Ann. Probab. 15, 767-775 (1987; Zbl 0624.60049)] for the case of an i.i.d. sequence \((\zeta_i)_{i\in \mathbb{Z}}\).