Time-frequency analysis of locally stationary Hawkes processes (Q1740528)

The concept of locally stationary Hawkes process was introduced in [the first author et al., Stochastic Processes Appl. 126, No. 6, 1710--1743 (2016; Zbl 1336.60094)]. The model contains a time-varying baseline immigrant intensity function \(\lambda_c(t)\) and time-varying fertility function \(p(t)\). In this paper, concepts of local mean density function and Bartlett spectrum (the Fourier transform of the autocovariance) of the process that are associated with these functions are introduced and their estimators are provided. Bounds on the bias and variance of these estimators are established under some regularity conditions. The approach involves approximation of the locally stationary process by a suitable stationary Hawkes process. Connection between these results and existing literature, and applications are discussed. Supplementary material is provided in this article.