Some results concerning the asymptotic distribution of sample Fourier transforms and periodograms for a discrete-time stationary process with a continuous spectrum (Q2742775)

The stationary process \(\{X_t,\;t=0, 1, \ldots\}\) with continuous spectrum is considered. The author supposes that \(\{X_t\}\) is linearly non-deterministic, so that it has a canonical moving-average representation \(X_t = \sum_{s=0}^\infty C_s \varepsilon_{t-s}\), where \(\varepsilon_t\) are uncorrelated. The author studies the behaviour \((\text{for} n \to \infty)\) of a tapered Fourier transform NEWLINE\[NEWLINEd_{n,x}(\lambda) = h(t/n) X_t \exp\{ -i\lambda t\},\;|\lambda|\leq \pi\;\text{of} X_0, X_1, \ldots, X_{n-1},NEWLINE\]NEWLINE where \(h(\cdot)\) is a tapering function on \([0, 1]\). It is assumed to be of bounded variation but can be otherwise arbitrary. In particular, \(h(\cdot) \equiv 1\) gives the ordinary Fourier transform. Let \(f(\lambda)\) be the spectral density function of \(X_t.\) Let NEWLINE\[NEWLINE D_{n, x} (\lambda) = [{\text Re} \{ d_{n,x}(\lambda)\}, {\text Im} \{ d_{n,x}(\lambda)\}],\quad H_2 = \int_0^1 h^2(t) dt.NEWLINE\]NEWLINE The main result of this paper is the following.NEWLINENEWLINENEWLINELet \(f(\lambda)\) be continuous at \(\lambda = \lambda_1, \lambda_2, \ldots, \lambda_k (\neq 0).\) Then under some conditions on \((\varepsilon_t)\) the following assertion holds true: NEWLINE\[NEWLINE\biggl\{ D_{n,x} (\lambda_1)(2\pi n H_2)^{-1/2}, \cdots,D_{n,x} (\lambda_k)(2\pi n H_2)^{-1/2}\biggr\} \Rightarrow N(0, V),NEWLINE\]NEWLINE where \(N\) is a multivariate normal distribution with zero-mean and covariance matrix \(V\), \(V =\) diag \((V_1, \ldots, V_k)\) with \(V_r =2^{-1}f(\lambda_r) I_2\), \(\lambda_r \neq \pi\) and \(V_r = f(\lambda_r)\), \(\lambda_r = \pi_0.\) The results are extended to processes for which \(f(\cdot)\) is continuous except at \(\lambda =0\), with \(\lim_{\lambda \to 0} f(\lambda) \lambda^{2d} = K\), where \(0< d < 2^{-1}\), as is typical of certain types of processes with long-range dependence.