Contibution by Manfred Deistler

Abstract

We give a survey on system identification, emphasizing the structure of the problem
rather than algorithms. The focus is on what we call the linear mainstream case.
The main parts considered are:

  -Structure theory, in particular parametrization.

  -Estimation for a given subclass, with an emphasis on maximum likelihood
   estimation and its asymptotic properties.
 
  -Model selection with an emphasis on information criteria.

Linear non-mainstream cases such as cointegration and nonlinear system are
discussed in addition.


SLIDES