'Unobserved' Monte Carlo method for identification of partially observed nonlinear state space systems. Part II. Counting process observations

摘要

For Part I see Technical Report, School of Elec. Eng., Univ. N.S.W. (2000). In this paper we present a simple simulation method for generating an approximate likelihood function for fitting partially observed (with counting process observations) nonlinear stochastic differential equations. We discuss the use of the method to generate approximate maximum likelihood estimators. We also mention methods based on density evolution equations.