September 26, 2017

Exact Nonlinear and Non-Gaussian Kalman Smoother for State Space Models with Implicit Functions and Equality Constraints

In this paper, I present a novel implementation of the exact nonlinear and non-Gaussian Kalman smoother that can also deal with implicit functions in the measurement and/or state equations as well as equality constraints.

Read the accompanying paper CPB Discussion  Paper 360.

My approach has the additional advantage that it can be fully automated, on the basis of which I have developed a toolbox that can handle a wide class of discrete-time state space models. The toolbox is documented in an accompanying paper, while the technical details are presented in the current one.


Joris de Wind