Multivariate supOU processes (Q627238): Difference between revisions

The authors motivate why standard and fractional Ornstein-Uhlenbeck processes are not sufficient for modeling real data. In this connection, the first idea is to add up countably many independent OU-type processes to preserve both jumps and long memory. The resulting processes are called supOU processes. The next idea is to introduce and study multivariate supOU processes. Due to the appearance of matrices and related peculiarities, the latter theory is not a straightforward extension of univariate results. The main tools and topics are: a comprehensive introduction into Lévy bases and the related integration theory which is needed to define supOU processes; existence conditions and the second-order structure of supOU processes; several examples illustrating the behavior and properties of supOU processes, including long memory. Positive semi-definite supOU processes are defined and applied to stochastic volatility modeling.