Mathematics · Statistics and Probability
Statistical Inference for Discretely Observed Markov Jump Processes
Mogens Bladt, Michael Sørensen
See on the mapAbstract
Summary Likelihood inference for discretely observed Markov jump processes with finite state space is investigated. The existence and uniqueness of the maximum likelihood estimator of the intensity matrix are investigated. This topic is closely related to the imbedding problem for Markov chains. It is demonstrated that the maximum likelihood estimator can be found either by the EM algorithm or by a Markov chain Monte Carlo procedure. When the maximum likelihood estimator does not exist, an estimator can be obtained by using a penalized likelihood function or by the Markov chain Monte Carlo procedure with a suitable prior. The methodology and its implementation are illustrated by examples and simulation studies.
Cited by 2
2 papers in the index cite this.
The cited-by list is still filling in while reference lists are collected.