The standard filtered Poisson process model of shot noise has unbounded paths. This property of the standard model precludes its use in a variety of theoretical studies of nonlinear dynamical systems. In addition, the number of Unif(0,1) variates required for computer simulation of shot noise random fields based on the standard model has no fixed tipper bound. This is a significant disadvantage in Monte Carlo simulations involving many replicates. A model of shot noise based on a finite point process is presented which has neither of these liabilities but, yet, approximates to any desired degree of accuracy the Poisson statistics of the standard model with any given power spectrum.
展开▼
