A1226
Title: A trace approximation theorem for products of Toeplitz matrices and its applications
Authors: Tetsuya Takabatake - The University of Osaka (Japan) [presenting]
Abstract: The aim is to present recent progress on deriving precise approximation error bounds for the trace of products of Toeplitz matrices, where the entries are defined through the Fourier transforms of the spectral density function of a stationary time series and its derivatives with respect to model parameters. These results provide a theoretical foundation for developing asymptotically optimal estimation methods based on likelihood analysis in both the time and frequency domains, applicable to general, possibly non-centered stationary Gaussian time series. It further demonstrates how these bounds can be used to establish the asymptotic efficiency of the resulting estimators.
